We nearly published a number we would have spent the next two years apologising for.
The plan was simple. Pull thirteen months of campaign data across the lead generation accounts we manage, compare Performance Max against Search on cost per lead, publish the table. Almost nobody can run that comparison properly, because you need the same accounts, the same businesses, the same time period and both campaign types running side by side. We had it.
The first number came back at 10.5x. Performance Max was producing leads at $9.76 while Search was producing them at $102.36 in the same accounts over the same period.
That is a headline. It is also, as far as we can tell, mostly false.
This article is the full working: what we pulled, what the number was, why we did not trust it, what happened when we tested it, and the exact steps to check whether the same thing is happening in your account. If you only read one section, read the audit steps near the end. That is the part nobody publishes.
What we pulled
You should not trust a benchmark whose sample you cannot see, so here is ours before any of the numbers.
We pulled a campaign-level report from our manager account on 24 July 2026, covering 1 June 2025 onwards. That is roughly thirteen and a half months, which is long enough that seasonality mostly washes out. It came back as 36 active campaigns across 13 US accounts, carrying $163,732.04 in spend and 6,616 platform-reported conversions. We kept Search and Performance Max and threw out the Demand Gen, Display and Smart campaigns, because there was not enough of any of them to say anything useful.
Every account is lead generation. Most are home services: plumbing, fencing, iron doors, windows and doors, home remodeling, tree removal and land clearing. The rest are professional and health services, which is to say counselling, psychiatry, payroll services and a dental practice.
We left conversion values out entirely. Every account in this set assigns a value to lead actions rather than importing real revenue, so when the export shows a $903 “value” against 22 conversions, that tells you what somebody typed into a settings box thirteen months ago. It does not tell you what the business earned. Publishing ROAS off numbers like that would be worse than publishing nothing.
There is one gap we would rather flag than quietly work around. Our manager account reporting shows more total spend than these 36 campaign rows add up to, with roughly a third of manager-level spend sitting outside this export. We have not worked out why yet. So treat this as $163,732 of campaign spend rather than the whole book of business, and read every number below against that.
We work white label, so no account is named anywhere in this article. Where a specific account matters to the argument, we describe the business type instead.
The number that looked great
Aggregate everything, split by campaign type, and this is what comes out.
| Campaign type | Campaigns | Spend | Clicks | Conversions | Cost per lead | Conversion rate |
|---|---|---|---|---|---|---|
| Performance Max | 12 | $54,101 | 74,794 | 5,545 | $9.76 | 7.4% |
| Search | 24 | $109,631 | 21,093 | 1,071 | $102.36 | 5.1% |
Search cost 10.5 times more per lead than Performance Max.
If you have read any PMax content in the last two years you know exactly what article that number produces. Automation wins. Google’s machine learning finds cheaper conversions than your keyword lists ever could. Move budget.
We have written versions of that article ourselves. The number is real in the sense that it is what the platform reports. It is just not measuring what it appears to measure.
Why we did not publish it
The thing that stopped us was not scepticism about PMax. It was one column.
Conversion rate.
Look at the Performance Max campaigns individually, sorted by how often a click turned into a reported conversion.
| Business | Campaign | Clicks | Conversions | Conv. rate |
|---|---|---|---|---|
| Plumbing | Com PMax | 25,369 | 3,586 | 14.1% |
| Fencing | Aluminum North Region | 20 | 2 | 10.0% |
| Plumbing | Local PMax (market 3) | 7,996 | 533 | 6.7% |
| Fencing | Fencing Main | 2,001 | 117 | 5.8% |
| Plumbing | Local PMax (market 1) | 10,689 | 562 | 5.3% |
| Fencing | Aluminum South Region | 20 | 1 | 5.0% |
| Plumbing | Local PMax (market 2) | 17,629 | 635 | 3.6% |
| Iron doors | PMax North Region | 3,708 | 47 | 1.3% |
| Iron doors | PMax South Region | 4,863 | 51 | 1.0% |
| Home remodeling | Performance Max | 1,409 | 9 | 0.6% |
| Windows and doors | PMax | 671 | 3 | 0.4% |
| Fencing | PMax aluminium | 419 | 0 | 0.0% |
Read the top row again. A plumbing campaign converting 14.1% of its clicks. More than one in seven people who clicked became a recorded lead.
We have managed plumbing accounts for years. Plumbing is a strong lead gen category, genuinely high intent, and a good landing page with a visible phone number will do well. It does not do 14%. Emergency plumbing on a well-optimised page might reach 8 to 10% on the very best days in the very best markets. One in seven, sustained across 25,369 clicks and thirteen months, is not a plumbing conversion rate.
Meanwhile the same table shows a home remodeling PMax campaign at 0.6% and a windows and doors PMax campaign at 0.4%. If Performance Max were simply better at finding cheap leads, that spread would not exist. Something is different about how those accounts count a lead, not about how well the campaigns perform.
We already knew what it was, because we had flagged it in that account months earlier.
The primary conversion action was a phone click.
What a phone click actually is
This is the whole article in one section, so it is worth being precise about the terminology, because Google’s own naming makes it easy to get wrong.
There are three different things people casually call “call tracking” in Google Ads, and they are not equivalent.
Calls from ads. Somebody sees your ad, taps the call asset in the ad itself, and a call connects through a Google forwarding number. Google knows the call happened and knows how long it lasted. You can set a minimum duration before it counts. This is a real conversion.
Calls from a website. Somebody clicks your ad, lands on your site, and calls the Google forwarding number that a snippet on your site swapped in. Again the call connects, again the duration is measurable, again you can set a threshold. This is a real conversion.
Clicks on a phone number on your website. Somebody taps a tel: link on your page. Google records that tap. That is all it records.
The third one is not a call. It is a button press. It fires when someone taps and immediately hits cancel. It fires when a thumb brushes the number while scrolling. It fires when the call rings twice and gets abandoned. It fires when the person is comparing three plumbers and dials all three. It fires again if they tap the header number, then the footer number, then the sticky mobile bar, unless counting is set correctly. On a mobile-heavy site with a sticky call bar, which is standard for home services, one visitor can generate several.
The word “call” appears nowhere in the third one’s actual behaviour. Google calls it a click on a phone number, and it means exactly that.
In the plumbing account, the third type was set as a primary conversion. Everything downstream inherited that definition.
Removing one account
The honest test is straightforward. Take out the account with the conversion action we know is not measuring leads, and run the same comparison on everything else.
| PMax CPL | Search CPL | Gap | |
|---|---|---|---|
| All 13 accounts | $9.76 | $102.36 | Search 10.5x worse |
| Excluding the plumbing account | $49.57 | $97.62 | Search 2.0x worse |
The Performance Max cost per lead moves from $9.76 to $49.57. That is a 5x change to the headline figure from removing a single account out of thirteen.
Search barely moves, from $102.36 to $97.62, because the phone-click inflation was concentrated in PMax rather than spread evenly.
Performance Max conversion rate across the remaining accounts falls from 7.4% to 1.8%. Search sits at 4.8%.
So the corrected picture is not “PMax produces leads at a tenth of the cost.” It is closer to “PMax produced leads at roughly half the reported cost of Search in this sample, on a metric that is still not measuring the same thing in both channels, in accounts that vary enormously.”
That is a much less exciting sentence. It is also the one supported by the data.
Why Performance Max inflates more than Search
The obvious question is why one account’s tracking problem hit PMax so much harder than Search when both campaign types were running in the same account, on the same site, with the same conversion actions.
Four reasons, and they compound.
Placement mix. Search campaigns deliver clicks to your website from the search results page. Performance Max delivers across Search, Display, YouTube, Discover, Gmail and Maps. Maps and mobile local placements put a tap-to-call button directly in front of people whose intent is often “find a number” rather than “hire someone today.” Those surfaces generate phone taps at a far higher rate than a desktop search click does.
Mobile skew. PMax inventory skews heavily mobile. The tel: link problem only exists on mobile. A desktop visitor cannot accidentally tap a phone number, which is why the same conversion action can look perfectly reasonable in a desktop-heavy B2B account and catastrophic in a mobile-heavy home services account.
Volume of cheap clicks. Look back at the click counts. The plumbing PMax campaigns pulled 25,369, 17,629, 10,689 and 7,996 clicks. The Search campaigns in the same account pulled a few hundred to a few thousand each. Cheap display and local traffic arrives in bulk. When a percentage of a very large number turns into phone taps, the conversion count balloons while the cost stays low, and the CPL calculation does the rest.
Smart Bidding closes the loop. This is the part that turns a measurement error into a spending error. Once phone taps are the primary conversion, the bidding system optimises to produce more of them. It learns which placements, times, devices and audiences generate phone taps. Those are not the same signals that generate booked jobs. Over thirteen months, the campaign gets progressively better at the wrong objective, and it does so with the account’s money.
Search suffers from the same conversion action, but the placement mix and the traffic volume limit the damage. PMax amplifies it.
Nobody chose this setting. Google did.
The uncomfortable part of this whole exercise is that the account was not misconfigured through carelessness. It was configured the way Google sets it up by default.
Read Google’s own help documentation for phone number click tracking. It tells you plainly that this feature “only tracks clicks on your phone number, not actual phone calls.” Good. That is honest and it is exactly right.
Then read the setup instructions on the same page. When you create the conversion action, the Goal and action optimization dropdown arrives with “Phone call lead” already selected as the goal and “Primary action” already selected for optimization.
So Google tells you it is not a call, then defaults it to the goal category named after calls, then defaults it to primary so it drives your bidding.
Anyone setting this up in a hurry, and everyone sets these up in a hurry, gets a phone tap as a primary conversion without making a single decision. There is no warning, no confirmation step, nothing that says “this will now determine how your budget is spent.”
The same page carries the counting guidance too: every is best for sales, one is best for leads. It is one line in a long article, easy to skim past, and getting it wrong silently multiplies your conversion count.
If you have ever wondered why so many lead gen accounts have inflated conversion numbers, this is most of the answer. It is not that practitioners do not know the difference between a tap and a call. It is that the platform’s defaults quietly make the wrong choice for you and never mention it again.
What other practitioners find
We are not the first people to notice that PMax conversion counts in lead gen accounts should be treated with suspicion.
Jyll Saskin Gales spent six years at Google before becoming a Google Ads coach, which makes her worth listening to on this. Her position is blunt: “I do not recommend Performance Max for lead gen business.” Her reasoning is not that the campaign type is broken. It is that PMax needs four things to work, and the fourth is full-funnel conversion tracking, which she says most lead gen businesses she sees simply do not have. Without it, the system optimises toward whatever weak signal it can find.
Defined Digital Academy makes the same point more directly, warning advertisers not to use PMax for lead generation at all if the only things being tracked are form fills and phone call clicks. Their recommendation is to build conversion history on Search first, use campaign-specific goals rather than account-wide ones, exclude brand and competitor terms, and feed qualified lead data back through offline conversion tracking before letting PMax loose.
Menachem Ani, who runs JXT Group and writes for Search Engine Land, has argued for years that PMax lead gen failures usually trace back to inputs rather than the campaign type, and that disabling URL expansion, structuring asset groups by service rather than audience, and watching the landing page report are where the fixes live.
The common thread across all three is that nobody credible thinks PMax is inherently bad at lead generation. They think it is unusually punishing when your measurement is weak, because it has more places to find cheap conversions than Search does.
Our data is a fairly literal demonstration of that.
The independent data on phone taps
There is a reasonable objection to everything above, and it deserves an answer: maybe phone taps are a decent proxy for calls. Maybe most people who tap do call. If so, counting them is imprecise rather than wrong.
The best data we have found on this comes from The Data Driven Trades, which analysed Google PPC phone leads in home service businesses and matched them back to CRM records. Two findings matter here.
First, calls placed directly from the search results page matched to new CRM opportunities at 31.4%, while calls placed after someone clicked through to the website matched at 37.8%. So even among genuine connected calls, the ones with less consideration behind them convert to real business less often.
Second, and more striking, 12.8% of the direct-from-SERP calls came from new customers, against 23.4% for calls placed after a website visit. Roughly half the new customer rate.
That is measuring connected calls, not taps. A tap sits a full step below the weaker of those two categories, because a tap has not even established that a call happened. If connected calls with low consideration match to new business at around a third, the match rate for raw taps is necessarily lower, and nobody publishes it because almost nobody tracks it that far.
This is why we do not treat phone taps as an imprecise proxy. They are a different thing.
Three other ways Performance Max inflates conversions
Phone taps are the mechanism we found in our own data, but they are not the only one, and if you audit your account and find your phone conversions are clean, check these before you conclude everything is fine.
View-through conversions. PMax can claim credit when somebody saw an impression and later converted without ever clicking. The view-through window can run up to 30 days. Somebody scrolls past a Display placement, forgets it entirely, searches your brand name three weeks later and fills in a form, and PMax records a conversion. The event genuinely happened. The attribution is the problem.
Brand cannibalisation. Without brand exclusions, PMax will happily serve on searches for your own company name. Those queries convert at very high rates and cost very little, which flatters PMax’s numbers while representing demand you already had. If your PMax cost per lead looks suspiciously good and you have not excluded brand terms, this is the first thing to check.
Junk placements generating junk conversions. PMax’s Display and partner inventory includes low-quality sites, and where there is low-quality inventory there is invalid traffic. Bots that fire form submissions to make their clicks look legitimate are a documented problem in this campaign type, and the resulting spam leads teach Smart Bidding to buy more of the same. Placement exclusions and account-level content exclusions help, though PMax gives you less control here than most advertisers would like.
Each of these produces a number that is real in the sense that the event fired, and wrong in the sense that it does not represent new business. That distinction is the whole game.
How to check your own account
This is the section that matters, and it is the one that almost never appears in articles about PMax performance. Everyone tells you to “make sure your conversion tracking is clean.” Almost nobody tells you which screen, which column, and what a wrong answer looks like.
Here is the actual sequence. It takes about twenty minutes.
Step 1: List every conversion action and its type
Go to Goals > Conversions > Summary.
Add the Conversion action and Category columns if they are not showing. What you are looking for is the Source and the specific action type, not just the friendly name somebody typed.
Names lie. We have taken over accounts with a conversion action called “Phone Calls” that was a tel: click, and one called “Contact” that was a page view of the contact page. Ignore the label. Look at the source and type.
Write down every action and put each one into one of three buckets:
- Real lead: form submission with a thank you page or event, call from ads with duration threshold, call from website with duration threshold, booking widget submission, chat conversation that produced contact details
- Signal, not a lead: phone number click, email link click, directions click, page view, scroll depth, time on page, video view, add to cart on a lead gen site
- Unclear: anything you cannot immediately categorise. Unclear counts as “signal” until proven otherwise.
Step 2: Check which of them are primary
In the same screen, look at the Goal and the primary/secondary designation.
Only bucket one should be primary. Every action in buckets two and three should be secondary.
Secondary conversions still get recorded, still appear in “All conv.”, and are still available for analysis. They just stop driving bidding. You lose nothing by demoting them except the false comfort of a big number.
If a phone number click is primary, you have found your problem and you can stop reading this section. Fix that first.
Do not assume somebody made a bad decision here. As covered above, Google’s setup flow defaults phone number click conversions to primary. In most accounts where we find this, nobody chose it. It simply arrived that way and nobody went back to check.
Step 3: Check the counting setting
Open each primary conversion action and find Count.
For lead generation this should be One, not Every.
“Every” is correct for ecommerce, where one person genuinely can buy three times. For a plumbing lead, one person tapping the number in the header, then the sticky bar, then the footer is one lead who tapped three times. With counting set to “Every” that is three conversions from one person, and the CPL you are reporting is a third of the real one.
We find this misconfigured more often than the phone-click problem itself, and it is quieter because the number it produces is merely wrong rather than absurd.
Step 4: Check the call duration threshold
For any call-based conversion, open it and find the minimum call length.
Google’s default is 60 seconds. That is a reasonable floor for most service businesses. If it has been set to 0 or 15 seconds, every wrong number and every immediate hangup is counting as a lead.
Where you set it depends on the business. For emergency plumbing, 30 seconds may genuinely be enough to book a job. For a legal consultation, 60 seconds barely covers a greeting and we would push it higher. Ask the client how long their shortest real booking call actually takes and set it just below that.
Step 5: Check for a campaign-level goal override
This one is subtle and it is where Performance Max hides.
Open the PMax campaign, go to Settings, and look at Conversion goals.
Campaigns can be set to use account-default goals, or they can override with their own selection. A PMax campaign can be optimising toward a completely different set of conversion actions than everything else in the account, and nothing on the campaign list screen tells you that is happening.
If you have ever looked at a campaign and thought “why is the Conversions column disagreeing with All conv. here specifically,” this is usually why.
Step 6: Segment conversions by action
This is the diagnostic that settles the argument, and it is worth doing even if steps one to five came back clean.
In the campaigns view, apply Segment > Conversions > Conversion action.
Now every campaign’s conversion total breaks apart into the specific actions that produced it. You will see, per campaign, exactly how many came from forms, how many from calls, how many from phone taps.
This is where a campaign that looked like it was producing 600 leads reveals itself as producing 40 form fills and 560 phone taps. Nothing else in the interface shows you that as directly.
Run it across a full year, not a month. Tracking problems are often introduced by a site rebuild, a new form plugin or a tag manager change, and a twelve month view shows you the step change and roughly when it happened.
Step 7: Reconcile against the business
Every step above is inside Google Ads, which means every step above can be internally consistent and still wrong.
Ask the client for their actual lead count for one specific month. Their CRM, their inbox, their job book, whatever they have.
Then compare. If Google Ads says 600 and the client’s CRM says 45, the platform number is not measuring leads, whatever the settings screen says. We have never regretted doing this and we have frequently been surprised by it.
For the account in this study, that reconciliation is what confirmed it.
Making PMax and Search comparable before you compare them
Once your tracking is honest, there is a second problem, and it is the reason we are cautious about even our own corrected 2.0x figure.
Performance Max and Search are not competing for the same traffic, so comparing their cost per lead is not comparing like with like even when both are counting correctly.
Three adjustments we make before treating a comparison as meaningful.
Separate brand from non-brand. Branded search converts at several times the rate of non-brand and costs a fraction as much. Any Search CPL that includes branded traffic is flattered. Any comparison against PMax where PMax is also picking up brand queries is doubly muddled. In this dataset, one account’s branded Search campaign produced 22 conversions at a materially lower cost than its non-brand campaigns, which is exactly what you would expect and exactly why blending them produces a meaningless average.
Check for overlap. PMax will serve on search queries. If your PMax campaign and your Search campaign are both eligible for the same query, whichever wins the auction gets the conversion, and you are not measuring two channels so much as watching an internal competition. Search themes in PMax make this worse, because they function much like custom intent targeting and pull the campaign toward queries your Search campaigns are already covering.
Match the conversion definition. If PMax is optimising toward one goal set and Search toward another, the CPL comparison is arithmetic performed on two different units. Get both onto the same primary conversion actions before you compare, then wait for the learning period to settle before you read the result.
Do all three and the comparison becomes genuinely informative. Skip them and you get a number like 10.5x.
What we actually think about Performance Max now
None of this means PMax does not work. It means the case for it in lead generation is more modest and more conditional than the raw numbers suggest.
Looking at the accounts individually, with the tracking caveat applied, the pattern is not “PMax wins.”
Of the five accounts running both campaign types over the same period, PMax produced a lower cost per lead in three and a higher one in two. In the windows and doors account, PMax cost more than twice as much per lead as Search. In the home remodeling account, both channels were expensive, but PMax’s 0.6% conversion rate over 1,409 clicks says the traffic was not close to purchase intent.
The accounts where PMax performed best had two things in common: enough existing conversion history for the system to learn from, and a genuinely local, high-intent service with a wide geographic footprint. The accounts where it performed worst were higher consideration purchases where somebody does not decide to buy custom iron doors or remodel a kitchen from a Discover placement.
That is not a controversial conclusion. It is roughly what a careful practitioner would have guessed. The point of the exercise was not to reach a surprising conclusion about PMax, it was to find out that our own headline number was manufactured by a settings error in one account.
What this says about benchmark reports generally
There is a broader point here that applies well beyond our thirteen accounts.
Every published Google Ads benchmark is built on platform-reported conversions. Every one of them therefore inherits the conversion tracking configuration of every account in the sample, and nobody auditing thousands of accounts is checking whether each one has a phone tap set as primary.
We found a 5x distortion in a thirteen account sample from a single misconfigured account. In a sample of thousands, that kind of error does not disappear, it averages. It becomes a slightly wrong number that looks perfectly plausible and gets cited for years.
We are not saying published benchmarks are useless. We use them. We are saying that when your account’s cost per lead comes in at three times the published average, the first question is not “what is wrong with my campaigns.” It is “are we counting the same thing?”
Quite often, you are not.
Methodology and limitations
This report covers 36 campaigns across 13 US lead generation accounts running between 1 June 2025 and 24 July 2026, representing $163,732.04 in spend and 6,616 platform-reported conversions. We included Search and Performance Max campaigns with recorded activity. We excluded Demand Gen, Display and Smart campaigns because the volume was too thin to say anything about, and we excluded conversion value and ROAS entirely because the values in these accounts are typed in by hand rather than imported from real transactions.
Cost per lead throughout means campaign spend divided by platform-reported conversions for the same campaigns. Apart from the one account discussed at length above, we have not checked those conversion counts against client CRM records, which is a real weakness and worth holding in mind.
The limitations are worth stating plainly rather than burying. Thirteen accounts is a small sample by any standard. One account carried 53% of the spend and 86% of the conversions before adjustment, which is the whole point of the article, but it also leaves the other twelve individually thin. Our manager account reporting shows more spend than these campaign rows account for, so this is a subset and we have not yet established why. Nearly everything here is US home services in the southeast, so a B2B SaaS account or an ecommerce account would behave differently and none of this transfers to them.
We should also be clear that the corrected 2.0x figure has not itself been audited account by account. Other accounts in this set may have tracking problems we have not found. If they do, the true gap between PMax and Search moves further toward parity rather than away from it, which means our corrected number is more likely to be too generous to PMax than too harsh.
We would rather publish a small honest sample with its problems described than a large one with its problems hidden.
FAQ
Is Performance Max actually cheaper than Search for lead generation? In our sample, after excluding the account with a known broken primary conversion, PMax produced leads at roughly half the reported cost of Search. But that varied enormously by account, and PMax was more expensive than Search in two of the five accounts running both. Anyone quoting a single multiplier is not telling you about the variance.
What is a phone click conversion and why is it a problem? It records somebody tapping a tel: link on your website. It does not know whether a call connected, how long it lasted, or whether the person hung up immediately. On a mobile-heavy site with a sticky call bar it can fire several times for one visitor. As a secondary conversion it is useful signal. As a primary conversion it teaches Smart Bidding to buy taps rather than customers.
How do I know if my account has this problem? Goals > Conversions > Summary, then check the type and the primary designation of every conversion action. Then apply Segment > Conversions > Conversion action in your campaigns view to see what each campaign’s conversions are actually made of. Then compare a month of Google Ads conversions against the client’s real lead count.
Should phone number clicks be turned off entirely? No. Set them to secondary. You keep the data for analysis and you stop them driving bids. Turning them off entirely loses information you might want later.
Why did this hit Performance Max harder than Search? Placement mix, mobile skew, and volume. PMax serves on Maps and mobile local surfaces where tap-to-call is prominent, its inventory is mobile heavy, and it delivers cheap clicks in bulk. A percentage of a very large number of cheap mobile clicks becomes a very large number of phone taps.
Does a low PMax conversion rate mean PMax is not working? Not on its own. The two accounts here with sub-1% PMax conversion rates were both high consideration purchases, custom iron doors and home remodeling, where somebody is unlikely to convert from a Discover or Display placement. Low conversion rate plus high cost per lead plus a considered purchase usually means the campaign type is wrong for the business, not that the campaign is badly built.
What should I fix first if I find several of these? In order: primary conversion designation, counting set to One rather than Every, call duration threshold, then campaign-level goal overrides. Fix the designation first because everything else is downstream of it. Expect a drop in reported conversions and a rise in reported CPL. Neither is a performance decline. You are seeing the real number for the first time.
How long should I wait after fixing tracking before judging performance? Give Smart Bidding a full learning period, and do not compare the fortnight after the change to the fortnight before. The bidding system has been optimising toward the wrong objective and needs time to re-learn. We would look at 30 days minimum before drawing conclusions, longer on lower volume accounts.
Is this Google’s fault or the advertiser’s? Both, but the defaults do most of the work. Google’s own documentation states that phone number click tracking does not measure calls, and the same setup flow then defaults that conversion action to the “Phone call lead” goal and to primary optimisation. An advertiser who accepts every default ends up bidding toward taps without ever making a decision about it.
Do I need offline conversion tracking to run PMax for lead gen? It is the difference between PMax working and PMax being expensive. Practitioners who recommend PMax for lead generation, including Menachem Ani and the team at Defined Digital Academy, consistently condition that recommendation on feeding qualified lead or closed deal data back into the platform. Without it, the system optimises toward whatever is easiest to generate, which is rarely what the business wants.
What if my phone conversions are already clean? Then check view-through attribution, brand cannibalisation and placement quality, in that order. Any of the three can inflate PMax numbers on its own, and all three are easier to miss than a phone tap because the resulting figures look plausible.
Should I just not use Performance Max for lead generation? That is a defensible position and several respected practitioners hold it. Our view is narrower: do not use it until your conversion tracking measures qualified leads rather than website behaviour, and do not compare it to Search until both are optimising toward the same actions. Get those two things right and the question becomes answerable with your own data instead of somebody else’s opinion.
About The Author
