Performance Max will deliver volume. Whether those enquiries are worth answering is a different question, and Performance Max lead quality is the most common complaint I hear about Google Ads right now.
The diagnosis is usually the same. The campaign is doing exactly what it was told, and what it was told was to find people who fill in forms. Filling in a form and wanting to buy are not the same thing. Once in a while the cause is something else entirely, and there is an account further down where I ended up switching it off.
This is for you if your Performance Max campaign looks efficient and your sales team disagrees.
Why Performance Max lead quality goes wrong
Automation optimises towards the signal you give it. Give it a weak signal and it will chase that weak signal with real determination.
The default conversion in most lead generation accounts is a form submission. So the system builds a model of people likely to submit forms, then goes and finds more of them across search, display, video and everything else it has access to.
People who fill in forms readily are not the same population as people who buy. There is overlap, and there is a large group who enquire casually, compare five suppliers, or were never serious at all.
If your CRM does not send closed-won data back to the ad platform, you are asking the algorithm to optimise blind and then blaming it for the result.
Performance Max lead quality starts with the signal
Everything else in this article is secondary to this.
Send conversion data back from your CRM so Google learns which enquiries became customers. Offline conversion import does this, and enhanced conversions for leads makes the matching work better.
Once that is flowing, the system is optimising towards money rather than towards form fills. That single change does more for Performance Max lead quality than every exclusion and control combined.
If you genuinely cannot connect your CRM yet, use a proxy. Fire a separate event only when an enquiry passes a basic qualification check, and optimise on that instead. It is cruder and it is far better than a plain form fill.
The controls that shape Performance Max lead quality
Once the signal is right, shape where it looks.
Account-level negative keywords now exist, and campaign-level lists too. Use them. The early absence of this is precisely why Performance Max earned its reputation for buying irrelevant searches, and plenty of accounts still have empty lists.
Brand exclusions matter more than people expect in lead generation. Without them, the campaign takes credit for people who were already searching your name, which makes its reported cost per lead look excellent while telling you nothing about its ability to find new demand.
Then check your asset groups actually describe distinct things. One catch-all asset group covering every service gives the system a muddled picture of who you want, and muddled input produces muddled output.
Comparing it fairly against search
Performance Max usually looks cheaper than standard search, and the comparison is often rigged in two ways.
It may be absorbing brand traffic, which is the cheapest traffic in any account. And it may be counting easier conversions if your conversion setup differs between campaign types.
Exclude brand from both, use the same conversion definition for both, and compare on cost per qualified lead. That comparison regularly reverses the ranking, and it is the only version worth taking to a budget meeting.
For outside context, WordStream’s 2026 study of 13,474 US search campaigns put the all-industry search average at $66.69 with an 8.18% conversion rate. Any Performance Max figure well below your search figure deserves the brand check before you celebrate it.
The account where I dropped it
The clearest Performance Max lead quality problem I have run into was not a tracking fault, and no amount of signal work would have fixed it.
A UK staffing and childcare business. Performance Max was tested alongside standard search, and what it brought back were people looking for visa sponsorship to come and work in the UK. What the client actually needed were candidates already in the country who could start.
The cost per lead looked reasonable. The enquiries were real people, filling in the form honestly, asking for exactly what they said they wanted. They were simply not who this business could hire.
So it was dropped. Search alone went on to produce 6,070 leads at £11.40 each on that account, in a competitive UK market.
That is the case I would put to anyone who assumes a bad Performance Max result must be a bot problem. Nothing about that traffic was fraudulent. The campaign matched an intent that looks identical to the target intent from the outside, and only somebody who knew the business could tell the two apart.
Nobody has actually measured Performance Max lead quality
Worth saying plainly, because the internet sounds confident about this and the confidence is not earned.
There is no independent, quantified study of Performance Max lead quality anywhere I can find. No disclosed sample, no methodology, no controlled comparison against another campaign type. What exists instead is a set of pages published by click-fraud prevention vendors, every one of whom sells a product whose value depends on the answer being bot traffic.
That does not make them wrong. It makes them interested, and it means the percentages they publish should be read as marketing rather than as findings.
So when this article says the usual cause is a shallow conversion signal, that is one practitioner’s experience across the accounts he runs. It is offered as that, not as a statistic. If somebody quotes you a bad-lead rate for Performance Max, ask them where the number came from.
Run it fed and fenced and it can earn its place. Run it on defaults and it will spend your budget with great efficiency on the wrong people. And sometimes, as above, the honest answer is that this campaign type and this business do not fit, and the right move is to stop paying to find that out again.
Give it time to settle
One practical warning. Changing the signal makes performance worse before it makes it better.
You are asking the system to unlearn a model it spent months building. Expect a fortnight of wobble, and do not reverse the change in week one because the numbers dipped.
That is also why doing this early matters. The longer a campaign optimises towards form fills, the more history there is to undo.
What Performance Max lead quality complaints usually turn out to be
Five causes account for nearly all of them, and only the last is really about the campaign type itself.
The conversion is too shallow. A form fill counts, so form fillers arrive. This is the big one and it is a setup decision rather than a Performance Max failing.
Brand is included, so the campaign looks efficient on people who already knew you, while its performance on new demand goes unexamined.
The negative list is empty, so it is free to buy searches that could never convert. Early versions gave you no way to prevent this. Current ones do, and plenty of accounts have not caught up.
Or the asset group is a catch-all covering every service you offer, so the system has a blurred picture of who you want.
The fifth is the one from the staffing account above, and it is the only one you cannot configure your way out of. The campaign is matching a nearby intent that reads as identical from the outside and is useless to the business. Broad automation is very good at finding adjacent demand, and adjacent is not always wanted.
Genuine algorithmic misbehaviour, once those five are ruled out, is rare in my experience. Almost every complaint I investigate turns out to be one of the above.
What to change this week
Four steps, in order.
Check what your Performance Max campaign counts as a conversion, and whether anything downstream of the form is included.
Set up offline conversion import or a qualified-enquiry event, so the signal points at money.
Add brand exclusions and check your account-level negative list is not empty.
Then re-run your Performance Max against search comparison with brand excluded from both, on qualified leads.
The wider piece on judging lead cost covers the ceiling, reading the search terms report by hand covers the manual work that still pays, and why search costs fell this year gives the market backdrop. For what a properly fed account produces, the case studies have the numbers. If you want this set up rather than explained, Google Ads management covers it, and you can book a teardown.