Brands that pit performance max vs search campaigns against each other as a budget competition almost always end up with an account that underperforms on both.
The question is not which to run. It is how to run them together without letting one cannibalise the other.
Without This System, You Are Paying Google Twice for the Same Buyer
Search campaigns capture people who typed what you sell into Google. Performance Max captures that intent plus everyone across Shopping, Display, YouTube, Discover, and Gmail the algorithm predicts will convert. These are not interchangeable demand pools. Pulling budget from one to fund the other forfeits demand the other channel was built to find.
The problem most accounts hit: PMax launched without guardrails pulls budget toward exact queries your Search campaigns already own at lower CPA - and toward branded terms your existing customers search - inflating conversion counts while the actual new-buyer rate stays flat.
This is the Search-PMax Stack: a setup where performance max vs search campaigns run simultaneously, each doing the job it was built for, with clear rules that keep the two from bidding against each other.
Step 1: Agree on One Conversion Goal Before Either Campaign Launches
The most common structural error in Google Ads is having Search campaigns optimize for one goal and PMax optimize for another - or both optimizing toward micro-conversions (add-to-carts, page views) that do not map to the metric the business actually tracks.
Both campaigns must point at the same conversion action. For ecommerce, that is a completed purchase tracked server-side. For lead generation, a qualified form submission. For SaaS, a trial sign-up or demo booked.
When the goal differs between campaign types, the algorithm's decisions diverge. One system bids to drive checkouts; the other bids to drive add-to-carts because add-to-carts were cheaper and the volume metric looked stronger. Two campaigns, two definitions of success, one confusing blended report that obscures both.
One goal. Both campaigns. Confirmed before either goes live.
Step 2: Launch Exact-Match Search First
The single most important google search ads best practices principle for a mixed Google account is this: Search should accumulate conversion history before PMax is introduced.
Run exact-match and phrase-match Search campaigns targeting your highest-intent query clusters for four to six weeks. The purpose is not primarily volume - it is building a conversion signal Google's systems can learn from. PMax at launch needs data to orient its bid decisions. An account with 50 or more relevant conversions in the past 30 days gives PMax a meaningful starting point. An account with eight conversions over two months gives it noise.
During this Search-only phase, review the search terms report weekly. The queries triggering your campaigns and producing conversions become the structural input for Step 3.
Step 3: Build PMax Asset Groups Around Intent, Not Product Catalogues
With pmax asset groups explained correctly, each asset group is a separate creative and audience signal container - not a product category folder. The mistake most accounts make is building one asset group per SKU or product line. The better structure is one group per buyer intent.
A DTC/Apparel brand might run three asset groups: one for gift-occasion searches, one for seasonal-need searches, and one for general brand discovery. Each gets headlines, descriptions, images, and video that speak to that specific intent. Each gets audience signals seeded with people who have already shown that behavior.
Asset groups structured around intent give the algorithm genuinely different creative to test across genuinely different demand signals. Asset groups built around product categories give it different product images pointed at the same audience pool.
For Shopping accounts, a clean and fully populated product feed is the prerequisite. PMax draws from the feed for product surfaces, and missing attributes, outdated pricing, or low-quality images constrain what the algorithm can show before a single bid decision is made. The Performance Max setup guide covers the full feed requirements before launch.
Step 4: Protect Branded Queries From PMax Cannibalization
PMax will bid on branded search terms. It will also compete against your own exact-match Search campaigns for the same queries. When both campaigns win the same branded auction, you pay more for a click your Search campaign would have captured at lower cost - and the conversion credits to PMax, making PMax look more efficient than it is.
The fix is specific, and it is not done through standard negative keyword lists:
- Add brand exclusions in PMax campaign settings - this is a separate feature from the negative keyword tool
- Run branded Search as its own campaign, separated from non-brand campaigns, so branded performance is always readable on its own line
- Pull the Insights tab from PMax weekly and check whether branded queries are appearing in the search category breakdown
Brand cannibalization is the issue BAVai surfaces most often in the Google Ads accounts we audit. When PMax launches without brand exclusions, the dashboard shows strong conversion volume while the new-customer acquisition rate stays flat. That is a setup problem, not a PMax problem - and it takes about ten minutes to fix.
Step 5: Hold Spend Steady Through the Learning Phase
The operators who cut PMax budget in week two are the ones who conclude the campaign type does not work. The system was not broken. It was interrupted before it could finish learning.
PMax needs approximately 50 cumulative conversions to stabilize bid decisions. The learning window runs two to six weeks depending on daily spend and conversion volume. During this period, performance variance is expected - costs per conversion will swing, impression mix will shift, and the algorithm will test combinations that look inefficient before narrowing on what converts.
Material changes during learning - budget cuts above 20%, bid strategy shifts, asset swaps - partially reset the clock. Following google search ads best practices, weekly budget changes above 20% in either direction force re-calibration at the new spend level. An account that responds to week-two volatility with cuts and restarts can stay in the learning phase indefinitely.
The ad account health check framework includes a monitoring approach for identifying when learning-phase interference is happening - as opposed to genuine underperformance that warrants a structural change.
What Good Looks Like
A well-structured Search-PMax Stack shows a clear division of labor in the account data. Search campaigns hold high-intent branded and category queries at a predictable CPA. PMax expands reach across Google's full inventory at a higher average CPA, but drives new buyers the Search campaign would not have reached on its own.
The combined number to watch is blended new-customer CPA - not platform ROAS on either campaign independently. BAVai tracks this daily across the Google accounts we manage: when PMax CPA drifts above target, the first checks are whether brand cannibalization has increased and whether pmax asset groups have degraded creative quality. Both are fixable within hours. Neither is visible in the standard dashboard without someone specifically looking for them.
The structural signal that confirms the performance max vs search campaigns setup is working: Search CPA holds or improves after PMax goes live (confirming no cannibalization), and total Google conversion volume increases, indicating PMax is finding demand Search was not reaching.
The Checklist
- One shared conversion action confirmed across both Search and PMax - same goal, same tracking
- Search campaigns launched first, building toward 50 or more conversions before PMax is introduced
- PMax asset groups built around buyer intent themes, not product category folders
- Brand exclusions added to PMax from day one via campaign settings
- Branded Search campaign running separately from non-brand campaigns
- Learning phase protected - no budget changes above 20% in weeks one through four
- Weekly: Insights tab checked for branded query cannibalization, Search terms reviewed, new-customer CPA trend confirmed
Most accounts running both PMax and Search today have either PMax competing against their own Search campaigns for branded queries, or Search running without PMax coverage on the inventory Search cannot reach. Which gap is costing you more new buyers right now?