Broad match: reporting gaps, audience signals and targeting controls

Correction, 30 September 2026: The previous article asserted a universal less-than-half reporting share without supporting account evidence and mixed broad-match Search with controls documented for other campaign types. Its earlier wording is retained below as withdrawn history.

Measure the reporting gap instead of assuming a percentage

Google's search terms report does not expose every query. That does not establish that fewer than half are visible in every account, or that all hidden queries are poor quality. Review the search terms documentation, then record the date range, filters and metrics used when assessing your own reporting coverage.

Visible clicks or spend are not the same denominator as the number of distinct queries. Do not turn a percentage of reported spend into a percentage of query strings. Preserve that distinction in an audit.

Observation is not a targeting restriction

Google distinguishes Targeting, which narrows reach, from Observation, which does not. First-party audience segments added under Observation can inform Smart Bidding, but they do not become a hard audience boundary merely because they are present. Check the campaign's actual setting. Google's targeting and observation guidance.

Use controls for the campaign you actually have

Google's optimized-targeting documentation lists Display, eligible Video and Demand Gen campaigns in its compatibility section. It should not be used as instructions for a standard Search campaign's broad-match keywords. The same page uses broad language about defaults; the specific compatibility list is the relevant scope. Compatible campaign types.

A practical Search review

• Identify the campaign type, match types, bidding strategy and audience mode.
• Review available search terms and relevant negative keywords without assuming the report is complete.
• Check business outcomes and conversion-goal configuration.
• Document uncertainty and the evidence needed before proposing changes.

For campaigns with AI Max, use the reporting guide to separate individual queries from grouped insights. Adding an audience is not a substitute for checking what the account is actually configured to do.

Historical wording retained

Withdrawn, 30 September 2026: The following headline, excerpt and body document the earlier version, not current guidance.

Broad match search-term reports show less than half the queries that actually triggered your ads

Google stopped reporting all matched search terms in 2020. Most practitioners know this in theory. Fewer have internalised what it means for broad match in 2026, when match logic is driven by intent modelling and audience signals rather than lexical similarity. Negative-keyword lists built from search term reports are incomplete by construction. Audience signals matter more than negatives now.

The search term report in your broad match campaigns is showing you less than half of the queries that actually triggered your ads.

Google stopped reporting all matched search terms in 2020. Most practitioners know this in theory. Fewer have internalised what it means in practice for broad match in 2026, when the match logic is driven by intent modelling and audience signals rather than lexical similarity.

The invisible query set is not random noise. It is directional. The algorithm is expanding toward query clusters it believes are relevant based on who is searching, not just what they typed. You are seeing the queries that converted. You are not seeing the full distribution of what matched.

Why negative-keyword strategy stopped scaling

This creates a specific problem: negative keyword lists built from search term reports are incomplete by construction. You are suppressing what you can see. The queries that matched but did not appear in the report keep running.

The implication is uncomfortable. The most rigorously maintained negative-keyword list on the planet is still operating on partial information. The discipline that worked in 2018, weekly review of every triggered query, has hit a hard limit imposed by the reporting layer. You can do the work perfectly and still miss most of the matched query distribution.

The accounts that manage broad match well do not rely on negatives as the primary control mechanism any more. They layer audience signals first.

If your broad match campaign is well-targeted by audience, the invisible query expansion is more likely to hit in-market users even when the query itself looks loose. The algorithm uses both the query intent and the user context to decide whether to serve. Audience signals shape the user context.

Negatives clean up the visible tail. Audience targeting shapes the invisible one.

What "audience-first broad match" looks like

The setup that actually works on broad match in 2026:

• Layered first-party audiences: customer match lists segmented by purchase history, RFM segment, or product category. Bid adjustments based on segment value, not blanket inclusion.
• In-market and custom intent audiences specific to the campaign goal. Not "all in-market" broadly, but the in-market segments that align with your product.
• Optimised targeting either off or used deliberately with the audience signals as the primary instruction, not optimised targeting expanding the audience back out into the broader pool.
• Demographic exclusions where the historical data shows clear non-converting segments (age bands, household income brackets, parental status, etc.).
• Negative keyword lists at the account level for the obviously irrelevant queries you can see, not as the primary lever.

The change in mental model: audience signals are no longer a "nice to have" layer on top of keyword targeting. On broad match in 2026, audience signals ARE the targeting. The keyword is the seed, but the audience signal is the constraint.

What to actually do

• Review which audience signals are attached to your broad match campaigns this month. If the answer is "none" or "the default optimised targeting", your broad match is genuinely unconstrained.
• Add customer match lists to every active broad match campaign. Even on accounts without much first-party data, the segments you do have are better signal than nothing.
• Set up in-market and custom-intent audiences specific to the campaign's product or service. Generic "in-market for software" is barely better than no audience signal. "In-market for project management software at companies of 50 to 500" is.
• Audit what optimised targeting is doing. If it is on, it is expanding past whatever audience you set. Turn it off if your audience signals are good. Leave it on only if you are deliberately testing audience expansion.
• Stop building bigger and bigger negative-keyword lists as the only control mechanism. They are necessary but not sufficient. The audience layer is where the real governance happens now.

When did you last review which audience signals are attached to your broad match campaigns?

If you want a view on whether your current broad match setup is appropriately audience-governed or running open, book a free audit and we will map the audience layer against the keyword strategy.