Skip to content
whitead

Fix: Negative keywords missing or insufficient

finding google ads updated 2026.05.06 5 min read
// contents · 7 sections
  1. Why this matters
  2. How to verify the issue
  3. How to fix it
  4. How to confirm the fix worked
  5. Diagnostic checklist — re-measure 14 days after deploy
  6. Related rules + concepts
  7. Sources

A Search campaign running broad or phrase match without a working negative-keyword layer is paying Google to learn what not to show — and feeding garbage queries straight into Smart Bidding's training set.

Why this matters

HIGH severity. Modern Search relies almost entirely on broad match plus Smart Bidding, and Google has retired close-variant exact match — phrase and exact now match looser intent variants (Google Ads Help, 2024-09). Without a negative layer the campaign serves on questions, jobs, free-tier seekers, competitor research queries, and adjacent verticals that share lexical overlap but zero buying intent.

Two costs compound:

  1. Direct waste. Every irrelevant click is paid spend with near-zero conversion probability.
  2. Training-data poisoning. Smart Bidding learns from the conversions it sees. When 20-40% of impressions go to non-converting queries, the bid model bakes that noise into its baseline and pulls bids down on genuinely commercial intent — the lift you'd expect from broad match never materializes.

Google's own "Use broad match with Smart Bidding" guidance assumes a maintained negative list as a precondition (Google Ads Help, 2025). Skipping that step is the single most common reason broad-match pilots underperform their phrase-match baseline.

How to verify the issue

  1. Open Campaigns → select the Search campaign → Audiences, keywords, and content → Negative keywords. Count active negatives at campaign + ad-group level. Fewer than 10 active negatives on a campaign that has been live >30 days = finding confirmed.
  2. Pull the Insights and reports → Search terms report for the last 90 days. Sort by impressions descending. Read the top 50 — flag every term that is off-intent (informational, navigational to a competitor, job-seeker, free-tier hunter).
  3. Calculate irrelevant query share: irrelevant_impressions / total_impressions. Above 10% = active leak. Above 25% = the bid strategy is being actively miseducated.
  4. Check for a shared negative list at Tools → Shared library → Negative keyword lists. Absent = there is no account-wide brand-safety / off-vertical guardrail.

How to fix it

Total time: 2-4 hours for the initial sweep, then a recurring 30-minute weekly cadence.

  1. Export 90 days of search terms. From Insights and reports → Search terms, download CSV. Filter for impressions ≥ 5 to focus on volume drivers.

  2. Cluster non-relevant queries by intent bucket. Typical clusters:

    - informational ("what is", "how to", "vs", "review", "tutorial")
- job-seeker ("jobs", "careers", "salary", "hiring")
- free-tier ("free", "open source", "cracked", "torrent")
- competitor (their brand names — only if you don't bid on them)
- adjacent vertical (lexically close but wrong audience)

  3. Build the negative list with correct match types. Use phrase match for multi-word patterns, exact for single high-volume terms, broad only for tokens that should never appear (free, jobs).

    # Shared list: "Global off-intent guardrail"
"free trial download"     # phrase
"open source"             # phrase
[free]                    # exact - only when "free" never has commercial intent
jobs                      # broad token
careers                   # broad token

  4. Apply at the right scope. Account-wide guardrails (jobs, careers, free, competitor names) → shared list attached to all Search campaigns. Campaign-specific drift → campaign-level negatives. Ad-group cross-pollination between tightly-themed groups → ad-group negatives.

  5. Cross-link to Search Query Mining for the recurring workflow — this finding is a one-time fix, but the underlying discipline is weekly.

"If you use broad match, regular review of the search terms report and addition of negative keywords is essential to maintain efficient performance."
— Google Ads Help, About broad match (accessed 2026-05-06)

How to confirm the fix worked

Diagnostic checklist — re-measure 14 days after deploy

  • Active negatives count ≥ 50 per Search campaign (or shared list with ≥ 100 entries attached).
  • Irrelevant query share in the 14d post-deploy search terms report < 10%.
  • Shared negative list present and attached to every Search campaign in the account.
  • CTR on remaining queries trending up (irrelevant queries usually had below-average CTR — removing them lifts the campaign average).
  • Conv rate stable or rising; CPA flat or down within 14-21 days as Smart Bidding re-stabilizes.

Re-run the audit — negative_keywords_missing moves from failed → passed once the campaign has any active negatives, but real impact requires the full sweep.

Methodology note. Whitead's negative_keywords_missing rule is an existence check (check_type: existence, data_source: negative_keywords) — it fires when a Search campaign has zero negatives at the campaign or ad-group scope, regardless of shared-list coverage. The rule is intentionally permissive: a single negative is enough to pass, because the rule is a tripwire for accounts that never opened the negatives panel at all. Pair this finding with the Search Query Mining playbook for the deeper "how many negatives, of which type, against which queries" diagnostic — that workflow uses the search terms report rather than the negatives list as its data source, and catches accounts that have a token list (5-10 entries) but no real coverage.

Sources

← Back to Wiki