What Google quietly taught marketers about quality control, and the lesson goes well beyond local search.
There is a moment every performance marketer knows. A lead comes in, you pay for it, and within ten seconds you can tell it was worthless. Wrong service. Wrong city. A robocall. For years, the deal across most pay-per-lead platforms was simple enough: you paid for volume, but you kept the right to point at the junk and say, that one does not count. That right was the pressure valve. It kept the relationship between advertiser and platform honest, because the platform had a reason to care about quality it could be charged back for.
In the middle of 2024, Google changed that deal for one of its products, and almost nobody outside the local-services world noticed. I think they should have, because the change is a small, clean case study in something that is now happening across our whole industry. When a platform automates a process, it does not just make the process faster. It quietly decides who is allowed to disagree with the outcome, and how loudly.
The change itself
Google Local Services Ads is the pay-per-lead product that sits at the very top of the search results for home service categories: the plumbers, electricians, HVAC companies, and roofers. You do not pay per click. You pay per lead, meaning per phone call or message from a potential customer. It is a large and growing business, and for the contractors who depend on it, the monthly spend is real money.
Until mid-2024, that product worked the way most pay-per-lead systems do. If you got a bad lead, you filed a dispute, picked a reason, and Google reviewed it. A credit usually followed within a couple of days. Then Google announced it was moving every advertiser to an automated lead-crediting system. Machine learning models would review every lead, decide which ones were invalid, and issue credits without anyone having to ask. The framing was generous: less paperwork, faster reviews, and fairer treatment for the many advertisers who never bothered to dispute anything and were leaving credits on the table.
Read on its own, that sounds like a genuine improvement, and for some advertisers it may well be one. But the announcement carried a quieter clause. The manual dispute path was being removed. In its place was a feedback survey, a form where you rate a lead and explain why you were unhappy. And the survey, by Google’s own description, is an input that helps tune future targeting. It is not a dispute. Filing it does not reliably get your money back.
The part that matters
Here is the detail that turns a routine product update into something worth writing about. Under the automated system, Google stated that two specific categories of bad lead would no longer be credited at all: leads for a job type you do not service, and leads from outside the geographic area you serve.
Sit with that for a second, because those are not edge cases. Ask any contractor what their most common junk lead looks like and they will describe one of exactly those two things. A roofer getting calls for gutter cleaning. An HVAC company two towns over getting a call Google should never have routed to them. These were, for years, the textbook examples of a lead you could dispute and reliably win. They are now, by policy, your cost to absorb.
The official reasoning is that the new models credit the same number of leads or more on average. That may be true in aggregate. But aggregate is doing a great deal of work in that sentence. Average outcomes are cold comfort to the specific advertiser watching a specific category of waste become non-refundable, with the dispute mechanism that used to address it switched off. Throughout 2025, advertisers reported exactly that experience: out-of-area and out-of-category leads arriving with, in their words, no real recourse.
Why this is a marketing story, not a local search story
It would be easy to file this under niche operational trivia for the home services crowd. I would argue the opposite. What happened to Local Services Ads is a preview of a negotiation every marketer is now having with every automated platform they buy from, whether they have noticed it or not.
Automation in advertising is almost always sold on the same two promises: it saves you time, and it removes human inconsistency. Both promises are usually real. The manual dispute process genuinely was tedious, and it genuinely did reward the squeaky wheel over the quiet advertiser. Those were real flaws and the automated system addresses them. None of that is in dispute.
But automation does a third thing that rarely makes the announcement. It moves the definition of quality inside a model you cannot see, audit, or argue with. Before, the question why was this lead not credited had an answer a human had written down. After, the answer is the model decided, and there is no second sentence. The feedback survey gives you a place to express displeasure, but expressing displeasure and having recourse are different things, and the gap between them is precisely where advertiser money now sits.
The lesson is not that automation is bad. The lesson is more specific and more useful than that: when a platform automates a quality-control process, read the announcement for what it removes, not just for what it speeds up.
What good operators are actually doing about it
The contractors and agencies handling this well have not rejected the automated system. You cannot reject it. It is the only system there is. What they have done instead is rebuild, on their own side, the accountability the platform stopped providing. Three habits stand out, and all three generalize to any automated channel.
They keep their own record
When the platform stops adjudicating quality, you have to adjudicate it yourself, which means you have to measure it yourself. The strongest operators log every lead independently of the platform’s dashboard: what it was, whether it was real, whether it became a job. They are no longer asking the platform to grade its own homework. They have their own gradebook, and it is the gradebook that informs every budget decision.
They separate the platform’s number from the true number
A platform that no longer credits your junk leads will quote you a cost per lead that quietly includes the junk. The real number, the cost per lead that was actually usable, is always worse, and only you can calculate it. Operators who track this discover their true cost is often twenty to thirty percent above the figure the dashboard celebrates. That gap is not a rounding error. It is the entire basis for deciding whether the channel is still worth it.
They move waste upstream, from disputes to settings
This is the most important shift, and the most counterintuitive. When you could dispute a bad lead, the bad lead was an annoyance you cleaned up after the fact. When you cannot, every preventable bad lead is permanent spend, so prevention stops being hygiene and becomes the main event. The good operators now spend their energy tightening service-area boundaries and pruning job categories, because that is the only lever left that actually moves the number. The platform took away the mop. The response is to fix the leak.
The broader takeaway
Google did not do anything villainous here. It automated a clunky process, and like most automation it delivered real efficiency alongside a quieter transfer of risk. The advertiser gained time and lost a measure of control, and the announcement understandably led with the half that sounds better.
That trade is going to land on your desk again and again, in product after product, for the rest of this decade. Bidding, budgeting, creative testing, audience selection, and lead quality are all being absorbed into systems that are faster and more consistent than the humans they replace, and also less interrogable. The marketers who stay sharp will not be the ones who refuse the automation or the ones who trust it blindly. They will be the ones who read every one of these announcements with a single quiet question in mind: this is faster, yes, but what did it just become my job to catch?
In local services, the answer arrived in a clause most people skimmed past. In your channels, the clause is probably already there. It is worth finding before it finds you.
Julian Diep is a Paid Acquisition Consultant at Blue Grid Media, where he works with home service businesses on Google Ads and Local Services Ads strategy.
