A search terms report is a table in a search advertising platform that lists the queries people entered before an advertiser's ad appeared, alongside the keyword that matched each one and the money it consumed. It exists because advertisers do not buy queries. They buy keywords, and the platform decides which searches are close enough to those keywords to trigger an ad. The report is the only routine view of that decision. Google Ads and Microsoft Advertising both offer one, and advertisers use it to find wasted spend, to block unwanted queries and to discover new keywords.

What the report contains

Google defines a search term as "a word or set of words a person enters when searching on Google," and a keyword as the word or set of words an advertiser adds to an ad group, according to its help page on the report. The gap between the two is the reason the report matters. A phrase match keyword such as "running shoes" can trigger an ad for "best trail running shoes for women", and an exact match keyword can trigger ads for close variants that share its meaning rather than its words. The Explaining keyword entry describes the split simply: the search term is what a person typed, the keyword is what the advertiser bought.

Each row in Google's report is one search term, with columns for campaign, ad group, impressions, clicks, click-through rate (CTR), cost and conversions. A match type column shows how closely each search term relates to the keyword it triggered, according to Google, with close variants labelled separately. A keyword column, which shows the exact keyword that matched, is hidden by default and has to be switched on. The report draws only on Search Network traffic. Keyword-based placements on the Display Network belong to contextual advertising and surface elsewhere.

Microsoft Advertising exposes the same data through its search query performance report. Its column reference distinguishes BidMatchType, the match type the advertiser bid on, from DeliveredMatchType, the one actually used to serve the ad. Close-variant values for exact and phrase match were added in October 2019, with data back to July 2019.

How practitioners use it

The workflow is called negative keyword mining. An analyst sorts the report by cost, finds queries that spent money without converting, and adds them, or a word within them, as negative keywords so the ad stops appearing for them. The reverse use is keyword discovery: a converting query not yet in the account is added as its own keyword to gain bid control. Both depend on the report showing what people actually searched.

The stakes are measurable. WordStream's analysis of 15,666 accounts, which found accounts with at least one negative keyword converting at 13% a month against 4.6% for those without, also found that a quarter of businesses had added no negatives at all. The sample was self-selected from users of a free grading tool, so the gap shows association, not cause.

Origin and evolution

Google introduced the Search Query Performance report in the AdWords Report Center on July 31, 2007, according to the Inside AdWords blog. It showed only queries for ads that were clicked, covered the Search Network only, and held data from May 2, 2007. The blog warned that its figures would "most likely not match" keyword reports, a caveat that still applies.

The decisive change came in September 2020. Google limited the report to terms "that a significant number of users searched for, even if a term received a click", citing privacy and declining to define "significant". Seer Interactive measured the effect across its clients the next day. According to Seer, the share of search spend with visible search terms fell from 98.7% on August 31, 2020 to 71.0% on September 1, and click visibility fell from 98.3% to 77.9%. Google partly reversed course on September 9, 2021, when it expanded the report to more queries meeting its thresholds, from February 1, 2021 onwards, according to a Google Ads announcement.

Automated campaign types widened the gap. Performance Max launched without a full report, offering search terms insights instead: auto-generated categories of related queries. According to Google's documentation, the insights show clicks, impressions, CTR, conversions, conversion value and search volume, but not cost, and group low-volume or unclassifiable terms into "Other search terms". In January 2025 Search Engine Land reported a "Private search term" label in Performance Max for queries with fewer than 50 unique searches over 90 days, a definition it attributed to Google's explanation. Google announced fuller Performance Max search term reporting on April 30, 2025, weeks after it raised the campaign's negative keyword limit from 100 to 10,000. Google Ads Editor 2.11 added the reports to the desktop tool in November 2025. Microsoft rolled out search term insights covering Performance Max in September 2024.

AI Max and inferred queries

AI Max for Search, announced on May 6, 2025, extends Search campaigns beyond their keyword lists. Google added "AI Max" as a distinct match type in search terms reports on July 2, 2025, and in September 2025 added aggregate rows to the keywords report for traffic from AI-generated broad matches and from matches driven by landing pages. A documentation update dated June 23, 2026 made a landing page view the default and warned that filtering for the AI Max match type understates its traffic because the filter excludes "Other search terms". A separate view pairs each query with the headline and landing page that served.

The more consequential change concerns what a "search term" now is. Google's documentation on ad group prioritisation says that for searches on Lens, AI Mode, AI Overviews and autocomplete, the search term shown in reporting "represents the best approximation of the user's intent." PPC Land reported in May 2026 that the report now shows inferred intent rather than what users typed for those journeys. Chris Chambers, head of paid search at Understory, said negative keyword work becomes "playing whack-a-mole with shadows."

Limitations and disputes

The central dispute is the hidden share: spend that buys queries the advertiser cannot see. Estimates vary widely because studies measure different things. Seer's 2020 figure was 29% of cost. Mark Meyerson of One Egg Digital, writing in Search Engine Land in September 2024, found about 40% of search term clicks and spend hidden across his agency's accounts, with up to 85% on some keywords, and said colleagues reported ranges from 20% to 80%. Doctor Ads, analysing 88 accounts in August 2026, found that the median account placed 44.0% of clicks in the "Other search terms" row of the insights category report, against 29.9% pooled across all clicks. That last figure concerns categories, not the search terms report itself, and cannot be expressed in cost because the category report has none.

Accuracy is a second dispute. Brad Geddes of Adalysis documented identical queries reported under both an exact match keyword and AI Max, and AI Max traffic credited to exact and phrase keywords. Ginny Marvin, Google's Ads Product Liaison, responded that Google was "increasingly determining relevance by inferred intent." Brian Lasonde of PPC Boost published a report showing an exact match keyword containing "hypoallergenic" matching 13 queries that did not contain the word, each labelled as a close variant. The match type labels in the report are therefore Google's own classification, not an independent check.

Regulators have touched the issue without resolving it. In their November 20, 2024 proposed final judgment in the US search monopoly case, the Department of Justice and states sought improved search query reporting and a ban on limiting data export. Judge Amit Mehta's remedies ruling of September 2, 2025 rejected that element, holding that Google will not be required to share "granular, query-level data" with advertisers. Google maintains that thresholds protect user privacy. Critics note that the same company sells the clicks and decides how much of the query data to show, an argument PPC Land set out in its analysis of how Google optimises its own revenue.

Not the same as

  • Keywords report: shows performance per keyword the advertiser bought. One keyword row can aggregate hundreds of search terms, and totals rarely reconcile with the search terms report.
  • Search query: the user-side string itself. A search terms report is a filtered, thresholded and, for some surfaces, inferred account of queries.
  • Search Console performance report: Google's free organic search report for site owners. It covers unpaid results, withholds rare queries as anonymised, according to PPC Land's Explaining Search Console, and has no cost data.
  • Match type: the keyword setting that governs how loosely queries are matched. The report displays its outcome; the setting is configured on the keyword.

Recent developments

As of October 2026, AI Max coverage keeps changing what appears in the report. Smarter Ecommerce found across 383 million EMEA impressions that about 29% of exact match impressions in AI Max campaigns were expanded matches by July 2026, up from roughly zero in early 2025. Google moved the automatic upgrade of Dynamic Search Ads campaigns to AI Max to February 2027. Microsoft announced search term landing page reporting with its AI Max pilot in June 2026 and has begun retiring Predictive Matching in favour of AI Max, with no published effective date.

Timeline

  • May 2, 2007: earliest data available in Google's Search Query Performance report.
  • July 31, 2007: Google announces the Search Query Performance report in the AdWords Report Center.
  • July 2019: earliest data carrying close-variant delivered match types in Microsoft Advertising search term reports, added in October 2019.
  • September 1, 2020: Google limits the search terms report to terms searched by "a significant number of users".
  • September 9, 2021: Google expands the report to more queries meeting its thresholds, for data from February 1, 2021.
  • March 2023: earliest date for custom ranges in Google's search terms insights.
  • September 2024: Microsoft rolls out search term insights covering Performance Max.
  • November 20, 2024: DOJ and states propose improved search query reporting as a remedy.
  • January 2025: "Private search term" label reported in Performance Max.
  • March 2025: Performance Max negative keyword limit rises from 100 to 10,000.
  • April 30, 2025: Google announces search terms reporting for Performance Max.
  • May 6, 2025: Google announces AI Max for Search.
  • July 2, 2025: AI Max appears as a match type in search terms reports.
  • September 2, 2025: Judge Mehta declines to order query-level data for advertisers.
  • September 2025: AI Max aggregate rows added to the keywords report.
  • November 2025: Google Ads Editor 2.11 adds Performance Max search term reports.
  • May 13, 2026: Search Engine Roundtable reports Google's "best approximation of the user's intent" wording.
  • June 23, 2026: Google's AI Max reporting documentation makes the landing page view the default.
  • July 2026: Smarter Ecommerce data shows about 29% of exact match impressions expanded by AI Max.

Summary

  • Who: Google Ads and Microsoft Advertising produce search terms reports; advertisers, agencies and in-house search teams read them; researchers such as Seer Interactive, Smarter Ecommerce and Doctor Ads have measured what they omit; the DOJ and Judge Amit Mehta have addressed query data in the US search case.
  • What: A report listing the user queries that triggered search ads, with the matched keyword, match type, impressions, clicks, cost and conversions, subject to privacy thresholds and, for some surfaces, Google's inferred version of the query.
  • When: Introduced by Google on July 31, 2007; restricted in September 2020; partly expanded in September 2021; extended to Performance Max in 2025 and reshaped by AI Max reporting through 2025 and 2026.
  • Where: In the Google Ads and Microsoft Advertising interfaces, their APIs and Google Ads Editor, with category-level search terms insights alongside.
  • Why: Keywords and queries diverge, so the report is how advertisers see what they paid for, block irrelevant searches and find new keywords; its thresholds and inferred terms are why its completeness is disputed.