Google Ads Location Settings: Who They Let In and Who They Keep Out
Google Ads location settings read two facts about each searcher: where they are, and which places they've shown interest in. Use a setting that reads the wrong one and you either pay for strangers elsewhere or cut off the people elsewhere who buy for your area.
Last verified November 2026
Google Ads location settings decide which searchers can see a campaign's ads. They work from two different facts about each person: where the person is, and which places the person has shown interest in, through a city named in their search, places they've searched for or visited before, or the page they're reading. Each setting reads one of those facts, or both, and most location mistakes come from using a setting that reads the wrong one. A florist switches to the Presence option to stop paying for clicks from other states, and stops reaching the people in other states who order flowers for their mothers. An advertiser excludes a region and keeps paying for searchers elsewhere who are looking for it. A manager opens the Matched locations report, sees only the cities the campaign targets, and decides nothing is leaking, when that report can show the city a person searched for rather than where the person was.
This guide is for people who manage real-spend Search, Shopping or Performance Max campaigns for a business that serves a defined area or set of countries: local and multi-location businesses, multi-country ecommerce and B2B advertisers, and businesses whose buyers search from somewhere else, such as florists, tour operators, venues and agents selling property to people who live elsewhere. It shows you how to choose the location option for each campaign, reach remote buyers without paying for everyone elsewhere, find where out-of-area clicks came from, keep areas out with the right tool, and structure, steer and test by area. It assumes you know Google Ads, and it doesn't cover a first setup or Local Services Ads. If location settings are new to you, read from the first section, because every later section uses its model. If you came with one question, start here:
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| If you're here because… | Go to |
|---|---|
| You're choosing between Presence and Presence or interest | Presence or "Presence or interest": choose for each campaign |
| Your customers order or book for your area from somewhere else | Reach people who buy for your area from somewhere else |
| Analytics, a session tool or your leads show people outside your area | Clicks from outside your area: find where they came from |
| You want a city, region or country kept out | Keep an area out: exclusion, place-name negative or a change of option |
| You're deciding whether an area gets its own campaign | One campaign per area, or several areas in one |
| You advertise in more than one country | Countries: structure, exclusions and Shopping offers |
| You need more spend in one area and less in another | Move spend between areas when Smart Bidding sets the bids |
| You've just changed a location setting, or you're planning a geo test | Change location settings without losing the read |
The short version. Google keeps two locations for each searcher. The default option, Presence or interest, admits someone on either one. Presence admits only people who are in, or regularly in, your target. An exclusion keeps out only people in the excluded area. By Google's own definitions, neither option keeps out someone inside your area who searches for a place you don't serve; a negative keyword naming that place does. So choose each campaign's option by who pays and where they are when they search:
- Buyers in or regularly in the area, with few elsewhere: Presence.
- More buyers elsewhere, on Search, when a split is worth its cost: a local campaign on Presence, plus a separate Search campaign for the remote buyers who name your place.
- More buyers elsewhere, in a Performance Max or Shopping campaign, which can't be split that way, or on Search when a split isn't worth it: Presence or interest, with a monthly check that the traffic pays.
- A whole country, with every buyer inside it: Presence.
Before you tighten an existing campaign, find out what the people it reaches through interest cost per conversion, using the three-step check in the section on choosing between Presence and Presence or interest. If that cost is within what you can afford to pay for a conversion, many of them are customers.
How Google decides whether a searcher is in your target area
Google Ads keeps two locations for every person who searches. Google's pages call them physical location and location of interest, and this guide uses the same names.
Physical location is where Google believes the person is, or regularly is. Google usually works it out from the IP address of the person's connection, including a Wi-Fi network's address or a mobile carrier's, and, when the person's device settings allow it, from GPS, Wi-Fi, Bluetooth and Google's database of cell towers (Google: About targeting geographic locations). The word "regularly" matters. The Presence option covers people who are in, or regularly in, the places you target, so a commuter can count as present while sitting at home somewhere else. Google doesn't say how often counts as regularly.
Location of interest is a place the person has shown interest in. Google reads it from:
- terms in the search that name or point to a place;
- earlier searches;
- places the person has been;
- the content of the page where an ad appears;
- searches on Google Maps;
- a custom location the person has set for search results.
Location of interest isn't limited to the person's own country or to the Google domain they search on. Google's example is someone in Paris who searches google.fr for a Los Angeles taxi and may be counted as interested in Los Angeles.
Two more behaviors apply under either option. Google may show ads to people in nearby, closely related areas that can't be targeted on their own, as when a Portland target reaches the suburbs. And Google calls all of this its best effort: because the signals vary, it says, no setting is 100% accurate (Google: About advanced location options). Expect a few strays under any setting. The decisions in this guide are about which groups of people you pay for on purpose.
What each control reads
The four controls that decide who gets in each read a different part of that picture. The last two columns matter most: a place named in a search can let someone in under one control, and keep someone out under another.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Control | Where you set it | Reads where the person is | Reads the place they're interested in | Does a place named in the search let someone in? | Does a place named in the search keep someone out? |
|---|---|---|---|---|---|
| Presence or interest, the default option | Campaign | Yes | Yes | Yes, as one signal of interest | No |
| Presence | Campaign | Yes | No | No | No |
| Location exclusion | Campaign | Yes, and keeps out people in the excluded area | No | No | No |
| Negative keyword that names a place | Ad group, campaign, list or account | No | No | No | Yes, wherever the searcher is |
Sources: Google's pages About advanced location options, About targeting geographic locations, Exclude ads from geographic locations and About negative keywords.
Two lines in that table need their reasons. An exclusion keeps out people Google believes are in the excluded area, and nobody else: Google has withdrawn the option to exclude by interest in Search, Display, Shopping and Performance Max campaigns, and its page says people outside an excluded area may still receive your ads. And the "No" in the last column for the two options is an inference: Google doesn't say this directly, but it follows from how Google defines the two options. Presence admits on physical location, and Presence or interest admits if either location is inside your target, so someone standing inside your target is eligible under both, whatever place they search for.
There is a fifth control, but only for Search campaigns that have AI Max for Search turned on, a set of optional settings Google offers in Search campaigns. In those campaigns, each ad group can carry its own locations of interest. When it does, Google considers the place named in the search first, and if the search names no place, it uses where the person is. The person must also meet the campaign's own location targeting. This control can keep out someone inside your area who searches for another place, because the place in their search doesn't match. The section on reaching remote buyers explains when to use it.
Five people search for flowers
To see what the table means for real searchers, take one business, which this guide follows throughout. Ashgrove Flowers is invented for illustration. It has shops in Columbus and Dayton, Ohio, and delivers the same day to 24 cities and towns around the two shops. Its Search campaign, called Delivery, targets those 24 places. Its Performance Max campaign, called Shop, uses its product feed and targets the same 24 places. Both are on the default option, Presence or interest.
Here are five people who might search for flowers today, also invented. Before you read the last two columns, decide which of them each option lets in.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Person | Where they are | What they search | What Ashgrove wants | Presence or interest | Presence |
|---|---|---|---|---|---|
| Dana | Columbus | flower delivery | In | In | In |
| Sam | Seattle; his mother lives in Columbus | flower delivery columbus ohio | In | In: the search names Columbus | Out |
| Priya | Chicago; she was in Columbus last month | flower delivery | Out: she wants flowers in Chicago | Possibly: a past visit is a signal of interest | Out |
| Lee | Lives outside the delivery area; works in Dayton every weekday | florist near me, at home on a Sunday | Out: Ashgrove can't deliver to his home | Likely, if Google counts him as regularly in Dayton | Likely, for the same reason |
| Alex | Columbus | flower delivery cleveland | Out: Ashgrove doesn't deliver to Cleveland | In | In |
The table is what Google's documentation predicts for each person, not a record of what any campaign served.
Read down the last two columns. Switching to Presence keeps Priya out, and it costs Ashgrove Sam, who is exactly the customer a florist wants: someone far away, paying for flowers to be delivered inside the delivery area. Neither option keeps Alex out, because both let in anyone who is in the area, whatever place that person searches for. And Lee gets in under both if Google counts him as regularly in Dayton, and Google doesn't say how it decides that.
The other controls don't change those answers by themselves. Excluding Cleveland from the campaign wouldn't stop Alex, because Alex is in Columbus and an exclusion reads only where people are. A negative keyword, cleveland, would stop Alex, and it would also stop every searcher, anywhere, whose search contains that word. That's fine for a campaign that never wants Cleveland searches, and wrong for one that does.
So the choice between the two options isn't a choice about how strict to be. It's a choice about which strangers you're willing to pay for: people elsewhere who may be buying for your area, or nobody elsewhere at all.
What the location reports can and can't show
The same two locations explain why Google's location reports seem to disagree with each other and with your analytics.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Report | What a row's location means | What it can't tell you |
|---|---|---|
| Targeted locations | One of the locations you target | Anything about people outside your targets |
| Matched locations | The location that matched the ad: a physical location or a location of interest | Where an interest-matched person physically was. Sam appears under Columbus |
| User location | Where the person physically was | Which searches those people made |
| Location type (interest or presence) | Whether the click came through physical location or through interest | Where either kind of person was. An interest-matched person may have been inside your area, such as a local who typed your city's name |
The first two are the Targeted and Matched views of the Locations report, documented on Google's measurement page. The other two need more work, and the section on out-of-area clicks gives the paths and how firmly each is documented. Two questions run through this guide, and they need different reports. "How much are we paying for people outside our area?" is answered from user location. "How much would switching to Presence remove?" has no exact answer in any report. Location type comes closest, but Google doesn't say how it labels a click from someone inside your area whom it also matched through interest, so the interest rows can include locals. Matched locations will put Sam in Columbus every time. And an analytics or session-recording tool counts every visit, including ones that didn't come from an ad, using IP data that can differ from Google's. Google's measurement page gives both of those reasons.
Presence or "Presence or interest": choose for each campaign
Google recommends Presence or interest, the default, as a best practice on Search, and calls it recommended for Performance Max too. Its evidence is one figure: advertisers in travel, real estate and education who switched from Presence to Presence or interest saw 5% more conversions on Search campaigns. The data is Google's own, global, for May 21 to June 1, 2022, and the page gives no cost or value (Google: About advanced location options). Much practitioner advice runs the other way. Mark Meyerson, head of paid media at the agency One Egg Digital, wrote in Search Engine Land in 2025 that "in most cases, it's better to choose Presence."
Read with the two locations in mind, both positions are right for someone. Travel, real estate and education are businesses whose buyers often search from somewhere else: a family planning a holiday, a buyer about to move, a student choosing a university in another country. A plumber's customer is standing in the flooded kitchen. So the option should follow where the person who searches and pays is when they search, and whether their search names your place.
The decision, row by row
Find the rows for your campaign type, along with the rows marked Any, and take the first one that fits. A standard Shopping campaign uses the Performance Max rows: Google says Shopping ads "use your existing Merchant Center product data (not keywords) to decide how and where to show your ads", so they can't confine serving to searches that name the place either. The rows marked Any apply to Shopping campaigns as written, and so does the condition below the table about Shopping offers in other countries. Where a row asks how many of your buyers are outside the area, count them in your own order records; the check below this table says how, and where the line between few and more sits. For a campaign already on Presence or interest, that check, the check before switching to Presence, governs these rows: if every set you could read clears your limit comfortably, don't switch. A business can need different answers for different campaigns.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Campaign | Who searches and pays | Use | Why |
|---|---|---|---|
| Any | Your business is in a sensitive vertical with strict targeting limits | Presence, plus exclusions where compliance requires them | Google names this case itself. Compliance comes before reach and before any sample size, and Presence is still best effort |
| Any | You target a whole country, and the people who buy are inside it | Presence | Interest in a country isn't limited to people inside it. Targeting all the country's regions instead of the country stops only people abroad who search on that country's Google domain, not interest from abroad |
| Search | Few of your buyers are outside the area when they search, or remote buyers don't matter: residents, workers, commuters, local callers | Presence | Interest mostly adds people elsewhere who can't use you. If more buyers than that search from somewhere else and matter, one of the next two rows governs instead |
| Search | More of your buyers search from somewhere else, as with flowers and gifts sent home, tours, venues and property sold to remote buyers, and the test in the next section's "When the split is worth it" says a split is worth it | A local campaign on Presence, plus a separate remote-buyer campaign for the searches that name your place, built as the next section shows | Presence alone on the area blocks the remote buyers, and the default also pays for people elsewhere whose searches don't name you. The search terms report shows how many searches name your place |
| Search | More of your buyers search from somewhere else, and a split isn't worth it, or you can't tell yet | Presence or interest, with the monthly check in the last section of this guide | Only interest reaches remote buyers who don't name the place, and no report separates them from those who do. Splitting into the two campaigns of the row above is optional, and the next section says when it's worth it |
| Performance Max | Few of your buyers are outside the area, or remote buyers don't matter | Presence | Interest would mostly bring strangers in |
| Performance Max | More of your buyers are outside the area, and they matter | Presence or interest, with the monthly check. Presence once a Search remote-buyer campaign takes the searches that name your place, if the check before switching to Presence, below this table, shows this campaign's interest traffic falling short or few of its own buyers remote | Performance Max can't confine itself to searches that name the place: its search themes add signals rather than restrict which searches it serves. A Search remote-buyer campaign takes the searches that name the place, but remote buyers who don't name it reach Performance Max only through interest |
Three conditions sit beside that table.
- The test is where buyers are when they search, not where they must be to buy. A tour operator's customers must be in the city to take the tour, but they book from home. That puts them in the rows for buyers who search from somewhere else, not the row for buyers in the area.
- If splitting off a remote-buyer campaign would leave either campaign with fewer than about 30 conversions a month, or 50 for a campaign that bids on Target ROAS, don't split. Run one campaign on Presence or interest with your place named in the ad text, as Google suggests for cutting clicks from people outside the place who aren't interested in it (Google: Prevent clicks outside of your geo-targeted locations). Google's example is that naming Boston in an ad "minimizes clicks from people not located in Boston and those not interested in Boston". That changes who clicks, not who's shown the ad: your targets and location option still decide who can see it. The numbers come from Google's guidance to judge performance over periods with at least 30 conversions, such as a month or longer, or 50 for Target ROAS (Google: Bidding FAQ). Google publishes no threshold for splitting campaigns, so applying its guidance to a split is our judgment: a campaign below it would have too few conversions for anyone to judge. This guide calls this the volume rule from here on.
- For Shopping ads, in Shopping campaigns or in Performance Max with a feed, the option doesn't keep your offers out of other countries. The section on countries covers what does.
A note on small areas. Google warns that a small radius or a very small location may serve only intermittently or not at all, and the smallest radius it allows is 1 km (Google: Target ads to geographic locations). If a new or small area gets few impressions under Presence, check how many people search for your keywords there before you blame the option, and don't loosen the option to buy volume the business can't serve. A campaign with no impressions at all is a different problem, worked through in our answer on new campaigns that show Eligible but get zero impressions.
Check your own numbers before you switch to Presence
If a campaign is already on Presence or interest, the case for switching is that the people it reaches through interest are wasted money. That's true only if their clicks don't convert at a price you can carry. Before you switch, find out, in three steps.
Find the cost and conversions of the people a switch could remove. Pull two sets of rows for the same period: the location-of-interest rows, and the user-location rows outside your targets. In the interface, both are rows you add in Report editor: a Location type row for the first set, user-location rows for the second; the section on out-of-area clicks gives the paths and how firmly each is documented. Use at least the last 90 days, leaving out the most recent conversion cycle, the usual time between an ad click and the conversion it leads to, so that late conversions have arrived. For the interest rows, the documented route is the Google Ads API, whose geographic view splits results by location type into area of interest and location of presence. In the interface, Ginny Marvin reported in Search Engine Land in 2020 that Report Editor offered Location type as a row; Google's Help pages don't document it. Neither set is exactly what a switch would remove. The interest rows can include people inside your target whom Google matched through interest, and the outside rows include people Google counts as regularly in your area, whom Presence keeps. Treat both as estimates, and if only one is open to you, use that one. Google's pages don't say whether these views include Performance Max campaigns; if yours shows no Performance Max rows, let step 3 decide for that campaign. You should end with cost and conversions for each set you could read.
Compare each set with what you can afford. Your limit is the most you can pay in ads for one conversion. If you set it from a minimum return on ad spend, divide your average order value by that return. Divide each set's cost by its conversions and set the result against your limit. To judge on return instead, you need conversion value for those rows, which has a documented route only in the API. If you use conversion value rules, which raise or lower the value Google Ads reports for some conversions, use values without them. You should end with one comparison for each set: it clears your limit or falls short.
Count where your buyers are, from your own records. If your orders, bookings or leads record where the buyer is (a billing address, a phone number's area code, a form field), count the share that came from outside your service area over the same period. If your records also say which campaign each order came from, count each campaign's share separately. You should end with one share for each campaign. It tells you how much of your business Presence couldn't reach at all.
If every set you could read clears your limit comfortably, don't switch: Presence would cut off customers. Because neither set isolates the people a switch would remove, step 3 decides when either set is close to your limit or falls short. We draw step 3's line at 1 in 10: few buyers outside your area means fewer than 1 in 10 of the campaign's conversions in your records came from outside it, and more means 1 in 10 or more. That line is our judgment. If the shortfall sits in particular places, exclude those places instead of switching. Find them in the outside user-location rows, which show where people were; the interest rows show only the place a person was interested in, which is your own target. Exclusions act where people are, so they take effect under Presence or interest too. Otherwise, with few buyers outside your area, switch. With more, the default is mixing remote buyers with people who will never buy. For a Search campaign, the remote-buyer campaign in the next section can separate them, and that section's two reasons for a split decide whether you build it: interest traffic that falls short, or remote buyers who need their own budget, target or schedule. Without either, keep Presence or interest, with the monthly check. A Performance Max campaign can't be split that way, because it can't confine itself to searches that name your place. Give those searches to a Search remote-buyer campaign instead, which Google's page on Performance Max says is served ahead of Performance Max when a search matches one of its exact match keywords. Then, if the Performance Max campaign's interest traffic falls short, switch it to Presence: that traffic doesn't pay, and the Search campaign keeps the remote buyers who name your place.
Don't decide on a handful of conversions. To measure a set's cost per conversion or return, count its conversions over the whole date range you read, and give no verdict until the set holds about 30, or 50 if the campaign bids on Target ROAS. That's the same Google guidance as the volume rule, and we apply it here too. The reason is noise: with 30 conversions, the 95% margin on the count is about ±36% by our calculation, and order values add more. If 90 days holds too few, widen the range, but not across a date when you changed a location setting.
A set that plainly doesn't pay can be judged sooner, because Google's guidance is about measuring performance, not about spotting a shortfall. Divide the set's cost by your limit: call the result E, the conversions the set would have brought if it converted at your limit. The shortfall is clear when the set's actual conversions are at or below E − 1.65 × √E − 0.5. That line is our judgment. It marks a count that would happen less than 5% of the time if the set really converted at your limit, by a standard approximation that, for any E up to 1,000, errs toward caution by at most one conversion and never the other way. A set that cost 7 times your limit has E = 7, and a line of 7 − 1.65 × 2.65 − 0.5 ≈ 2.1, so 2 conversions or fewer is a clear shortfall and 3 isn't yet. The line E − 1.65 × √E − 0.5 is for an E that comes from your own figures, such as cost ÷ limit. When you compare a count with a measured count, such as a month's orders against the months before a split, the earlier count has its own noise, so use E − 1.65 × √(E × (1 + 1/m)) − 0.5 instead, where E is the before period's monthly average and m is the number of months it averages.
If the widest range you can use still holds too few to measure, and no shortfall is clear, the rows can't decide for you. In that case we'd decide from the decision table and step 3: switch to Presence if your records show few buyers outside your area, by the 1-in-10 line above, and otherwise keep Presence or interest and read the rows each month as they grow. None of this waiting applies to an area the business can't or mustn't serve. That area gets excluded now, whatever the numbers.
Cascader's Opportunity Scan includes a Performance Gaps check that compares parts of an account with a stronger benchmark, among them location presence versus interest, and frames each finding as a change to monitor, a controlled test or something to investigate, without changing any setting.
What Ashgrove found
Ashgrove's new account manager wanted Delivery on Presence. Analytics showed visits from Texas, California and India, and the campaign's clicks from outside the 24 delivery places looked like waste. Here are the three steps on Ashgrove's last 90 days. As with the rest of the account, the numbers are invented to show the reasoning, not taken from a real account.
Ashgrove's average order is $80 and its minimum return is 250%, so its limit is $80 ÷ 2.5 = $32 an order. Both of its campaigns bid on Target CPA.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Delivery campaign, last 90 days | Cost | Orders | Cost per order |
|---|---|---|---|
| Whole campaign | $12,000 | 600 | $20.00 |
| Location of interest rows | $2,400 | 150 | $16.00 |
| User-location rows outside the 24 places | $2,700 | 160 | $16.88 |
Both sets hold far more than 30 orders, and both clear the $32 limit comfortably, at about half of it. Ashgrove's order records explain why. Of 1,180 online orders in the same 90 days, from every channel, 390 (33%) were paid for from an address outside both delivery areas. Those were people like Sam, sending flowers home. Switching Delivery to Presence would give up customers, so Delivery stays on Presence or interest. Whether to split it into two campaigns is a separate decision, in the next section.
The search terms report turned up a problem neither option could touch. Ashgrove had paid $410 for searches naming Ohio cities it doesn't deliver to, such as Cleveland and Cincinnati, and those searches brought 3 orders, all refunded. The search terms report has no location breakdown, so it can't say where those searchers were. The searches are consistent with Alex's case from the five-person table: someone inside the delivery area searching for another city, whom both options let in. The fix is a negative keyword, in the section on keeping areas out.
Reach people who buy for your area from somewhere else
Some businesses' buyers search from somewhere else and name the place they're buying for, as Sam did when he searched from Seattle for flower delivery in Columbus. When the split is worth its cost, the decision table's answer for them is two campaigns. A local campaign, on Presence, serves people in your area. A remote-buyer campaign, also on Presence, targets the regions where your buyers live, excludes your own area, and uses only keywords that name your place. It works because an exclusion reads only where people are. The exclusion sends everyone in your area to the local campaign without blocking anyone elsewhere, and because every keyword in the remote campaign names your place, people elsewhere see its ads only when they search for your place. Google's API documentation describes the same combination in its own example: target a country, exclude one state, and the ads stop showing to people in that state while still reaching people outside it who are interested.
Two facts shape the build. A radius can't be excluded: Google Ads Editor and the API say so, and the web interface's exclusion steps take only named areas such as cities, regions or postal codes. And Google says a campaign's excluded locations shouldn't outnumber its targeted ones.
When the split is worth it
Splitting a campaign that works has costs, so it needs a reason. There are two.
- The check before switching to Presence found the interest traffic falling short, and your records show at least 1 in 10 of your buyers outside your area, the check's line. The split separates the remote buyers who name your place from the strangers the default was paying for.
- The interest traffic already pays, but your remote buyers need something one campaign can't give them: their own budget, their own target or their own ad schedule. A campaign has one of each. If the interest traffic pays and nothing needs to differ, keep one campaign on Presence or interest, because it's already earning its keep.
Either way, the split moves the local campaign to Presence, and that gives up the remote buyers who don't name your place, whom only interest reached in that campaign. The split also moves searches between campaigns. Google's page on Performance Max says a Search campaign is served ahead of Performance Max when a search matches one of its exact match keywords, so the remote-buyer campaign's exact keywords can take searches a Performance Max campaign used to serve. And remote buyers the local campaign no longer reaches can still reach a Performance Max campaign that's on Presence or interest. So judge a split by what it does to the whole business, not by each campaign's cost per conversion. After one to two conversion cycles, compare one month's total conversions and blended cost per conversion, across every campaign that can serve the same searches, Performance Max included, with the monthly average of the months before the split. For a seasonal business, make sure neither period holds a holiday the other doesn't. If the totals hold or improve, keep the split. A fall in total conversions counts only when it's clear by the line in the check before switching to Presence for comparing a count with a measured count, with E the before period's monthly average and m the number of months it averages. After a clear fall, the split has cost more remote buyers than it gained, and merging back into one campaign on Presence or interest is the fix. A smaller fall means keep the split and read another cycle.
The split also has to pass the volume rule beside the decision table. To estimate the remote campaign's volume, divide the current campaign's location-of-interest conversions into months. That's an estimate, not a limit in either direction. Some of those conversions came from locals, who stay with the local campaign, and some from remote buyers who don't name your place, who are lost. The remote campaign can also take more than the estimate: searches naming your place from people Google counts as regularly in your area, and searches that other campaigns had served through interest. If the estimate sits near the threshold, the first month decides. If it's well below, keep one campaign on Presence or interest with your place in the ad text. If the account already runs AI Max on that campaign, its locations of interest are another way to avoid a split, described after the build.
Build the two campaigns
Set the local campaign's target to named places, and switch it to Presence. Target a list of cities or postal codes that approximates your service area. Google Ads takes up to 1,000 locations in bulk, from one country at a time: Campaigns menu › Audiences, keywords, and content › Locations › the pencil icon › choose the campaign › tick "Add locations in bulk." Then set the option: Campaigns › Settings › expand Locations › Location options › Presence › Save. You should see your list under the campaign's targeted locations, and Presence selected. If you'd rather keep a radius, you can, but the remote campaign can then exclude only places wholly inside it, and the two campaigns will overlap at the edges.
Create the remote campaign on Presence, targeting where your buyers live. Target the countries or regions your remote buyers come from. If the exclusion list in step 3 is long, target the country by its regions, so that the excluded places don't outnumber the targeted ones. You should see those countries or regions under the campaign's targets, with Presence selected. Presence is right here because the campaign already targets where the buyers physically are; interest would only add people elsewhere again. That's our inference from the option definitions.
Exclude the local campaign's list from the remote campaign, and from nowhere else. Use Campaigns › Settings › Locations › Advanced search › "Add locations in bulk" › Exclude all › Save. Exclude the same cities or postal codes the local campaign targets. You should see them under "Excluded locations" in the remote campaign only. This is deliberate routing: the local campaign must keep those places, so don't copy the exclusion anywhere else.
Give the remote campaign keywords that contain your place name, in phrase or exact match. A broad match keyword can match searches without its words. Google's example is the broad keyword purple flowers matching the search "pink flowers", so a broad keyword with a place name can match searches that don't name the place. Use the forms of the name people type with your place's qualifier, such as the city with its state. If other places share your place's name, the city alone also matches searches for them. Google gives the same warning about targets: "You don't want to target Paris, France, if you're trying to sell cowboy hats to customers in Paris, Texas." Applying it to keywords is our inference. If your buyers often type the city alone and you add it, add the other places' state names to the remote campaign as negatives, with their abbreviations where an abbreviation isn't also an ordinary word (IN, for Indiana, is one). You should see only place-named keywords in the remote campaign, and those negatives beside them.
Name the place in the ads, with its state if other places share the name. Google suggests putting the location's name in the ad text to cut clicks from people outside it who aren't interested in it. A searcher looking for the other place with your name can then see that the ad isn't for them. You should see the place in every headline set.
Write the ads, and the landing pages, in the buyers' language if it isn't yours. Google has announced that from September 2026, Search campaigns lose their campaign language setting and match ads on the language the ad is written in, so in Search the ad's language is what does the work.
After a full conversion cycle, check that it worked:
- The remote campaign's search terms report (Campaigns › Insights and reports › Search terms) should show searches that all name your place, and none that names another place sharing its name.
- Its user-location rows should all sit outside your area. If they show material traffic inside it, check that the exclusion list covers the local campaign's whole target list. If they show clicks from another place with your place's name, exclude that place in the remote campaign. Expect a small residual: location is best effort, the edges of a city list only approximate a delivery zone, and rows below country level don't always add up.
- The local campaign's search terms should keep showing local searches, and searches for places you don't serve should start to fall once you've added the negatives described in the section on keeping areas out.
The cost of this construction is real. You run a second budget and maintain two keyword lists. The new campaign starts in Smart Bidding's Learning status as a new strategy, which Google says lasts up to 7 days. Google says Smart Bidding can draw on data from all of an account's campaigns, so a new campaign doesn't start from nothing, but its own results take a conversion cycle or two to read.
If the campaign already runs AI Max for Search
An AI Max campaign can do the same job inside one campaign, with the ad-group locations of interest described in the model section. It works only under two conditions:
- the campaign's own location targeting must cover where your remote buyers physically are, because a person must meet both the campaign's targeting and the ad group's locations of interest;
- every ad group in the campaign must carry the locations of interest, or the campaign must exist only for this purpose.
Google's own two examples show the effect. In a campaign targeting the US, an ad group set to Spain matches someone in Chicago searching "hotels in Madrid", and doesn't match someone in Spain searching "restaurants in Chicago." Applied to Ashgrove, an ad group set to its delivery places would take Dana, whose search names no place and who is in Columbus. It would take Sam, whose search names Columbus, only if the campaign's own targeting covered Seattle, where he is; Delivery's 24 places don't. It would leave out Priya and Alex. To set the locations of interest, with AI Max on: Ad groups › Ad group settings for AI Max › Locations of interest › add the places › Save. Locations excluded at campaign level can't be used.
If the campaign doesn't already run AI Max, we wouldn't turn it on only for this. In our judgment, AI Max changes more than location matching, and that's a larger decision than a location setting.
Ashgrove's two campaigns, and why Shop stayed as it was
Ashgrove's check found Delivery's interest traffic paying, so the first reason for a split doesn't apply. The second one does. A third of Ashgrove's orders come from remote senders, and they bunch up before Mother's Day and Valentine's Day, when Ashgrove wants to give them a larger budget without raising the local one. One campaign has one budget, so it can't do that.
The volume estimate comes from the same 90 days. The interest rows held 150 orders, which is 150 ÷ 3 = 50 a month for the remote campaign, and the local side keeps about (600 − 150) ÷ 3 = 150 a month. Both campaigns bid on Target CPA, so the volume rule's number is 30, and both clear it. On Target ROAS the number would be 50, and an estimate of exactly 50 would leave the decision to the first month.
The budget split follows the same rows. Delivery spent $12,000 in 90 days, about $133 a day. The interest rows were $2,400 of that, about $27 a day. So about $27 a day goes to the new remote-buyer campaign and about $106 a day stays local.
Ashgrove built Delivery – local on its 24 delivery places, on Presence. It built Gifts – sent from elsewhere targeting the 50 US states on Presence, with the 24 places excluded. That's 24 exclusions against 50 targets, so the exclusions don't outnumber the targets. Gifts holds phrase and exact keywords for columbus ohio and dayton ohio, plus the city names alone, because senders often type only the city. Both names are shared, by Columbus, Georgia, and Dayton, Texas, among others, so Gifts also carries the negatives georgia, ga, indiana, texas and tx. Its ads say "Same-day delivery anywhere in Columbus, Ohio" or "in Dayton, Ohio."
Ashgrove left Shop's settings alone on the day of the split, so that the first month would judge one decision, not two. With Gifts running, the decision table's Performance Max rows allow Presence for Shop only if its interest traffic falls short or few of its own buyers are remote. Google's pages don't say whether the location-type rows cover Performance Max, so Ashgrove had no interest rows to show a shortfall, and let step 3 of the check decide. Its order system records the campaign each order came from. Over the same 90 days, Shop brought 240 orders for $5,400, $22.50 each, and 75 of them (31%) were paid for from outside both delivery areas, well over the 1-in-10 line. Gifts can reach only the remote buyers whose searches name Columbus or Dayton, and many of Shop's placements, such as YouTube videos and Display sites, aren't searches at all. So Shop keeps Presence or interest, with the monthly check. Priya, in Chicago, is the cost of that choice: Shop can still show her an ad. Leaving Shop's settings alone doesn't keep Shop's traffic still, though. Gifts' exact keywords can take searches Shop used to serve, and remote buyers whom Delivery – local no longer reaches can still reach Shop through interest, so Shop's orders belong in the split's test.
The first month after the split held no holiday, and neither did the 90 days before it. The invented numbers, with each campaign's 90 days divided into an average month:
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Ashgrove's campaigns | Cost | Orders | Cost per order |
|---|---|---|---|
| Before the split: Delivery, average month | $4,000 | 200 | $20.00 |
| Before the split: Shop, average month | $1,800 | 80 | $22.50 |
| Before the split: both together | $5,800 | 280 | $20.71 |
| First month: Delivery – local | $3,150 | 157 | $20.06 |
| First month: Gifts – sent from elsewhere | $800 | 44 | $18.18 |
| First month: Shop | $1,850 | 82 | $22.56 |
| First month: all three together | $5,800 | 283 | $20.49 |
The test is the last row against the third. For the same $5,800, Ashgrove took 283 orders against 280, at a blended $20.49 an order against $20.71. The split cost nothing measurable, and Ashgrove now has the separate budget it split for. Gifts' 44 orders are enough to judge Gifts on its own, and they fell short of the estimate of 50, which means nothing by itself, since the estimate was never a limit. Gifts' search terms all name Columbus or Dayton, and none names Georgia or Texas. Its user-location rows showed 9 clicks from Columbus, Georgia, so Ashgrove excluded that city in Gifts. With E = 280 averaged over 3 months, the line sits at 280 − 1.65 × √(280 × 4/3) − 0.5 ≈ 247.6. A month of 260 wouldn't have been a clear fall. A month of 247 or fewer would have, and then the split would have cost more remote buyers than it gained, whatever each campaign's cost per order, and Ashgrove would have merged back.
Clicks from outside your area: find where they came from
Out-of-area traffic usually arrives as a screenshot: visits in analytics from another country, repeat visitors in a session tool from another state, or a lead from someone you can't serve. Work through these steps in order. Each one rules out a cause before you pay to fix the next.
The diagnosis, in order
Confirm the visits are Google Ads clicks. Analytics and session tools count all traffic, not just ad clicks, and use IP data that Google updates on its own schedule (Google: About measuring geographic performance). Check for Google's click identifier or your own tagged landing-page URLs on the sessions in question. Out-of-area sessions without an ad click aren't Google Ads traffic. Handle them with your site's bot and analytics filtering, not with location settings or Google's invalid-click process, which covers ad clicks. If the ad clicks themselves look automated, our answer on click-fraud blockers covers what Google filters and what a tool can stop.
Read where the ad clicks physically were. Use the user-location rows described below, not Matched locations, which mixes physical locations with locations of interest. Classify each row as inside or outside your targets, by the method in the table below. You should end with the clicks, cost and conversions that came from outside your targets.
If the outside clicks are on Shopping ads, check their cost. Merchant Center says Google might show your offers to people in a country you've excluded or don't target, and clicks are free from a country you don't target in Google Ads (Merchant Center: Set up your target country of sale). Applying that to Performance Max's Shopping ads is our inference. If the outside country's rows show zero cost, you're not paying for them. If they show cost, the country matched your targeting, so carry on to step 4.
If a country target shows clicks from outside the country, check the country-domain rule. Google says a country target may show ads to people searching on that country's Google domain even if their IP address is outside the country. Presence, or targeting all the country's regions instead of the country, keeps ads from them (Google: Prevent clicks outside of your geo-targeted locations).
Check how much came through search partners. Search partners are sites outside Google that show Search ads, and a new Search campaign includes them by default. Google says its location options apply to ads on the Search Network, which includes search partners, so partner clicks should follow your location settings. Google doesn't say that in one sentence, and it doesn't say separately how it locates people on partner sites. In April 2025, Google announced a change to search-partner location targeting meant to stop spend from clicks by people in unexpected locations, without saying how it works or which campaign types it covers. To see partner traffic in the interface, segment a campaign table by "Network (with search partners)". That shows partner clicks, cost and conversions for the whole campaign; Google documents no way to combine it with user-location rows in the interface, while the API's user location view can be split by network. If the outside clicks sit mostly on partners and partners fall short of your limit, turn partners off rather than change a location setting: Campaigns › hover over the campaign › the gear icon › Networks › uncheck "Include Google search partners" › Save. Shopping campaigns have the same setting. Performance Max reports search partners as a channel in its channel performance report and lets you give them their own priority tier and an optional daily budget allocation. Set search partners to Low priority with a 0% budget allocation, and "Google won't spend on that channel", though Google says allocation "works on a best-effort basis". That control is rolling out progressively from October 15, 2026, so it may not be in every account yet.
Name the suspect by option. On Presence or interest, interest is the first suspect, and the location-type rows from the check before switching to Presence estimate it. Under either option, nearby related areas can be served. On Presence, the suspects are people Google counts as regularly in your area, related areas, and IP or carrier geolocation. VPNs are a plausible suspect too, though Google's pages don't mention them.
Decide. If the outside traffic comes from an area the business can't or mustn't serve, exclude it now, under either option: no sample size applies, and Presence being on doesn't make the exclusion unnecessary. If it's a question of whether the traffic pays, run the check before switching to Presence, with its rule on when the rows can give a verdict. If the sub-country rows don't add up to the campaign total, that's expected: some impressions are matched only at country level.
Where to see user location, and how firmly each route is documented
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| What you want | Where | How firmly it's documented |
|---|---|---|
| Results by the country users were physically in | Campaigns menu › Insights and reports › Report editor › the predefined Distance report › remove the default rows › add a "Country/Territory (user location)" row | Google documents the path to this row (Google: About jurisdiction surcharges, as of November 2026). The cost and conversion columns beside it rest on the practitioner captures in the next row |
| Region, metro, city or "Most specific location target (User location)" rows, with Cost, Clicks, Impressions and Conversions | Report editor, adding those rows | A practitioner capture: Ginny Marvin, Search Engine Land, 2020, which also shows Campaign and Ad group rows beside them. Mark Meyerson's 2025 capture in Dashboards shows Clicks and Cost |
| Whether each row was inside or outside your targets | Not in the interface | Classify rows yourself: export the report and compare it with your target list |
| Conversion value by user location, a targeted or not-targeted flag, and a split by network | The Google Ads API's user location view (user_location_view), with its targeting_location field, cost, conversions, conversion value, network segment and region, city, metro and postal-code segments | Documented in the API reference |
| Location type, interest or presence | API geographic view (location_type); Report Editor's Location type row | API documented; interface row per Marvin, 2020 |
If you work only in the interface, Google documents how to reach the country rows, and practitioners' captures show the cost and conversion columns beside them. Below country level, the rows themselves rest on those captures too. You'll be judging on cost per conversion; return by location needs the API. Google's pages don't say whether the API's user-location and location-type views return Performance Max campaigns.
If you target a radius, an export can't be compared with a list of targets. The API's targeting flag classifies each row for you. In the interface, classify by hand: check each row's place against your radius on a map. Places that straddle the edge will come up every month, so decide once whether each counts as inside, write the list down and use it every time, so that your outside totals stay comparable from month to month.
Keep an area out: exclusion, place-name negative or a change of option
Three tools claim to keep people out, and each removes a different group, because each reads a different thing. Choose by who you want to stop.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| You want to stop | Use | Why it works | What it doesn't do |
|---|---|---|---|
| People physically in an area you can't or don't want to serve | A location exclusion, in every campaign that could serve them | Exclusions act on physical location | It doesn't stop people elsewhere who are interested in that area. For Shopping ads, offers can still show in an excluded country |
| People elsewhere who reach you through interest in your target | Presence | Presence admits only on physical location | Under Presence or interest, you can instead exclude the places where those people physically are. An exclusion can't remove "interest" as such |
| People inside your target searching for places you can't serve | A negative keyword naming each such place | Negatives read the words of the search | By Google's definitions, the location options won't stop these searches |
| Your products in a country's Shopping ads | Merchant Center's excluded-country attribute, or removing the country from the data source | The attribute keeps a product out of Shopping ads in the countries named | A location exclusion doesn't do this, for the reason in the first row |
A negative keyword naming your own target is never the fix for interest from elsewhere: it would block the locals who search for your city too.
Rules that decide where an exclusion takes effect
- If an area is both targeted and excluded, the exclusion wins, including nested areas: exclude a state and you exclude the cities you targeted inside it (Google: Exclude ads from geographic locations).
- When the goal is to stop serving an area, exclude it in every campaign that could serve it, or another campaign will take the traffic. A manager who excludes a region only from a non-brand campaign can still serve that region's searches through a brand campaign whose keywords match them. When the exclusion is deliberate routing, as with the remote-buyer campaign, only the routed campaign excludes.
- Under Presence, you don't need to exclude untargeted areas in advance. Presence is still best effort, though: it can serve people Google counts as regularly in your area, nearby related areas and people whose IP or carrier location is wrong. So when the user-location rows show an untargeted area being served, and the business can't or mustn't serve it, exclude it now. Under Presence or interest there's a further reason to exclude: someone physically in an untargeted country can reach you through interest in your target, and Google's page on jurisdiction surcharges recommends excluding such a country if your ads serve there.
- Exclusions take named places, not a radius.
- A campaign that targets "all countries and territories" can carry at most 122 country exclusions, since March 31, 2024.
- Google takes up to 1,000 exclusions at a time in bulk, and says a campaign's exclusions shouldn't outnumber its targets.
Exclude an area
- Open Campaigns › Settings › Locations. Search for the place and select Exclude, or use Advanced search and Exclude. For many places, use Advanced search › "Add locations in bulk" › Exclude all. Save. You should see each place under "Excluded locations."
- Repeat in every other campaign that could serve the area, unless the exclusion is routing for a remote-buyer campaign. You should be able to list every campaign that targets the area and see the exclusion in each.
- After a few days, check the user-location rows for that area. You should see clicks fall to a small residual.
Add place-name negatives for places you can't serve
A negative keyword blocks searches that contain its words in that form, and it doesn't match close variants (Google: About negative keywords). Add each form people use: accented and unaccented spellings are two different negatives, and common abbreviations need their own. Test each against the searches you want, because a place name can sit inside a wanted search. The reach test in our guide to negative keywords is the procedure.
Put place-name negatives at campaign level, or in a shared list attached only to the campaigns that shouldn't serve those searches. A Performance Max campaign takes campaign-level negatives too, and they act on its Search and Shopping ads only; the path is in the interface paths table near the end of this guide. Never put place-name negatives at account level: an account-level negative applies to all eligible Search and Shopping inventory, Performance Max included, and that includes a remote-buyer campaign that needs searches naming your own place.
Ashgrove added phrase negatives for the Ohio cities it doesn't deliver to, at campaign level in Delivery – local and Shop, after checking that none blocked a search it wanted. That's what stopped the spend on Cleveland and Cincinnati searches.
One campaign per area, or several areas in one
Location targeting is a campaign setting, so the question of whether an area needs its own campaign comes down to whether the area needs its own campaign-level decision. Our essay on when two Search campaigns should be one, and one should be two gives the general rule: a campaign line needs an independent decision behind it. These are the location-specific conditions.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| If the area… | Do this | Because |
|---|---|---|
| Must be funded, capped or scheduled independently | Give it its own campaign and budget, if each campaign would pass the volume rule beside the decision table. Below that rule, only a hard business requirement overrides it. Our test for hard: the requirement is set outside the ad account and binds whatever the results show, such as a client contract or budget rule that fixes the area's own amount or cap, a law or licence, or ads that can't run outside set hours, in whatever time zone. Wanting to see, steer or fund an area separately isn't hard, and neither is a new market on its own: run it in the shared campaign and read it in the location reports until it would pass the volume rule | A budget is campaign-level. Google's own example is a separate campaign for a better-performing state |
| Must be switched off at times, such as when a branch is at capacity | Keep it in the campaign, and switch it off when needed. Under Presence or interest, remove it from the targets rather than add a dated exclusion. Add a negative keyword naming it, too. Ashgrove's Dayton week, below, works through each step | Google says a targeting change takes effect at the next auction and never puts the campaign into Learning. Under Presence or interest, an exclusion stops only people "likely to be located in" the area, and Google's API guide says "People outside the excluded areas, even if interested in them, may still see your ads". Removing the area from the targets also stops interest in it from admitting anyone; that's our inference from Google's definition of the option. Searches that name the area, from people still inside the targets, need a negative keyword naming it |
| Is in another time zone, and its ads must run at set local hours | Give it a campaign with its ad schedule offset for the time difference. Below the volume rule, that campaign needs the first row's test for a schedule: the business can't let the area's ads run outside those hours. Use a separate account only if its reporting or billing must be in that time zone or currency | Ad schedules run on the account's time zone, and Google advises adjusting them "to account for any difference in your target customers' time zone". An account's time zone can't be changed once it's set |
| Speaks another language | In Search, write its ads in that language; language alone doesn't need a separate campaign. In Performance Max, the language setting still separates markets on YouTube, Display, Discover and Gmail | Google has announced that from September 2026, Search matches ads on the language they're written in, not on a campaign setting. In Performance Max the setting no longer applies to Search or Shopping ads, but still applies to those other channels |
| Only needs to be seen separately in reports | Keep it in the campaign, and read it in the location reports | Reporting doesn't need a campaign line |
| Needs a different feed label or country of sale | Its own Shopping or Performance Max campaign, whatever its volume | The campaign selects one country of sale or feed label |
| Would leave a campaign below the volume rule | Don't split, unless a hard business requirement forces it, by the first row's test | A campaign below the rule has too few conversions to judge |
If a hard requirement does force a small campaign, remember that the budget caps the area's spend without guaranteeing it. Google rolled out a change between August 17 and 27, 2026: budget-limited Target CPA and Target ROAS campaigns now optimize more consistently toward their target, so a strict target can hold spend below the budget (Google: Bidding FAQ). A small campaign on Manual CPC or Maximize clicks avoids that problem. Shared budgets aren't available to Performance Max, so a Performance Max campaign's budget is always its own.
Ashgrove's Dayton shop at capacity
Before Mother's Day, Ashgrove's Dayton shop runs out of delivery slots for a week, while Columbus still has room. Dayton brings about 25 orders a month across the campaigns (another invented input), below the volume rule, so Dayton doesn't get its own campaign. The limit is on deliveries to Dayton, not on people in Dayton, and that decides which tools work. Before you read Ashgrove's changes, decide which of these four searchers each tool would stop that week:
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Person | Where they are | What they search | What Ashgrove wants that week |
|---|---|---|---|
| Rae | Dayton | flower delivery | Out: she wants flowers delivered in Dayton |
| Alex | Columbus | flower delivery dayton | Out |
| Sam | Seattle, ordering for a friend | flower delivery dayton ohio | Out |
| Jo | Dayton | flower delivery columbus ohio | In: Columbus has room |
Location changes read where people are, so they're the tool for Rae. Alex is in Columbus, which stays targeted, so only a negative keyword, dayton, stops him. Gifts targets Sam's state, so in Gifts only the same negative stops Sam. The negative has to sit in every campaign that could serve their searches. Jo needs a campaign that still admits someone in Dayton who searches for Columbus. For the week, Ashgrove's manager:
- removes the Dayton-area places from the targets of Delivery – local and Shop, which stops Rae. In Shop, which is on Presence or interest, removing them also stops interest in Dayton from admitting anyone, which an exclusion wouldn't do;
- adds the negative keyword dayton to Delivery – local, Gifts and Shop, which stops Alex and Sam. In Shop it covers Search and Shopping ads only;
- removes the Dayton-area places from Gifts' exclusion list, so Jo, who is in Ohio and searching for Columbus, can reach Gifts. That doesn't let Rae back in: Gifts holds only keywords that name Columbus or Dayton, so Rae's search, which names no place, doesn't match, and the dayton negative stops searches that name Dayton;
- puts everything back the following week.
Shop can also reach Jo, through her interest in Columbus. The ads don't control orders, though. Ashgrove's checkout closes Dayton delivery dates for the week, and the ad changes stop Ashgrove paying for clicks that would meet a full calendar. They're operational switches, not changes to evaluate, so they don't wait for a conversion cycle.
Countries: structure, exclusions and Shopping offers
Everything above applies to countries, with three differences: a country target can reach people abroad through the country's Google domain, Shopping offers follow Merchant Center's country rules as well as your targeting, and markets can differ in feed, language and time zone, which the table on one campaign per area covers. Here is a second invented account that shows how they combine.
Halden Audio sells headphones online and ships to the United States and Canada. One Merchant Center data source targets both countries. Halden runs a Search campaign and a Performance Max campaign with its feed, both targeting the US and Canada on the default Presence or interest, and it's about to start selling in the United Kingdom from a second data source with its own feed label. Over 90 days, its user-location country rows show the US and Canada, plus:
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Country | Clicks | Cost | Ad type |
|---|---|---|---|
| India | 210 | $180 | Search text ads |
| Philippines | 95 | $70 | Search text ads |
| Mexico | 60 | $0 | Shopping ads |
| United Kingdom | 40 | $35 | Search text ads |
Halden's decisions, country by country
- India, the Philippines and the UK, on Search. These clicks cost money, so they matched Halden's targeting. Two documented routes explain them: people abroad searching on the US or Canadian Google domain, and interest. Halden can't ship to any of these countries, the UK included until its launch, so it excludes all three now, in the Search campaign and in Performance Max, with no sample size. Presence alone would stop both of those routes. Presence is still best effort, though, as the rules on where an exclusion takes effect, earlier in this guide, say: it can serve people Google counts as regularly in your area, nearby related areas and people whose IP or carrier location is wrong. So an area the business can't serve, which the rows show being served, is excluded now, under either option. Halden also switches the Search campaign to Presence, the decision table's whole-country row, because every buyer is inside the two countries, and interest from countries the rows haven't shown yet would bring more of the same.
- Mexico, on Shopping ads. These rows show no cost: Merchant Center says Google might show offers in a country you don't target, and clicks from it are free. There's nothing to pay for. If Halden didn't want its products shown there at all, the excluded-country attribute would keep them out of Shopping ads in Mexico.
- Performance Max. Every buyer is inside the two countries, so remote buyers don't matter, and the decision table's first Performance Max row gives Presence. That decides who sees Halden's other formats. It doesn't fully control where Shopping ads appear, because of the Merchant Center behavior above.
- Canada. Halden wondered about a separate Canada budget. Canada brings about 20 conversions a month, below the volume rule, and a wanted budget isn't a hard requirement by the first-row test in the table on one campaign per area, so Canada stays in the shared campaigns and gets read in the location reports.
- The United Kingdom in Performance Max. The UK products carry their own feed label, so the UK needs its own Performance Max campaign, whatever the volume. That campaign targets the UK and mustn't exclude it. Once the UK campaigns start, the UK exclusion stays only in the US and Canada campaigns, as routing that sends UK searchers to the UK campaigns. If a data source ever targets several countries, a campaign using it must include every country you want the products to show in.
- The United Kingdom in Search. Halden gives the UK its own Search campaign in the same account, with its ad schedule offset for the time difference. The reason is the table's time-zone row. Halden runs its ads only while its support is staffed, and its UK support is staffed in UK time, so the UK ads can't run outside UK hours. That passes the first-row test for a hard requirement, so it overrides the volume rule, which Halden couldn't show the UK passing anyway, with no UK sales yet; a wanted separate budget alone wouldn't have been enough. The UK campaign's budget caps the UK's spend without guaranteeing it. Halden needs no separate account, because nothing requires UK reporting or billing in UK time or currency, and the language row doesn't apply, because the UK market speaks English.
For a country-level account, the decision table's whole-country row and the check before switching to Presence still govern. When buyers abroad pay for delivery inside your country, as with flowers or gifts sent across a border, use the remote-buyer construction, or keep Presence or interest and run the monthly check on the countries it reaches. For the Merchant Center side of products that show in Merchant Center but not in Google Ads, see our list of Merchant Center beliefs Google Ads doesn't share.
Move spend between areas when Smart Bidding sets the bids
Before Mother's Day, Ashgrove, the invented florist this guide follows, wants more spend in Gifts, its Search campaign for people elsewhere in the US who send flowers into its delivery area. That's why Ashgrove gave Gifts its own budget. Gifts bids on Target CPA. Before you read the table, guess what moves Gifts' spend: a location bid adjustment, a bigger budget, or something else.
Location bid adjustments were the old way to put more money into one area. Under Smart Bidding they aren't supported, and what does work depends on the bid strategy.
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Your bidding | Location bid adjustment | The area's own campaign and budget | Conversion value rule by location |
|---|---|---|---|
| Manual CPC, on Search or Shopping | Works: −90% to +900%, set per campaign | Caps the area's spend | Not a bidding lever here |
| Maximize clicks | Works, by Google's table | Caps the area's spend | Not a bidding lever here |
| Target CPA, Maximize conversions or Target Impression Share | Not supported | Caps the area's spend. Under Target CPA, a strict target can hold spend below the budget, so a less strict target in that campaign lets it spend more | Not used by these strategies |
| Target ROAS or Maximize conversion value | Not supported | Caps the area's spend. Under Target ROAS, a strict target can hold spend below the budget, so a lower target in that campaign lets it spend more | Works, when an area's conversions are worth more or less than you report |
| Performance Max | None exist | Caps; its budget is always its own, and its target, if it has one, works as above | Works, with the value-based strategies above |
Sources: Google's page About bid adjustments and the Bidding FAQ's sections on conversion value rules and budget-limited campaigns.
So for Gifts, the answer is the budget and the target together. Its budget sets the most it can spend, and its Target CPA decides how much of that budget it actually spends: a bigger budget alone may change nothing if the target is holding spend below the old one. How to set and reset a target is the subject of our guide to Target CPA and Target ROAS; this section covers only where the location levers are.
A few details change what you'd do:
- Bid adjustments. The most specific location's adjustment applies to the people in it, and combined adjustments can't go above +900% or below −90%. A −100% location adjustment isn't possible, so to stop an area, exclude it.
- Smart Bidding already uses location. Google says Smart Bidding already uses geography as a signal, and its signals include location intent as well as physical location. If an area converts better and your reporting shows it, Smart Bidding accounts for that already. A value rule is for value your reporting doesn't capture, such as a higher margin in one area. Google notes that some value rule conditions are unavailable for housing, employment or credit ads.
- Priority that changes weekly or faster. Use budgets or dated location changes, not targets. Google advises against several target changes within one conversion cycle, and against frequent budget changes during a campaign's 7–14 day learning phase.
Change location settings without losing the read
Google's Bidding FAQ says that a change to targeting settings, location included, takes effect in the next auction, never removes historical data and never puts a campaign into Learning; Smart Bidding performance then adjusts over your conversion cycle. So a location change resets nothing, but its effect still takes a conversion cycle to show.
For any location change you intend to judge:
- Date it, in a change log outside Google Ads as well as in the account's change history.
- Judge it after one to two conversion cycles, which is Google's repeated guidance, and only once the judged period holds enough conversions to measure, or shows a clear shortfall, by the rule in the check before switching to Presence.
- For Performance Max, allow at least 2–3 weeks. Google gives that advice after significant changes, naming budget and bid changes as its examples; treating a location change the same way is our inference.
- Make one change at a time, and don't change the same setting again before you've judged the last change. Ashgrove left Shop's settings alone on the day of the split, so the first month judged one decision, not two; Shop's orders still went into the test, because the split changed which searches reached Shop. The exceptions are correcting a mistake, such as an exclusion list that missed places, and removing an area the business can't or mustn't serve. Operational switches you don't intend to evaluate, like Ashgrove's Dayton week, sit outside this rule.
Geo tests
A geo holdout turns ads off in some regions to measure what they add. Exclusions act on physical location, so excluding the holdout regions works whichever option a campaign uses. But the holdout leaks unless you exclude those regions in every campaign that could serve them, Search, Shopping and Performance Max alike. If you also want the regions left on to run on Presence, switch them before the test's pre-period, the stretch before the test that its results are compared with, so the test start is one change, not two. For Shopping ads, in Shopping campaigns or in Performance Max with a feed, a country-level holdout can still show offers in the held-out country. The full steps for a Performance Max test are in our guide to which campaign should own a search when Performance Max runs beside Search or Shopping.
What you'll hear, and what to check
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| You hear | Check | What you can say |
|---|---|---|
| "Google recommends Presence or interest." | The cost per conversion of the location-of-interest rows | "It does, on the strength of one 2022 figure for travel, real estate and education. Our interest traffic costs this much per conversion against our limit." |
| "Always use Presence." | Your own records of where buyers are | "A third of our orders are paid from outside the area. Presence would block them." |
| "Just exclude the countries that don't convert." | Whether the business can serve each country, and its user-location rows | "Any country we can't serve gets excluded as soon as the rows show us paying for it, whichever option we're on. A country we can serve gets excluded once its rows show it falls short: on enough conversions to measure, or sooner if its shortfall is already clear by our line." |
| "Matched locations says we're only in our cities." | User location | "Matched shows the place that matched, which can be the place they searched for. User location shows where they were." |
| "Analytics shows traffic from India." | Whether those sessions were ad clicks | "Most of those sessions aren't ad clicks. Of the ad clicks, here's what India cost." |
| "Raise the bid adjustment for Columbus." | The bid strategy | "On Target CPA, location bid adjustments aren't supported. If Columbus needs more spend, it needs its own campaign, with a budget that caps it and a target loose enough to spend it." |
Reference: interface paths and where Google's pages disagree
Interface paths
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| To | Go to | Source |
|---|---|---|
| Change the location option | Campaigns › Settings › expand Locations › Location options › Presence or Presence or interest › Save | Google: Prevent clicks outside of your geo-targeted locations |
| Add location or radius targets | Campaigns menu › Audiences, keywords, and content › Locations › pencil › choose the campaign › enter a place, or Radius with an address and distance › Save | Google: How do I set up location targeting |
| Add targets in bulk | The same, then "Add locations in bulk", up to 1,000 from one country at a time | Same |
| Change locations in several campaigns | Campaigns › Settings › tick campaigns › Edit › Change locations › Preview › Apply | Same |
| Exclude places | Campaigns › Settings › Locations › search and Exclude, or Advanced search › Exclude (bulk: "Add locations in bulk" › Exclude all) › Save; or the Locations page's Location exclusions tab › + Edit locations | Google: Exclude ads from geographic locations |
| See Matched locations | Campaigns › Insights & reports › When and where your ads showed › Matched locations tab | Google: View matched locations and distance reports |
| See user location by country | Report editor › Distance report › remove default rows › add "Country/Territory (user location)" | Google: About jurisdiction surcharges |
| Set AI Max locations of interest | With AI Max on: Ad groups › Ad group settings for AI Max › Locations of interest › Save | Google: Target ads to geographic locations |
| Add negatives to a Performance Max campaign | Campaigns › Keywords › Negative keywords tab › plus button › select the Performance Max campaign › add keywords or a negative keyword list › Save | Google's Help page on negative keywords for Performance Max |
| Turn off search partners in a Search campaign | Campaigns › hover over the campaign › gear icon › Networks › uncheck "Include Google search partners" › Save | Google's Help page on the Search Network and search partners |
| See search-partner traffic | Any statistics table › Segment › "Network (with search partners)" | Google's Help page on segmenting data |
| See location bid adjustments | The Locations page in the Campaigns menu, Bid adjustments column | Google: About bid adjustments |
| See search terms | Campaigns › Insights and reports › Search terms; the Keyword and Match type columns, and Segment › Networks | Google's search terms report help page |
Google moves these menus, and these paths are as of November 2026.
Where Google's pages disagree
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Topic | The disagreement | What this guide does |
|---|---|---|
| The "Search interest" option | Help says it can no longer be set for Search, Display or Shopping; the API guide still describes it as usable | Doesn't advise it |
| Excluding untargeted areas | The exclusion page says there's no need, except when you target by interest and exclude by likely location | Reads the exception as covering today's default, Presence or interest, as two other current Google pages do, and excludes any unservable area the user-location rows show being served, on either option |
| Learning after a location change | One passage says targeting changes never trigger Learning; another advises against frequent changes during a learning phase | Treats them as compatible: a status, and how long to wait before judging |
| Language | The location page says to keep language and location settings consistent; the newer language page says Search loses its language setting from September 2026 | Follows the newer page for Search |
| Which search-partner sites showed your ads | Google's Search Network page says Google Ads gives no site-level detail outside Performance Max; an August 2025 announcement adds search-partner placement reporting for Search, Shopping and App campaigns, under Report Editor › Content suitability | Follows the newer announcement: look there first |
The location check to rerun each month
This is the routine the guide comes down to. Run it for each account that targets an area or a set of countries.
- For each campaign on Presence or interest, rerun the check before switching to Presence on the last 90 days: the location-of-interest rows and the outside user-location rows against your limit, with your own buyer records beside them. Give a verdict only by that check's rule: on enough conversions to measure, or earlier when a shortfall is already clear. When they fall short, follow that check's branches: exclude the particular places the user-location rows show falling short, switch if your records show few buyers outside your area by that check's 1-in-10 line, or, with more, take that check's route for a Search or a Performance Max campaign.
- For each remote-buyer campaign, check two things. Its search terms should all name your place, and none should name another place that shares the name. Its user-location rows should all sit outside your area and outside those other places, apart from a small residual.
- Read every campaign's search terms for places you can't serve, and add campaign-level place-name negatives, Performance Max included, tested against searches you want.
- For every campaign, read the user-location rows outside your targets. Exclude any area the business can't or mustn't serve now, whichever option the campaign is on. For other countries, check the ad type and the cost first: Shopping-ad rows at zero cost are free, and charged rows mean the country matched your targeting.
- Check that each exclusion is where it should be. A stop-serving exclusion belongs in every campaign that could serve the area. When an exclusion is deliberate routing, only the routed campaign excludes.
- Undo any dated operational changes to targets, exclusions or negatives whose reason has passed.
- Write down what you changed, and the date, so next month's read starts from the right day.
Most months at Ashgrove, the check changes nothing. Here are the five people from the start of this guide, searching again with the account as it now stands:
If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.
| Person | Where they are | What they search | Which Ashgrove campaign can show them an ad | Why |
|---|---|---|---|---|
| Dana | Columbus | flower delivery | Delivery – local and Shop | She's in a targeted place |
| Sam | Seattle | flower delivery columbus ohio | Gifts and Shop | Gifts targets where he is, and his search names Columbus. Shop admits him through interest |
| Priya | Chicago | flower delivery | Possibly Shop | Shop still admits interest, so that it can reach remote buyers who don't name Columbus or Dayton |
| Lee | At home outside the delivery area | florist near me | Possibly Delivery – local and Shop | Google may count him as regularly in Dayton, and no setting changes that |
| Alex | Columbus | flower delivery cleveland | None | Delivery – local and Shop carry the cleveland negative, and Gifts excludes Columbus |
Gifts keeps buying searches that name Columbus or Dayton from people in other states, Delivery – local keeps serving the 24 delivery places, and Sam's mother still gets her flowers.