€55 a lead, €870 a deal: what PMax cost once I counted the closed deals.
Google’s lead report says €55 a lead. The client’s own sheet of closed deals, matched to the ad clicks, says €870 a deal. Same Performance Max campaign, at one of my lease clients, counting ad clicks from mid-July to the end of September.
And the €55 was never the bargain. Most of the main non-brand Search campaigns there ran about €28 to €30 a lead, and together they came out at about €380 a deal. That average leans on one unusually cheap campaign (more on that below), but even next to a plain one at about €370 a deal, PMax cost more than twice as much.
How I counted
In October I traced closed deals on two lease accounts I manage. Each lead stores its ad click ID, or gclid, if it has one. I matched every closed deal that carried one to its click, and every click to its campaign, with no attribution model in between.
About half of all closed deals carry no click ID, and GA4 suggests most of those came through other channels (paid social, organic search, direct, email) rather than lost Google clicks. So every deal figure here counts only the deals that traced back to an ad click.
What the lead count hid
PMax spent about €3,500 there. Google’s report (data-driven attribution) credits it with about 63 leads, 16% of the account’s ad leads, and GA4, which goes by the session the form was filled in, lands on about the same share. So I can’t blame the lead count. Neither tool measures what PMax adds, though.
The trouble is what those leads turned into. PMax took 21% of the spend and 16% of the leads, and its click was on 4 of the 34 deals (12%), Search’s on the other 30. Its leads cost more than Search’s, about €55 against about €35 across the main non-brand campaigns, and they didn’t close any better, so they stayed more expensive all the way to the deal. To be fair, PMax wasn’t the worst campaign per deal, a couple of small Search campaigns cost more.
The lead report flattens Search too. Two campaigns with almost the same cost per lead came out at about €370 and €160 a deal, twins on the lead report and nothing alike at the deal. On 7 and 10 deals, with the cheaper one spotted in this same data, that’s a strong candidate and no more, and I hold PMax’s 4 deals to the same standard.
What I wrote in June
On June 29 I published “Everyone says shift budget to PMax. In lead-gen, I do the opposite.” The gist: in lead-gen, much of PMax’s low cost per lead is borrowed from warm, branded and retargeted traffic that would have converted anyway, so I move its converting search terms into Search and let PMax catch the mixed-intent people I keep out of Search on purpose.
Going into the count, my suspicion was that PMax inflates its lead numbers and leeches off Search, and that the real cost shows up at the deal. The lead numbers held up better than I expected. The cost at the deal did show up, in this account anyway. The leeching this data can’t show, because nothing in it tells me PMax took anything from Search.
The account that argues back
Then there’s another lease account I manage, which is a lot less convenient for my argument. PMax there is a remarketing campaign, aimed at people who were already on the site. Google’s count gives it about 85 leads at about €8 each, and its click was on 4 deals, so about €175 a deal against an account average of about €375.
PMax brought in about a quarter of the ad leads here, by Google’s count and by GA4’s, and its click was on 4 of 26 deals (15%), Search’s on the other 22. It also has the lowest close rate in the account, about 5 deals per 100 leads against about 8 for the main Search campaigns. At €8 a lead it can afford that, which is how it still comes out cheap per deal.
As for those cheap deals, my suspicion (and I haven’t measured it) is that most of those returning visitors first came in through Search, and PMax’s click was simply the last one before the form.
What the numbers can’t show
Small numbers, so this is a first picture and a pattern, not proof.
I also haven’t confirmed whether these sites keep the first or the last ad click. If it’s the last, a late PMax click on someone Search brought in first hands PMax the whole deal and makes it look cheaper. If it’s the first, that late click gets nothing. So a deal count can hide leeching as easily as expose it, which is why I lean least on the second account’s per-deal figure.
The click-ID blind spot can work against PMax too. Some iPhone and app clicks carry a different identifier, one the site never stored. With 4 deals, one extra takes the first account’s PMax from about €870 to about €700. GA4 suggests roughly four deals there lost their click ID, and if all of them were PMax’s it would still land around €440, just above the main Search campaigns at about €380.
What I’d do
I still run PMax, as a layer on top of Search rather than a budget destination. And the only way I know to learn what it adds is a test on the account total: run it at half budget or pause it for about four weeks, and watch whether total leads drop, then whether total deals follow. That test hasn’t been run on these accounts, and budget calls sit with the client anyway, so for now this stays a first picture.
The lead report is where PMax looks its best. The deal is where I judge it.
The client doesn’t get paid per lead. The first account points the way my June edition did, the second argues back, and only a test on the total can settle it. I’ll keep counting.