Fix: Performance Max search terms report not reviewed on a weekly cadence

Performance Max regained a usable search terms report in March 2024, but Google did not ship the operating model that goes with it [1][3]. Most accounts treat PMax search terms as a quarterly curiosity rather than a weekly hygiene loop — which is the practical reason wastage queries accumulate inside PMax for months before anyone notices. This rule checks whether the operator is actually running the cadence.

Why this matters

Before March 2024, PMax exposed only aggregated "search categories" — broad groupings that named the intent space but never named the actual query, which meant operators could not write targeted negatives. The March 2024 release restored query-level visibility for terms above a privacy threshold, with the rest still rolled into the aggregated view [1][3]. This was a structural unblock, not a hygiene tool: the report sits in Insights alongside diagnostic panes most teams open monthly at best.

The hidden trap is the difference between search themes and search terms. Search themes are an audience signal — you give PMax a phrase like "enterprise crm migration" and the model uses that as a hint about which auctions to enter, not as a hard targeting constraint [1]. PMax will still serve adjacent queries, off-intent close variants, and even brand competitors unless you explicitly block them. Operators who treat themes as targeting wake up to PMax bidding on "free crm tools" for a paid SaaS — the model interpreted "crm" loosely and the absence of a negative meant nothing stopped it.

The cost profile compounds because PMax does not show match type per query, does not separate Search-slice wastage from Shopping-slice wastage in the same table, and silently drops sub-threshold queries into the aggregated bucket where they are uncountable. A campaign can leak 10-15% of its Search slice to off-intent queries that never individually exceed the visibility threshold, so nothing looks alarming on any single line of the report — only the cumulative shape, week over week, reveals the leak (practitioner benchmark, not Google-published).

Account-level negative keywords (cap 1,000) cover the universal cases — jobs/careers/salary/free/torrent + competitor brand tokens — and apply across Search, PMax, App, Shopping, Smart, and Local campaigns simultaneously [2]. Campaign-level negatives, which landed for PMax in Q1 2026 with a 10,000-per-campaign cap, handle campaign-specific drift [3]. The rule this article describes does not test caps or attachment — it tests whether the operator is actually pulling the report on a recurring basis and seeding negatives at all.

How to verify the issue

1. Open Campaigns → pick a Performance Max campaign live > 30 days → Insights → Search terms. If the report is empty or only shows a single category-level row, you are looking at a campaign that has not generated enough Search-slice volume to surface terms — re-check the next-largest PMax campaign instead.
2. Note the date range the report defaults to. Most teams leave it on "Last 7 days" — that is the correct review window. Switch to "Last 30 days" only for a sanity check, then back to 7 days for the workflow.
3. Pull the change history: Tools → Change history, filter by Negative keywords and by the specific PMax campaign, last 30 days. Fewer than 4 negative-keyword additions in the last 30 days for a PMax campaign that has > $1k/month Search-slice spend = cadence gap confirmed (practitioner benchmark, not Google-published).
4. Cross-check account-level negatives: Admin → Account settings → Negative keywords. Confirm a list of at least 20-50 universal exclusions is present and not at the 1,000 cap [2]. An empty account-level list means even the universal guardrails are missing.
5. For PMax campaigns rolled out before Q1 2026 (most accounts): also check Campaign → Keywords → Negative keywords for campaign-level negatives. If this is empty and the campaign carries > $5k/month, the structural pattern is "operator never touched the campaign-level slot" — see PMax campaign-level negative keywords for what to seed first.
6. Confirm the operator can articulate the signal-not-targeting distinction. If the response to "what do your search themes do?" is "they tell PMax what to target", retrain before the next session — the workflow below will surface drift the operator does not expect because themes are not constraints.

How to fix it

Total time: 60-90 minutes for the first sweep, then a 30-minute weekly cadence per active PMax campaign.

1. Set the calendar reminder first. Before doing any analysis, create a recurring 30-minute weekly block — Monday morning is the practitioner default. Assign an owner. Without a calendar reminder and an owner name, the cadence regresses within 6-8 weeks (practitioner observation, not Google-published).

2. Pull the trailing 7 days of PMax search terms per active campaign. Campaigns → select PMax campaign → Insights → Search terms → date range "Last 7 days". Export to CSV. Repeat per PMax campaign that has > $500/week Search-slice spend; smaller campaigns can move to a biweekly cadence.

3. Bucket queries into four intent buckets. Read down the impression-sorted list and tag each query:

   • wanted — on-brief queries that are converting or plausibly will. Leave alone.
   • brand — your brand or close variants. Decide once: is this PMax supposed to capture brand, or is it generic? If generic, every brand query is a negative candidate (see brand search cannibalization).
   • competitor — competitor brand names. Default to negative unless you have an explicit competitor-conquest strategy.
   • irrelevant — informational, job-seeker, support-seeker, adjacent vertical, free-tier seekers. Default to negative.

4. Add the negatives at the right scope. Use a three-layer model:

   • Account-level (cap 1,000) for universal exclusions that apply to every campaign: jobs, careers, salary, free, torrent, tutorial, pdf download. Admin → Account settings → Negative keywords [2].
   • Campaign-level (cap 10,000 per PMax campaign, Q1 2026) for campaign-specific drift — long-tail wastage that only one PMax campaign is exposed to [3]. Campaign → Keywords → Negative keywords.
   • Asset-group level is not available in PMax — only the two scopes above exist.

5. Use phrase match for competitor names and irrelevant patterns. Single-word universal exclusions (free, jobs) as broad. Multi-word patterns (open source alternative, free download) as phrase. Specific brand misspellings as exact. Avoid exact-only negatives for competitor names — variants slip through.

6. Log each weekly addition in a changelog. Date, campaign, query bucket, match type, negative added. The changelog serves two purposes: it proves the cadence is happening (the audit signal this rule looks for), and it lets you reverse a negative that turns out to block converting traffic.

7. Reinforce the signal-not-targeting model with the team. When seeding negatives, name the theme that almost certainly invited the off-intent query. Example: a search theme of "crm software" invited "free crm tools" — the negative is free, and the team learns that themes widen the auction surface rather than narrowing it.

8. Quarterly: review and prune the account-level list. Negatives that block converting queries are a hidden tax. Cross-check the account-level list against the last 90 days of search terms across all campaigns and remove any negatives that now match a known-good query.

How to confirm the fix worked

Diagnostic checklist — re-measure 14 days after deploying the cadence

• Weekly review documented in calendar with named owner.
• Negative-keyword additions ≥ 4 in the trailing 30 days per active PMax campaign with > $1k/month Search-slice spend.
• Account-level negatives populated with ≥ 20 universal exclusions, not at the 1,000 cap.
• Campaign-level negatives present on every PMax campaign live > 30 days that carries > $5k/month (see PMax campaign-level negative keywords for what to seed first).
• Changelog of weekly additions exists — written record, not memory.
• Search-slice wastage share trending down week-over-week in the trailing 7-day report.
• Team can articulate that search themes are signals, not targeting, when asked.

Re-run the audit — pmax_search_terms_report_hygiene moves from failed → passed once the trailing-30-day negative-keyword-addition count crosses the threshold AND a changelog or equivalent process artifact exists. Both conditions must hold.

Edge cases — when this finding does not apply

• PMax campaigns with < $500/month Search-slice spend can move to a biweekly or monthly cadence. The fixed time cost of the review exceeds the leak it would prevent at low spend levels.
• Brand-only PMax campaigns (intentionally constrained to brand queries via brand-only asset groups + brand search themes) need a different workflow — the off-intent question is competitor names and brand misspellings, not generic intent buckets. See performance-max brand cannibalization.
• PMax campaigns < 14 days old are still in the learning phase and the search terms report will be sparse and unrepresentative. Begin the weekly cadence at day 30, not day 7.
• Accounts with single-PMax-campaign setups can run the entire workflow at the account-level negative list and skip the campaign-level layer, since there is only one PMax campaign for the cap to constrain.

Why this rule is about cadence, not cap mechanics. The whitead audit splits PMax negative-keyword hygiene into two rules deliberately. The cap-and-scope mechanics — what is the account-level cap, what is the campaign-level cap, when did campaign-level negatives roll out — live in the glossary article [[pmax-negative-keywords-campaign-level]]. This rule asks the harder operational question: is the operator actually pulling the search terms report on a recurring basis and acting on what they see? The audit signal is two-part: a trailing-30-day count of negative additions per active PMax campaign, plus a written changelog or calendar artifact that proves the cadence is more than a one-time burst. Either signal alone is insufficient. A team that added 50 negatives in a single Monday in week 1 and zero in weeks 2-4 fails the same way a team that adds 1-2 per week with no documentation fails — the failure mode is a cadence that does not survive operator turnover, vacation, or quarterly priorities. The fix is process, not configuration. Sequence this rule before any broader PMax restructure (over-segmentation, asset-group consolidation) — restructuring a PMax campaign that is leaking 15% to off-intent queries means restructuring around a polluted Smart Bidding training signal, and the new structure inherits the same leak under a different campaign name.

Related rules + concepts

• PMax campaign-level negative keywords — companion glossary covering the 10,000-per-campaign cap mechanics, Search/Shopping inventory scope, and rollout timeline. This rule references that one for cap detail rather than re-explaining it.
• Negative keywords — base concept: match-type semantics, scope hierarchy across account/campaign/ad-group.
• Fix: negative keywords missing — existence tripwire for accounts with zero Search-campaign negatives; this rule is the PMax-specific cadence layer.
• Fix: PMax search themes missing — companion rule on the signal side. Search themes guide PMax to the right auctions; negatives keep it out of the wrong ones. Both are needed.
• Brand search cannibalization — when PMax search terms reveal that the campaign is poaching brand queries, this is the playbook.
• Performance Max — base concept page on the campaign type and its 2025-26 transparency surface.
• Search query mining — the same workflow applied to traditional Search campaigns; the analytical pattern transfers directly to PMax.

Sources

1. Google Ads Help — Negative keywords in Performance Max campaigns. https://support.google.com/google-ads/answer/15726455 (accessed 2026-05-27)
2. Google Ads Help — About account-level negative keywords. https://support.google.com/google-ads/answer/11396330 (accessed 2026-05-27)
3. PPC Land — Google raises negative keywords limit to 10,000 for Performance Max campaigns. https://ppc.land/google-raises-negative-keywords-limit-to-10-000-for-performance-max-campaigns/ (2025-03-11, accessed 2026-05-27)
4. Optmyzr — Search Terms Available Again in Performance Max: What This Means & How Can You Benefit. https://www.optmyzr.com/blog/performance-max-search-terms/ (accessed 2026-05-27)
5. Practical Ecommerce — 3 Performance Max Updates for 2026. https://www.practicalecommerce.com/3-performance-max-updates-for-2026 (2026-02-06, accessed 2026-05-27)