For two years, "Performance Max is a black box" was the single most common objection agencies and in-house marketers raised about the campaign type, and Google spent 2025 and 2026 answering it. Search term visibility rolled out. Negative keywords became addable straight from the report. Even accounts spending as little as $50 a day now get access to data that used to require an enterprise-size account manager relationship to even ask about, according to reporting on the rollout.
The conclusion most advertisers are drawing from that is wrong. PMax reporting being more visible is not the same as PMax reporting being complete, and the gap between those two things is exactly where ecommerce brands keep losing budget to campaigns that look like they are working.
What actually changed in the PMax search terms report
Google began rolling out a dedicated search terms report for Performance Max in March 2025, expanding it through 2026 to the vast majority of accounts, according to Search Engine Land's coverage of the update. It lives under Insights and reports in the campaign menu, and Google's own documentation for the report confirms it shows the actual query, which ad format it triggered (Shopping versus text), and standard performance metrics per term.
For the first time, a PMax advertiser can pull up a term, see it converted at a real cost, and decide whether it deserves a negative keyword or a dedicated asset. That is a genuine unlock. A Google Ads audit that used to stop at "PMax is a mystery, trust the ROAS" can now actually inspect what the campaign spent against, the same way it always could for a standard Search campaign.
What still does not show up
Read Google's documentation closely and the limits are stated plainly, not hidden. Google's own search terms report documentation confirms that terms without enough query activity get grouped into subthemes or an "other queries" bucket rather than listed individually, to protect the privacy of anyone whose search was rare enough to be identifying, and that the report only surfaces terms used by "a significant number of people," a bar Google does not publish an exact number for. The PMax-specific version of the report confirms the same exclusion applies here, on top of leaving out shop visits and shop sales conversions entirely. And per Brandography's account of the current state, the data is not yet supported in automated reporting scripts, so pulling it into a recurring dashboard still means manual work every time.
None of that makes the report useless. It means treating it as a sample of what happened, not a ledger of everything that happened, which is a different mental model than the one most advertisers carry over from standard Search reporting.
The number the search terms report still cannot fix: brand contamination
Here is the part visibility alone does not solve. Excluding a brand term as a negative keyword inside a PMax campaign only blocks the Search and Shopping surfaces, the ones that are keyword-matched. Display, YouTube, Gmail, and Discover are not keyword-matched, so a brand-aware audience keeps qualifying for ads on those surfaces even after the exclusion is live everywhere it can technically apply.
The effect on blended cost per acquisition is measurable, not theoretical. Haus, an incrementality-testing company, ran controlled brand-inclusion-versus-exclusion experiments across a set of PMax advertisers and found that excluding brand terms lowered new-customer CAC in every experiment that tracked it, by 19 to 60 percent depending on the account, averaging roughly 40 percent. One featured brand, Caraway, saw the opposite pull in the other direction: including branded queries generated 1.75x more new-customer revenue than excluding them, while Google's own platform reporting overstated that branded lift by roughly 33 percent once checked against actual revenue. The pattern that held across accounts was not "always exclude brand," it was that whichever way a brand went, the blended, brand-included PMax number was not a trustworthy read of prospecting efficiency on its own.
Isolating that number takes a deliberate step PMax's own dashboard will not do for you: segmenting reported conversions by whether the underlying query (where visible) or the asset group is brand or non-brand, and comparing CPA between the two rather than trusting the single blended figure at the top of the campaign.
Why the platform's own conversion count is not the same as contribution
Search term visibility answers "what did the ad show for." It does not answer a harder, more consequential question: how many of the conversions PMax claims credit for would have happened anyway. Media-mix agency Opascope reports that across every account in its portfolio, PMax's native click-based attribution underreports incremental revenue by 50 percent or more, with platform-reported ROAS sometimes reading above 1,000 percent on a campaign whose actual profit-and-loss contribution came in below 100 percent for the same spend. The gap traces to how PMax mixes channels: it takes view-through credit on Display and YouTube that a click-based read misses, while simultaneously claiming last-click credit on Search traffic, including brand searches, that would have converted regardless.
That distinction, attribution versus contribution, is the one the new search terms report was never built to answer, because it is a reporting feature layered on top of an automated bidding system whose core job is still to claim credit for conversions it can plausibly connect to an impression, not to prove which of those conversions were incremental.
Reconciling PMax reporting against what actually landed
The most reliable check does not live inside Google Ads at all. In a delivered tracking and reporting audit, we found platform-reported ROAS of 4.2x on Meta and 3.8x on Google running against blended revenue and spend of roughly 1.9x for the same store in the same window. Both platforms were individually "correct" by their own attribution rules and both were describing a business that, added together, was performing at roughly half the efficiency either dashboard implied on its own.
A blended MER, total revenue across every channel divided by total spend across every channel, cannot double count the way platform-reported ROAS can, because it does not depend on either platform's attribution window. Running that number alongside PMax's own reported figures, every week rather than after a budget decision, is the single habit that keeps a newly visible search terms report from becoming a new source of false confidence instead of an old one.
If your Meta and GA4 numbers already disagree in a way that makes this kind of reconciliation hard to trust, that is worth resolving first, and the same discrepancy shows up constantly between Facebook Ads and GA4 reporting for reasons that have nothing to do with PMax specifically.
Where this leaves an ecommerce PPC account in 2026
The search terms report is real progress, worth checking weekly the same way a standard Search Query Report always deserved a weekly look. Use it to catch obviously irrelevant terms, confirm which asset groups are actually earning their budget, and stop treating "PMax is a black box" as an excuse to skip diagnosis.
What it does not do is replace the two checks that were true before the rollout and remain true after it: isolating brand from non-brand CPA, because a negative keyword does not reach every surface PMax serves on, and comparing platform-reported performance against blended revenue, because attribution and contribution are not the same claim even when they come from the same dashboard. A paid media audit that starts from those two checks, rather than from the campaign's own reported ROAS, is what actually tells you whether Performance Max is earning its budget or just reporting well. For ongoing accounts, that same discipline is what measurement-led ads management is built to run every month instead of once a year.
