Google Ads Search Terms Audit Checklist

SEARCH-TERM EVIDENCE REVIEW
Scope: [anonymous offer, Search campaigns and network, dates, timezone, currency, cost definition, snapshot]
Source exports: [campaign total; individual query rows; exclusions and completeness]
Normalization: [minor-unit precision, rounding policy, row grain, duplicate checks]
Relevance rubric: [offer-fit, out-of-scope, ambiguous, owner and version]
Amounts: [T total; V visible rows; R reviewed rows; I reviewed out-of-scope]
Required order: [0 <= I <= R <= V <= T]
Partition: [R-I fit; I out; V-R visible unreviewed; T-V not represented]
Coverage: [V/T; R/T; I/R, null on zero denominator]
Logical bounds: [I/T to (I+T-R)/T; no probability or savings interpretation]
Residual checks: [network/filter/date/snapshot/rounding/disclosure hypotheses]
Action candidates: [opaque row key, reason, evidence, ambiguity; no automatic negatives]
Performance boundary: [conversion definition, maturity, tracking quality; no causal claim]
Decision: [reconcile scope / review visible rows / inspect residual / review classifications]
Owner and review date: [human approver, smallest evidence needed, next check]

Start with the decision, not a negative-keyword list

Use this original review sequence when a Search campaign's cost is larger than the sum of its individual search-term rows, or a query audit reports a suspiciously precise “waste” rate. The audience is a paid-search operator, founder or analyst who can access aggregate campaign costs and review an offer's intended demand. Do not use it to audit mixed-channel account spend without a valid Search cost slice.

The output is an evidence ledger and a shortlist for human review, not permission to block traffic. It cannot reconstruct undisclosed searches or estimate profit. Existing acquisition economics and conversion measurement remain separate.

Use four gates before proposing a change

1. Scope gate: the analyst freezes campaigns, network, dates, timezone, currency, cost definition and export time. The campaign owner confirms that the two reports cover the same activity.
2. Row gate: remove report totals and other-query aggregates from individual rows. Check the actual export grain and overlapping segments. Preserve exclusions rather than silently deleting differences.
3. Classification gate: the offer owner defines fit and out-of-scope intent. Reviewers keep ambiguous rows unreviewed. A second reviewer challenges high-cost or high-impact candidates; no arbitrary percentage threshold is required.
4. Decision gate: reconcile the four-part partition in the calculator, inspect unknown spend and choose one evidence action. Campaign changes need the usual account owner and a separate performance check.

Worked synthetic review

A closed-month Search campaign has 12,000 total cost. Individual term rows contain 8,400; 6,000 has been reviewed and 900 of that fails the stated offer-fit rubric. The partition is 5,100 reviewed fit, 900 reviewed out, 2,400 visible unreviewed and 3,600 not represented in the supplied rows. It reconciles to 12,000.

Visibility is 70% and review coverage 50%. Out-of-scope cost is 15% of reviewed spend but only 7.5% of total spend. Its possible total share is 7.5%–57.5% until the unclassified 6,000 is resolved. This is a logical range, not an estimate of waste. The next step is to review material visible rows and reconcile the residual; it is not to announce a 57.5% saving.

Record a small, accountable next step

If the amounts violate the subset order, reconcile scope before classifying. If visible rows remain unreviewed, start there. If all visible rows are classified, investigate filters, report freshness and available aggregate insights without guessing hidden text. If everything is classified, challenge the rubric and downstream performance separately.

Common failure modes are adding subtotals to detail rows, counting the same segmented cost twice, treating zero conversions as irrelevant intent and projecting a visible-row rate onto a privacy-selected pool. Keep the export and rubric version so a later change in visibility is not mistaken for better traffic quality. The editable worksheet above makes ownership and uncertainty explicit.

Sources and limitations

Checked October 22, 2026: Google Ads search terms documentation explains low-activity query omissions and differences from search terms insights. Publication/update date is not displayed. Microsoft Advertising's query-report request documentation, dated November 13, 2024, describes significant-click selection, reporting timezone and completeness controls. These current references establish reporting constraints, not a new October release or measured keyword demand. This original editorial method is not endorsed by either platform and does not infer hidden query identities, causal waste or guaranteed savings.