What is a PPC manager for when Google sets the bids? Deciding what counts as a win

Smart Bidding can price each auction. The harder work is deciding what the campaign should learn, what the business can verify, and when a change would only look like work.

A person considers three blank inquiry cards at a desk while an out-of-focus laptop displays performance shapes behind them.

A client, your boss or your own doubt has asked what you're for now that Google sets the bids. Fair question, and a slightly annoying one. Our answer: Google picks the price in each auction. You decide what Google is bidding for, and you keep checking that against what the business actually sells.

Here's a made-up example of what that means. A company that cleans offices runs search ads, and every submitted quote form counts as a conversion. On Monday the conversions column is up. Then the sales team goes through the actual requests. One came from a homeowner. Another came from someone who wants a job cleaning offices, not a company to clean theirs. Both filled in the same form as the facilities manager who might actually sign a contract. Two conversions, no customers.

Then comes the client meeting. Bidding used to be the part of your job a client could see: every bid you raised or cut sat in the change history. Now the client sees the name of Google's bid strategy and asks, "So Google does the bidding. What did you do?" You can point at the conversion count. Sales can describe the people behind it. Which do you lead with?

The client's question stings because the old job gave you feedback you could watch. You chose a bid, moved it and saw traffic respond. That could fool you (a click isn't a customer, and a chart that moved after your change doesn't prove your change helped). Still, it felt like a conversation with the account. Smart Bidding prices each auction using more contextual signals than a person could sensibly combine by hand. That's a genuine release from repetitive guessing, and we won't pretend to miss it. But the feedback went with it. You need a harder way to learn whether your choices made the business better.

In the cleaning account, we'd start by turning down the easy explanation. The campaign isn't necessarily "finding bad traffic," and Smart Bidding isn't necessarily broken: it may be doing exactly what you asked. You made a submitted form a success, even when sales can't quote the job. Google Ads lets advertisers choose which conversion actions are used for bidding, and in a standard goal, the primary conversion action helps define the result the system pursues. If Google gets very good at finding people who fill in forms, the gap between a form and a contract can get expensive. Deciding what counts as a win is the part of the job Google left with you.

The obvious fix is to bid on qualified leads instead. Sometimes that's right. But "qualified" hides an argument you haven't had yet. Does sales reject jobs outside its territory? Jobs below a minimum size? Jobs it could have taken if the schedule weren't full? Does one rep write off a request after a single missed call ("rang once, no answer, unqualified") while another keeps trying for a week? Signed contracts may be truer to the business, but they're slow and sparse, so they may give Smart Bidding little timely evidence. And switching is a real intervention: Google notes that Smart Bidding may need one or two conversion cycles to adapt after a conversion goal change. So if the first week after the switch is quiet, you've learned nothing yet about which goal was right.

So you go where the bid screen can't: through the rejected inquiries, with the sales manager. Not to announce that lead quality matters (everyone in the room knows). You're there to learn what this company will and won't sell, and which rejections are about the offer and which are about a tired Tuesday. If homeowners keep arriving, the landing page may need to say "commercial" before the form, not in an apology on the callback. If the crews only travel to certain areas, the page should name them. Making the page that clear may lower your form count, which makes for an awkward slide. The real test comes later: once enough of those leads have had time to become jobs or not, are you winning more qualified work per dollar spent? If you can't join that data reliably, say so, before a prettier ratio becomes the story.

Then someone in the meeting says, "Just add 'jobs' as a negative." The search terms report does show people plainly looking for janitorial work. It also shows a school-cleaning query containing "jobs," a word a facilities buyer with several jobs to award might use. You can't tell from the report who made that search or whether they asked for a quote, and you can't see the similar searches a blanket exclusion might block. The sales manager wants the word gone before one more applicant fills in the form. You can't call that a clean saving while a possible buyer might type it too.

Sales can't settle it either. That school search isn't sitting in their notes as an inquiry. They can say what would make a school contract quotable and why they rejected the real requests on their desk; they can't tell you who the searcher was or what work was behind those words. So you and the sales manager agree to exclude phrases that are plainly job-hunting and leave the bare word open. The client, whose money pays for any applicant clicks that still get through, accepts that cost because school work could fit what they sell, not because that search has produced a lead. Blocking the word would have been a choice too: any buyer using it would vanish from the next report along with the waste.

Here's the annoying part: with no bid changes to point to, this work is harder to defend. If your report has a section listing the changes you made (say, "Optimizations made"), a target change fits there neatly. A conversation about which leads are real sounds like a meeting about having meetings. But a tweak made mainly to fill that section is a small professional surrender. You hand the client an action to look at instead of an answer to trust, and neither the change history nor the green form count shows whether that trade was wise.

Don't judge the exclusion on next week, either. An applicant can submit today; a facilities buyer may take longer to choose a contractor. So fewer forms next week wouldn't prove that excluding the job-hunting phrases helped. In its Smart Bidding measurement guidance, Google asks advertisers to account for conversion delay and ordinary variation. Later, the client might decide the applicant clicks cost too much to keep "jobs" open, or sales might establish that the company can't take school contracts. Either would reopen the "jobs" decision. With a separate record joining inquiries to quotes and won work, you could learn more about what the account is worth, if the company can make that join at all. We'd rather the account sit with a costly uncertainty than pretend Google settled it.

So your job isn't proving that a human hand touched every auction. It's deciding what counts as a win, and keeping that answerable to the business outside the account. That may mean arguing over a lead, changing an offer, narrowing an exclusion or refusing to read a fresh chart as a verdict. None of it feels as reassuring as moving a bid. Judge each move against the business result it claims to serve, when you can trace that result.

Google can choose the price for the next auction. It can't tell the cleaning company whether another full calendar of quote calls will become another full calendar of work. That answer takes longer. It's still worth having someone whose job is to find out.