Why PMax and Search shouldn't share a scorecard

Open a typical account dashboard and Search, Performance Max, Meta, and LinkedIn often sit in the same table, under the same columns: impressions, CTR, CPC, ROAS. It reads as consistency. It is actually the reason a real problem in any one of them is easy to miss -- because the columns that matter for one campaign type are not the columns that matter for another, and forcing them into the same row hides that.

Two different games

Search is an auction and intent game: someone is already looking for something specific, and the question is whether the account wins that exact moment efficiently against competitors bidding on the same query. Performance Max is a different game entirely -- the campaign type itself decides, largely automatically, which inventory (search, shopping, display, video, and more) the budget reaches, in proportions a single blended ROAS number does not separate out at all.

Why the same KPI misleads both

Judge a Search campaign on a metric built for PMax, or the reverse, and the number technically calculates but stops meaning what it looks like it means. A stable, on-target blended ROAS on a PMax campaign can hide spend quietly drifting toward lower-intent inventory that would look alarming the moment it was isolated -- and a generic dashboard showing one ROAS column for both campaign types has no way to surface that, because it was never built to ask the question.

A worked example

Illustrative case: a Search campaign and a PMax campaign both report a healthy, on-target ROAS this month. Read on the same scorecard, both look equally fine. Split apart: the Search campaign's ROAS is healthy because impression share and conversion efficiency both genuinely improved. The PMax campaign's ROAS is healthy on average only because a shrinking slice of well-converting Shopping inventory is subsidizing a growing slice of lower-converting Display and video placements -- a trend a blended number is specifically built to smooth over, not reveal. Same column, same number, two entirely different underlying stories.

What separate judgment actually looks like

The fix is not a wider table with more columns bolted on. It is refusing to force one comparison in the first place: what tells you a Search campaign is healthy is not what tells you a PMax campaign is healthy, and each deserves to be judged on its own terms before either gets compared to anything else, including its own past performance.

The honest limit

This only holds up if the data behind each campaign type is actually complete. A PMax campaign missing feed or asset-level data does not get a guessed score to fill the gap on the scorecard -- it gets marked as not assessed, which is a more useful answer than a confident-sounding one built over a hole in the data.

The habit worth building is small but consistent: before comparing two campaigns' numbers, check whether they are even playing the same game. If they are not, the comparison was never valid, no matter how similar the columns looked.