Can you see search terms in Performance Max?

Yes. Google publishes a report called Search terms and landing pages for Performance Max, under Insights and reports in the campaign menu, with ad format segmentation and landing pages, and you can add negatives from it. Its data goes back to March 2023; nothing I found dates the launch itself.

Why is my Performance Max CTR so low?

Usually because it is not a search CTR. Performance Max serves across Search Partners, YouTube, Display, Discover, Gmail and Maps as well as Search, so its impression count includes events a Search campaign would never have counted. Google warns that these rates can misstate the campaign, though its own example is of rates reading too high.

Should I compare Performance Max CTR with my Search campaigns?

No. A rate computed over mixed inventory is not comparable with a rate computed over search results, and averaging the two gives a number that describes your inventory mix rather than your performance. Compare cost per conversion instead, which is less distorted because spend is spend, though not undistorted: Google notes that some conversions arrive without a click or impression.

Where do negative keywords go for Performance Max?

Two different places. For Search and Shopping inventory, negative keywords at account or campaign level. For Display and Video inventory negative keywords are not the lever: you exclude content keywords in the Content Suitability centre instead.

What is still missing from Performance Max reporting?

Store visits and store sales conversions are not in the search terms report, which matters if footfall is the outcome you are buying: they stay visible in campaign reporting, but you cannot tie them to a search term. Data before March 2023 is absent. And nothing in the documentation I read labels which metrics are safe to compare across campaign types.

Is Performance Max still a black box?

Not in the way the phrase implies. Channel reporting, placement visibility, asset level metrics and search terms all exist now. The remaining problem is not missing data, it is unlabelled data: the reports arrived without the caveat that belongs on them.

Performance Max Reporting: The Missing Warning Label

Same word, two meanings·why the rates stopped comparingEIGHT PAGE-ONE GUIDES · READ 6 SEPTEMBER 2026

CTR in Searchclicks over search results

1,000 ÷ 20,000 = 5.0%

An impression is a search result and a click is somebody choosing your listing over the others on the page. The rate means what everybody assumes it means.

CTR in Performance Maxclicks over everything Google owns

2,000 ÷ 200,000 = 1.0%

The same arithmetic over Search Partners, YouTube, Display, Discover, Gmail and Maps as well as Search. Readable against itself over time; not readable against another campaign type.

The reports arrived: channels, placements, assets, search terms with data back to March 2023. What did not arrive is the label saying the click metrics inside them are computed over different inventory from everything you would compare them with.

What I checked

I opened the thirteen pages Google returns for two phrasings on 6 September 2026. Channable rate limited me twice, so eleven were read, and three of those eleven are Google's own marketing and developer pages rather than guides. That leaves eight advertiser guides, and the counts below are against those eight where the question is about advice.

The coverage is better than the genre's reputation. Four pages reach for the phrase black box and three of them use it about the past. The fourth refuses the framing outright: Refonte Learning writes that "an experienced specialist should resist saying 'Performance Max is no longer a black box'", and runs a section headed what still isn't visible. That is the most careful page in the sample and it is doing the job properly.

The sentence nobody carries

Google's help page on the Performance Max search terms report, under metric prioritisation:

"Prioritize conversion data over clicks and impressions when you evaluate performance max campaign performance. Since these campaigns are automated to drive conversions across various platforms, including YouTube, some conversions may occur without direct clicks or impressions. This can lead to high click-through rates (CTRs) or cost-per-thousand impressions (CPMs) that don't reflect the true value of the campaign."

Note the direction, because I got it wrong myself at first: Google is warning that these rates can read high, not low. And note the scope. That bullet sits inside a section about one report, so read strictly it is a caution about a column in that report rather than a general law about the campaign type. I think the general reading is right, but it is an inference and I would rather label it than let you find it.

What that does when you put two campaigns side by side

This next part is my argument, not Google's. Performance Max serves across Search, Search Partners, YouTube, Display, Discover, Gmail and Maps. A Search campaign serves on search results. Both report an impression count and a click count, and both divide one by the other.

Say Search returned 1,000 clicks on 20,000 impressions, a CTR of 5.0 per cent, and Performance Max returned 2,000 clicks on 200,000 impressions, a CTR of 1.0 per cent. The obvious reading is that Performance Max is five times worse at earning a click. It is not. Most of those 200,000 impressions were never a search result, and a video impression and a search result are not the same event with the same intent.

The rate is not readable across campaign types. It is readable against itself: the same Performance Max campaign month over month, or one asset against another inside an asset group, where the inventory mix is roughly constant. That is the use the per-asset reporting was built for, and it survives everything above.

Cost based figures hold up better, because spend is spend. They are not immune: Google's own sentence flags that some conversions arrive without a click or impression, so the conversion count in a Performance Max campaign is not counting quite the same events either. Less distorted is the honest claim, not undistorted.

Three other limits worth knowing

The search terms report's data starts in March 2023. Google states it: data before that date is not shown. That is a backfill window rather than a launch date, and I have not found anything that dates the launch.

Store visits and store sales are missing from that report. Google says those conversions "aren't available in the search terms report". They still appear in campaign reporting; what you cannot do is connect footfall back to a search term.

Negatives land in two different places. Search and Shopping inventory takes negative keywords at account or campaign level. For Display and Video, negative keywords are not the lever: you exclude content keywords in the Content Suitability centre. Four of the eight guides get you to the first half. None of them states both.

A correction to our own earlier article

In August we published a piece on the search terms report that says, as a heading, that Performance Max does not have this report at all. That was wrong, and it is wrong in a way this article is about: the reporting moved and the writing did not. It has been corrected rather than quietly edited.

Three things to do

Read rates within a campaign type, never across them

Segmenting to campaign type is one dropdown, and it stops the comparison that produces the wrong conclusion. It does not make the number clean, because a single Performance Max campaign is itself mixed inventory.

Judge Performance Max on cost and conversions

That is Google's own instruction on the page that also explains why the click metrics mislead. Clicks stay useful for pruning search terms; they are second order for judging the campaign.

Check both negative keyword surfaces

If you have been adding negatives at campaign level and still see irrelevant placements, you have been fixing half of it.

Where this sits in our own tool

This is the part I got wrong until I read my own source, so it is worth spelling out.

Google Ads Vitals has a page called Channels that splits every rate by campaign type, CTR included. That is the right thing and it was there before I wrote this. The campaign table carries conversion rate and cost per lead per campaign but no CTR column at all, which is defensible given everything above, and which means my own advice to read the rows would not have handed you the metric this article is about.

Two things I would change. The campaign type column is optional, and when it is missing the file guesses the type from the campaign name, so a Performance Max campaign called Brand is quietly filed as Search and then appears in the Search row of that comparison. And the impressions column accepts a header called views, so a YouTube view is read as an impression without a word, which is the inventory mixing in this article happening inside my own parser.

Neither is a fact about your account. Both are decisions in my file, and the fix is a label rather than a rewrite.

Google Ads counts your leads twice by default is the same problem one layer down
The search terms report that does not add up, now carrying the correction above
Auction insights, and what the share numbers are share of
Breakeven ROAS: the one field no calculator asks for

Questions people actually ask

Can you see search terms in Performance Max?: Yes. Google publishes a report called Search terms and landing pages for Performance Max, under Insights and reports in the campaign menu, with ad format segmentation and landing pages, and you can add negatives from it. Its data goes back to March 2023; nothing I found dates the launch itself.
Why is my Performance Max CTR so low?: Usually because it is not a search CTR. Performance Max serves across Search Partners, YouTube, Display, Discover, Gmail and Maps as well as Search, so its impression count includes events a Search campaign would never have counted. Google warns that these rates can misstate the campaign, though its own example is of rates reading too high.
Should I compare Performance Max CTR with my Search campaigns?: No. A rate computed over mixed inventory is not comparable with a rate computed over search results, and averaging the two gives a number that describes your inventory mix rather than your performance. Compare cost per conversion instead, which is less distorted because spend is spend, though not undistorted: Google notes that some conversions arrive without a click or impression.
Where do negative keywords go for Performance Max?: Two different places. For Search and Shopping inventory, negative keywords at account or campaign level. For Display and Video inventory negative keywords are not the lever: you exclude content keywords in the Content Suitability centre instead.
What is still missing from Performance Max reporting?: Store visits and store sales conversions are not in the search terms report, which matters if footfall is the outcome you are buying: they stay visible in campaign reporting, but you cannot tie them to a search term. Data before March 2023 is absent. And nothing in the documentation I read labels which metrics are safe to compare across campaign types.
Is Performance Max still a black box?: Not in the way the phrase implies. Channel reporting, placement visibility, asset level metrics and search terms all exist now. The remaining problem is not missing data, it is unlabelled data: the reports arrived without the caveat that belongs on them.

Olha, the analyst who builds and runs Lucid Vitals

WRITTEN BY
Olha · clinic data analyst

I build the reporting our managers open every morning at a multi-branch medical clinic — and package it so other practices don't have to start from scratch.

Published on 7 September 2026. Three limits worth stating outside the body text. The sample is two search phrasings and one day, and two of the thirteen rate limited me, so rerun the method rather than trust the counts. Performance Max reporting is changing quickly enough that a page written six months ago can be accurate and out of date at once, which is part of the subject rather than an excuse. And the gap I describe in my own $39 file is a gap I could close and have not; writing it here is cheaper than letting a buyer find it.

38ARTICLES WRITTEN

SOURCES8 CITED

The sample is the thirteen pages Google returns on page one for performance max campaigns and performance max reporting transparency what data you get 2026, on 6 September 2026. Channable answered HTTP 429 to two attempts with a pause between them, so eleven were read. Three of those eleven are Google’s own marketing pages and a developer doc for scripts rather than advertiser guides, so the counts about advice are against the remaining eight and the excluded three are named here rather than folded in.
Two corrections I made before publishing rather than after, both to my own first pass. I had counted six pages as saying where negative keywords go; two of them do not mention negative keywords at all and were matched on placement exclusions instead, so the number is four. And the paragraph about the black box named the wrong page: ALM Corp never uses the phrase, only "opaque", and about the Search Partner Network rather than Performance Max. The page that does use it and that I had missed is Benly.
The four that reach for the phrase: Relevant Audience ("it traditionally operated like a black box"), Strike Social ("beyond the platform’s historical black box reputation"), Benly and Refonte Learning. Three use it about the past. Refonte Learning does not, and is the most careful page in the sample: "an experienced specialist should resist saying 'Performance Max is no longer a black box'", followed by a section headed what still is not visible.
One of the eight names the search terms report as a working tool: Benly, with "add negatives directly from the search terms report with one click". DataFeedWatch mentions query visibility only to call it "not as useful as a traditional search terms report" and does not name the report, so it is not counted. The others in the sample are Search Engine Journal, Affect Group and ALM Corp.
None of the eight warns that the click metrics can misstate the campaign. Three mention CTR and none attaches that caveat: one lists click-through rate by placement among the new granular metrics, one lists it as a per-asset metric to read, and one notes only that attribution at placement level remains limited. Everything quoted from Google comes from the help page on the Performance Max search terms report and is checked against a saved copy before this article will build.
Three things worth stating about that Google quotation. It is reproduced whole rather than spliced, because the causal clause matters. Its direction is that these rates read too high, and the worked example in this article is about a low CTR, which is my argument and is labelled as mine. And the bullet sits inside a section about one report, so reading it as a general caution about the campaign type is an inference; the article says so.
The description of my own tool is read out of its source rather than its sales copy, and the first version of this section was wrong in three places: the campaign type column is optional and the file guesses the type from the campaign name when it is missing, the campaign table has no CTR column but does carry conversion rate and cost per lead, and the Channels page does split rates by campaign type, which I had said it did not.
The related pieces referenced above: conversion counting, the search terms report, auction insights and breakeven ROAS. The second of those carried a section saying Performance Max has no search terms report; it has been corrected in the same deploy as this article rather than quietly edited.