When a Search Term Deserves Its Own Keyword
A promising query in the search terms report is a clue, not an instruction. Four contrasting terms show when to add a keyword, fix the ad or page, exclude an intent, or leave a working route alone.
The search terms report has a button that says Add as keyword. It is a very convenient button. So is “Reply all.”
A search term is what someone typed; a keyword is one of the things you ask Google Ads to use when deciding whether your ad can enter a search auction. The report shows that a term has reached you. It does not, by itself, show that the term needs a new keyword. Before promoting a row, ask a more useful question: What would I do differently for the next person who searches this?
Consider a fictional restaurant-equipment retailer that sells new stainless-steel prep tables. Its search campaign already contains broader table keywords. Four terms appear during the same review.
| Search term in the report | What the term reveals | Best next move |
|---|---|---|
| 48 inch stainless prep table | The existing ad and 48-inch product page already answer the search; orders are sound. | Leave it served. Keep watching the query and its wider family. |
| custom stainless prep table | The store can fabricate and quote to a buyer’s dimensions, but the current stock ad and checkout cannot take that request. | Add a focused keyword to a truthful quote ad group, then check what actually serves. |
| stainless prep table with casters | The store sells a caster option, but the landing page makes buyers hunt for it. | Fix the page and ad first. Reassess keyword control afterward. |
| used stainless prep table | The store sells only new equipment; a recorded lead asked for a used table. | Exclude the unwanted intent. Investigate the lead before praising its conversion. |
These are four different answers to the same question, not steps on a conversion-count ladder. Start with the row that looks most like an easy win.
48 inch stainless prep table has made the operator happy. That is a good reason to inspect it, and a poor reason to copy it automatically. The current ad and page already serve the 48-inch buyer. If adding [48 inch stainless prep table] would merely put a gold star beside a successful query, leave the star in the report. A duplicate keyword is not a second bidder competing against yourself, and it may buy no useful control.
Before deciding that this row needs to be “made exact,” check the report’s Keyword column and the ad group that actually served it. The report’s search-term match type describes the relationship between this search and that keyword; it may differ from the match type assigned to the keyword. Google explains the distinction. Look at searches for neighboring sizes, too. If a particular size later needs its own accurate ad or destination, a narrower keyword may help make that treatment eligible. “Already working” is sometimes a complete diagnosis.
Custom stainless prep table is where the button earns its keep. This retailer really does fabricate to order and can quote from a buyer’s dimensions. But the broad table keyword has matched this search to an ad for stock tables ready to ship and a product checkout that cannot take dimensions. A custom buyer needs a quote, not an empty cart.
So the retailer builds a page that explains made-to-order work and asks for dimensions, with an ad promising a fabrication quote. It then adds [custom stainless prep table] to that quote ad group. This one earns a keyword: it can make the quote ad group eligible for matching custom searches. It cannot guarantee the ad or page will serve. Even exact match can include searches with the same meaning or intent; existing keywords may overlap, and Google's keyword priorities, relevance, Ad Rank, campaign eligibility and budget still matter. A keyword alone supplies no ad, page, budget or bidding goal.
At the next review, check the search terms report’s matched keyword and ad group, the quote page’s tracked visits, and whether fabrication can quote the requests. If custom searches still match chiefly through the stock route, the intended route is not serving. If custom buyers reach the quote page but fabrication cannot quote their requests, the route may be working while the offer is not. And if the retailer cannot fabricate or quote at all, the answer reverses: exclude that intent. A keyword cannot manufacture the offer.
Stainless prep table with casters could have its own keyword and still send buyers to the fixed-leg ad and a page that hides casters in a configuration menu: one more keyword, same puzzle. Put the caster choice where a buyer can see it, make the ad’s promise accurate, and examine whether people reach the option and buy. If those buyers later justify a distinct ad or destination, consider keyword control then. The question is not whether the search is relevant. It is whether the experience after the click is doing its job.
Used stainless prep table has a conversion in the report. Open the lead: it asks for secondhand inventory the retailer does not carry. Counting it as a win would let the conversion column appoint itself sales manager. If used equipment is outside the offer, add an appropriately scoped negative keyword and check what neighboring searches it could block. A new positive keyword would invite more of the unwanted search, not exclude it. Negative keywords match differently from positive keywords, so a broad negative chosen in haste can reject more than the single row that prompted it. Google’s negative-keyword guidance explains the scope and match-type choice. Before saving the negative, test it against the searches you want stopped and the wanted searches that share its words: if no version blocks the first without the second, narrow its scope, and if that fails too, record the cost you accept or don’t add it.
There is a fifth decision hiding behind any row: watch. A term with one promising order, uncertain lead quality or a recent landing-page change may deserve attention without yet deserving a new keyword. Compare the business value of the conversions, the cost of the traffic, nearby searches and the time the campaign has had to reveal a pattern. Conversion lag and privacy filtering can make the visible report incomplete; Google notes that low-activity queries may be omitted and that Search terms insights can aggregate them. A missing or thin row is not proof of absent demand. See the report’s stated limits.
A decision you can use at the next review
For the next tempting row, read across three things before touching the button:
- The searcher’s job. What is the person trying to buy or accomplish, and does your business want that job? Check the actual lead or order when the conversion label could conceal a mismatch.
- The current route. Which keyword, ad group, ad and landing page served this search? Does that path already keep its promise? Look beyond the single row to close neighbors that would inherit the change.
- The intended difference. Can you name the truthful ad, page, match eligibility, bid or campaign treatment you would change? If the answer is “I would have this term in my keyword list,” you have named a record-keeping preference, not a customer benefit.
Then choose the move the evidence supports—add and inspect serving, repair the ad or page, exclude, watch, or leave alone—and check the traffic and experience that follow. The test is not the satisfaction of having pressed a button.
The report is full of things people said. A keyword is a decision about what your account is prepared to do when they say them again.