Skip to content

Fix: Enable PMax brand exclusions

finding google ads updated 2026.05.28 5 min read

Performance Max launched account-level brand exclusion lists in May 2024 and expanded their reach to PMax plus Search inventory in November 2024, yet most advertisers still leave them off — meaning a non-brand PMax campaign quietly siphons branded clicks away from your Brand Search campaign at a cheaper effective CPC and inflates its own ROAS with conversions it did not actually generate.

Why this matters

PMax bidding does not distinguish "easy" branded conversions from net-new prospects when it allocates spend. If your brand is searchable, PMax will route a portion of impressions to those auctions because the predicted conversion rate is extremely high. The result is a three-way leak:

  1. Cannibalization of Brand Search. Your dedicated Brand Search campaign — which typically runs at a CPC 3–10x cheaper than non-brand — loses impression share to PMax. Net cost per branded conversion rises across the account.
  2. Inflated PMax ROAS. Branded conversions get attributed to PMax, masking the true performance of cold-traffic asset groups. Decision-making on budget allocation becomes unreliable.
  3. Cold-funnel starvation. Because brand clicks convert so well, PMax's bidder keeps doubling down on them and gives less budget to prospecting placements. The campaign stops doing the job you actually hired it for.

Search Engine Land's framework piece on PMax brand exclusions makes the same point from a different angle: without separation, you "deny [PMax] data on an essential layer of the lower funnel" because branded traffic crowds out the prospecting signal the algorithm is meant to learn from.[3]

How to verify

In Google Ads UI:

  1. ToolsShared libraryBrand lists. Confirm at least one brand list exists containing your brand name + common variants (typos, abbreviations, sub-brand names).
  2. For each PMax campaign that is NOT a brand campaign: Campaigns → select campaign → Settings → expand Brand exclusions. Verify a brand list is attached.
  3. Cross-check the Search Terms Report inside the campaign for branded queries that should not be there. If you see your brand name appearing as a top converting term and no brand list is attached, the leak is confirmed.

You can also infer the problem from a Brand Search campaign losing impression share to "ad rank" or "budget" without a budget cause — PMax is the silent competitor on the other side of that auction.

How to fix

  1. Build the brand list. ToolsShared libraryBrand lists+ New brand list. Add your primary brand name, all spelling variants, common typos, and any sub-brand or product-line names that should be treated as branded queries. If your brand is not in Google's verified brand database, request verification first — unverified brands cannot be excluded.[1]
  2. Apply to every non-brand PMax campaign. Open each PMax campaign → SettingsBrand exclusions → attach the brand list. The exclusion applies to both Search and Shopping inventory inside that PMax campaign.[2]
  3. Leave brand exclusions OFF for explicitly branded PMax campaigns (e.g., a "Brand Defense PMax" if you run one).
  4. Re-baseline. Expect Brand Search impression share to recover within 7–14 days and PMax CPA / ROAS to normalize to a less flattering but more honest number. Do not panic-edit bid targets during this rebaseline window.

Edge cases

  • Branded searches are more than ~50% of all traffic. Search Engine Land's framework argues that for brand-dominant accounts, blanket brand exclusion strips PMax of the lower-funnel learning signal it needs. In that case, run a separate Brand Search campaign on Target Impression Share AND keep brand exclusions OFF on PMax — accepting some attribution overlap as the cost of preserving signal.[3]
  • Small ecommerce or lead-gen accounts with limited conversion volume. Same principle: if removing brand traffic drops PMax below its learning-phase conversion floor (~30 conv / 30d), keep brand exclusions OFF and accept the cannibalization.[3]
  • You don't run a Brand Search campaign at all. Brand exclusions still help on PMax because they prevent ROAS inflation, but the cannibalization argument doesn't apply — prioritize standing up a dedicated Brand Search campaign first.
  • Multi-brand portfolio under one account. Build one brand list per brand and apply selectively. Do not lump them into one mega-list — you'll over-exclude across campaigns that should be allowed to convert on sister-brand queries.

Industry benchmarks

  • Optmyzr's 2024 study of 24,702 PMax campaigns found that 97% of advertisers do not use brand exclusion lists, despite the feature being live since May 2024.[5]
  • The same study notes that overly restrictive PMax filters can stifle automation, so the recommended practice is "precise, purposeful exclusions" — brand exclusions are the canonical example of a precise, high-ROI exclusion.[5]
  • Tinuiti's PMax best-practices guidance frames the same risk in operational terms: "Don't turn off your Search campaigns. Performance Max is designed to complement your existing Search campaigns, not replace them" — which is only true if you actively prevent PMax from absorbing branded traffic.[4]
  • [[fix-pmax-search-terms-report-hygiene]]
  • [[fix-pmax-campaign-level-negatives-seeded]]
  • [[pmax-negative-keywords-campaign-level]]
  • [[fix-pmax-over-segmentation]]

Sources

[1] Google Ads Help. "About brand exclusions in Performance Max campaigns." https://support.google.com/google-ads/answer/16669487

[2] Google Ads Help. "Apply brand exclusions to your Performance Max campaign." https://support.google.com/google-ads/answer/14505308

[3] Ani, Menachem. "How to think about brand exclusions for Performance Max." Search Engine Land. https://searchengineland.com/performance-max-brand-exclusions-395439

[4] Tinuiti. "Google Performance Max: Ad Examples & Tips." https://tinuiti.com/blog/search/google-performance-max/

[5] Optmyzr. "Performance Max Study: Data & Tactics." https://www.optmyzr.com/blog/performance-max-study/

// was this useful?
// anonymous · no personal data stored