How to fix: Brand and non-brand keywords mixed in one campaign
TL;DR
When brand keywords (your company name, product names, trademarks) sit in the same Search campaign as generic acquisition keywords, three things break at once: branded conversions are 3-10x cheaper and steal a disproportionate share of budget, blended CPA and ROAS become meaningless, and Smart Bidding trains on a corrupted signal that under-bids real acquisition auctions while over-paying on traffic you would have won anyway. The fix is structural — isolate brand into a dedicated campaign with its own budget, bid strategy, and copy; add brand tokens as negatives on every non-brand Search campaign to stop cross-pollination; pick bid strategies per intent (Target Impression Share for brand defence, Target CPA or Target ROAS for acquisition) [1].
Why it matters
Branded queries and non-brand queries are different products served through the same channel. A user searching your brand name has already chosen you — the click costs $0.10-$1.00 with conversion rates 5-10x higher than generic keywords, because the user is bottom-funnel rather than discovering you for the first time [2]. A user searching "best CRM software" is top-of-funnel and that click costs whatever the contested auction demands — often $5-$50 in B2B SaaS.
Mixing them breaks four mechanisms simultaneously. First, budget cannibalisation: branded queries are cheap and convert well, so Smart Bidding routes more spend through them, leaving non-brand keywords throttled before they accumulate the conversions they need to learn. Second, reporting distortion: blended CPA looks great because branded conversions average down the cost, but real acquisition CPA is typically 30-60% higher than reported. Third, Smart Bidding signal poisoning: Target CPA and Target ROAS calibrate against an unrealistically low CPA expectation when brand dominates the training set, then under-bid in non-brand auctions where the economics are different [3]. Fourth, lost defensive lever: a brand campaign needs 90%+ Search Impression Share to stop competitors intercepting high-intent branded traffic, but if brand and non-brand share a budget, brand impression share collapses whenever acquisition pulls spend.
Per Google auction mechanics, two ad groups from the same advertiser cannot both serve on one query — only the higher Ad Rank wins. The generic campaign's bid is usually higher (Smart Bidding treats branded converters as premium), so generic wins the auction and burns commercial-tier budget on a query that should have cost a fraction of that [1]. This is a campaign-structure problem, not a bid-modifier problem — the only fix is structural separation.
How to fix
Tag search terms. Pull the last 90 days search-terms report (Insights and reports → Search terms → Download CSV) and tag every term as brand or non-brand using your company name, product names, trademark tokens, common misspellings, and brand+modifier patterns ("brand reviews", "brand pricing", "brand vs competitor").
Identify mixed campaigns. Open each Search campaign (Campaigns → select campaign → Audiences, keywords, and content → Search keywords) and flag any campaign that has both branded keywords (or broad/phrase keywords pulling branded queries) and non-brand keywords.
Create a dedicated Brand Search campaign. New campaign → Search. Brand keywords in phrase and exact match only — never broad, which pulls non-brand variants and reintroduces the mix. Ad copy doubles down on brand promise (founders, awards, trust signals) — branded searchers are mid-funnel, the ad job is reassurance not discovery.
Add brand tokens as negatives on every non-brand campaign. Campaign → Negative keywords → add each brand token as a phrase or broad negative. Without this step the structural fix does not stick. For accounts with many non-brand campaigns, build a shared negative list (Tools → Shared library → Negative keyword lists) and attach it to every non-brand Search campaign.
Pick bid strategies per intent. Brand → Target Impression Share at 90% absolute-top, with a max CPC ceiling so the strategy cannot run away on contested auctions. Non-brand → Target CPA or Target ROAS, anchored to your true acquisition CPA — gate the strategy on 30 conversions / 30 days before switching from Maximize Conversions.
Set budgets independently. Brand needs enough budget to never lose impression share to cap (monthly branded query volume × expected CPC × 1.3 buffer). Non-brand campaigns pace separately against acquisition CPA targets. Shared budgets recreate the cannibalisation problem.
Verify after 14 days. Re-run the search-terms report on every non-brand campaign — branded query share should be 0% (or under 1% noise floor). Brand campaign Search Impression Share should be 90%+ at absolute top. Reported non-brand CPA will be higher than before — that is the correct number.
Common mistakes
- Broad match on brand keywords. Broad on "Acme" pulls "Acme alternatives", "Acme vs competitor", "Acme jobs" — the queries non-brand campaigns and shared negative lists should handle. Brand campaign uses phrase and exact only.
- Skipping negatives on non-brand campaigns. Both campaigns compete for the same queries, the higher-bid one wins, and the structural fix achieves nothing.
- Sharing a budget between brand and non-brand. Recreates cannibalisation at the budget layer — when acquisition is hungry, brand impression share collapses; when throttled, brand burns its share of the pool on traffic it would have won at 10% of the cost.
- Picking Target CPA on the Brand campaign. Will under-bid against contested branded auctions and lose impression share to competitors squatting on your terms. Use Target Impression Share for brand defence.
- Forgetting brand+modifier queries. "Acme pricing", "Acme reviews", "Acme login" are all branded — your brand token list must cover patterns, not just the bare name.
FAQ
Q: Do I need a separate Brand campaign if I am only running Performance Max?
A: Yes — and the PMax-specific version of this problem is documented in Performance Max brand cannibalization. For PMax, the fix is brand exclusions at the account level plus a parallel Search Brand campaign to actually win branded queries with control over copy and bid strategy.
Q: What if my brand has very low search volume — does the separate campaign rule still apply?
A: Below ~10 branded conversions per month, separation still matters for reporting accuracy but the bid-strategy choice changes — use Manual CPC or Maximize Clicks on the Brand campaign until volume justifies Target Impression Share. The negative-keyword block on non-brand campaigns matters regardless of volume.
Q: Should I bid on competitor brand names in the same Brand campaign?
A: No — competitor conquesting is a third intent bucket with different ad-copy rules (trademark restrictions, copy that compares without naming, often higher CPCs because the competitor outbids on their own terms). Run a separate Competitor campaign with its own budget and ad strategy if conquesting is part of your plan.
Q: My agency says brand and non-brand can stay together if Smart Bidding handles it. Are they right?
A: No. Smart Bidding cannot solve a structural problem — it bids per auction but cannot separate the reporting, budget pacing, or strategic-intent layers. Even with Target ROAS, blended reporting will hide real acquisition CPA, and you lose the budget lever for brand defence. The "Smart Bidding handles it" argument typically comes from agencies that do not want to do the campaign-restructuring work.
Q: How often should I review the brand-token list and negatives?
A: Once a quarter at minimum, plus immediately after any product launch, rebrand, or trademark filing. Search terms reports drift as Google expands match-type interpretation; quarterly audits catch new branded patterns before they leak budget.