Performance Max brand exclusions, negatives & search term control — stop brand cannibalization

Direct answer — how do you control brand and junk queries on Performance Max?

Fund branded Search, apply brand exclusions on PMax so it cannot rent your brand and call it incremental, build campaign-level and shared negative keyword lists for DIY/jobs/wrong-geo junk, and review available search term / category insights on a cadence — accepting that PMax will never be as transparent as a Search campaign. Without exclusions, PMax “ROAS” is often brand cannibalization plus Soft Primary theatre (attribution inflation).

This insight owns Performance Max brand exclusions, negative keywords, and search term control. Full setup OS: PMax SMB playbook. Search-campaign hygiene twin: search terms & negatives. Sequencing: PMax vs Search. Asset groups and search themes steer traffic; they do not replace exclusions — full OS: PMax asset groups & search themes. Lead quality: UK / US.

Working rule: Brand Search owns brand. Negatives own junk. PMax owns incremental reach — or it owns your dashboard lies.

What “good” looks like after 2–4 weeks: brand exclusions live; branded Search CPA still healthy and funded; core junk families negated; monthly shared-list review on the calendar; PMax efficiency judged after brand is protected — not before.

If you only remember three operator moves: (1) attach brand exclusions before you celebrate week-one ROAS, (2) seed a shared junk list the same day the campaign goes live, (3) never starve branded Search to “make room” for PMax. Those three prevent most cannibalization theatre this URL exists to stop.

What this URL owns vs siblings

Own here: brand exclusion strategy, PMax negative keyword systems, shared lists vs campaign lists, how to use limited search term visibility, anti-cannibalization tests.

Delegate: full PMax launch prerequisites → playbook + checklist; Search-only negatives OS → search terms insight; ROAS lies / view-through → attribution; Soft Primary quality → UK / US lead quality; search themes / asset groups → asset groups & search themes (steering, not blocking).

Out of scope: inventing channel budget sliders that do not exist; legal trademark advice for competitor bidding; claiming Google will someday give Search-level query dumps on every PMax impression path.

Entities: brand exclusions, brand lists, campaign-level negatives, account-level negatives (where available), shared negative lists, search themes, search term insights, category insights, branded Search, cannibalization, Diagnostics.

Why brand exclusions are mandatory for most SMBs

Performance Max can show on Search inventory. If your brand queries convert cheaply in a branded Search campaign, PMax will happily spend there too — then report elite ROAS on demand you already owned.

Default for disciplined SMBs: brand exclusions ON; branded Search funded and not budget-capped into irrelevance.

Rare exception: branded Search is severely limited by budget and you intentionally want PMax to help capture brand overflow. Document that choice. Do not stumble into it.

How to set them (operator view)

1. Build or select a brand list with your brand, common misspellings, and close variants you care about.
2. Apply brand exclusions on the PMax campaign(s).
3. Keep exact/phrase brand keywords (or strong brand themes) in Search with adequate budget.
4. Re-measure PMax CPA/ROAS and branded Search volume for 1–2 weeks.

If PMax “efficiency” collapses after exclusions while MER and CRM stay flat or improve, the previous ROAS was partly rented brand credit (attribution).

UI labels evolve — treat “brand exclusions / brand lists” as the control job, not a screenshot religion.

What belongs on the brand list

Include:

• Legal brand / trading name and common short forms customers actually type
• Frequent misspellings and spacing variants from Search terms on branded campaigns
• Product-line names that behave like brand (only when you already capture them cheaply in Search)

Usually leave off:

• Generic category words (“plumber,” “CRM”) — those are commercial demand, not brand
• Competitor names you are intentionally conquering in Search (handle with Search negatives and competitor structure, not by hiding them only inside a PMax brand list)
• Every trademark lawyer’s wishlist token that never appears in query data

If the brand list is empty because “we’ll add it later,” PMax will teach you what later costs.

After rebuilds and duplicates

Campaign copies, drafts → published, and “temporary” test PMaxes are where exclusions quietly die. After any rebuild:

1. Confirm brand exclusions are attached to the live campaign ID.
2. Confirm shared negative lists are still linked.
3. Confirm branded Search budget was not cut to fund the rebuild.

Diagnostics that look “fine” while exclusions are missing are not fine — they are unmeasured brand rent.

Negative keywords on PMax — what they are for

Campaign-level (and account-level, where your account supports them) negative keywords block irrelevant Search/Shopping-style queries from matching into PMax. Capacity is far higher than early PMax years — enough for real hygiene, not a 10-token toy list.

They are not:

• A full replacement for Search structure
• A way to force exact-match themes like a classic Search campaign
• A fix for Soft Primary lead quality (wrong event still finds junk on allowed queries)

They are:

• DIY / how-to / jobs / salary / training blockers for service businesses
• Wrong-city and wholesale blockers
• Competitor freebie / “near me” spam families where CRM proves waste
• Ecommerce blockers for returns, free shipping scavengers, wrong product language

Search campaigns still need their own hygiene OS (search terms). Do not expect PMax negatives alone to replace weekly Search terms work.

Seed list families (copy, then localize)

Use families, not one-off panic adds:

| Family | Typical tokens (adapt) | Why | |--------|------------------------|-----| | Career | jobs, salary, hiring, CV, resume | Hire-intent, not buyer-intent | | Education | course, training, certification, how to, DIY, wiki | Research / self-serve, not booked work | | Freebie | free, PDF, template, sample download | Lead magnets that never buy | | Wrong geo | cities/states/countries you do not serve | Budget bleed outside service area | | Wrong offer | wholesale, bulk, OEM, franchise (if you are retail/local) | Intent mismatch | | Spam / adult / illegal | obvious junk from Search twin | Protect brand + Quality |

Localize hard. A UK trades account needs different geo and slang tokens than a US home-services metro. A B2B SaaS PMax needs “open source alternative” and “pricing PDF” patterns more than “DIY.”

Shared negative lists vs campaign negatives

| Layer | Job | |-------|-----| | Shared list | Universals across PMax + Search: DIY, jobs, salary, free PDF, wrong country tokens | | Campaign negatives | Theme economics: this PMax sells installs, not training; this one is Phoenix-only geo language | | Account-level (if available) | Account-wide junk floor — still review so you do not block a new service line forever |

Using only shared lists either under-protects themes or overblocks everything the same way. Using only campaign lists duplicates work and drifts.

Review shared lists monthly. Last quarter’s “training” negative can block a profitable certification upsell you just launched.

How many negatives do you need?

Enough to cover junk families, not every zero-conversion row.

• Start with a seeded list (DIY, jobs, salary, how to, course, free, wiki, PDF, wrong geos).
• Add from search term insights and CRM notes weekly for the first 30–60 days.
• Prefer phrase negatives for families; exact when one token is toxic but cousins are commercial.
• Do not negate every thin zero-conversion term — creates bloat and accidental blocks (Search twin).

Exact UI caps change; build process for “capacity exists — use it for families,” not “I hit a folklore limit so I gave up.”

Match types on PMax negatives (practical)

• Phrase for most families ("how to", "salary", city names you never serve).
• Exact when a single token is poison but adjacent commercial language must stay live.
• Avoid dumping thousands of exact zeros from a one-week export — that is Search bloat culture transplanted onto a thinner report.

If a converting query is also a bad CRM lead, that is usually a Primary conversion problem, not a missing negative (UK / US).

Search term control — the tinted window

You will not get Search-campaign-level query transparency on PMax. Expect:

• Search term insights / reports where Google surfaces them
• Category / search category insights
• CRM and call notes (“they asked for jobs / DIY / another city”)
• Asset and landing feedback

Close the loop:

1. Export or review available search terms on a weekly cadence early, then biweekly/monthly when stable.
2. Map junk to negative families.
3. Map money language to Search themes / Search campaigns (steering and capture) — not only to negatives.
4. If converting terms are still bad leads, fix conversion definitions first (UK / US quality) — negating a Soft Primary “converter” just moves spam.

What “tinted” means in practice

You may see some Search terms, some categories, and long stretches of “unknown / other.” That is not a license to skip hygiene. It is a reason to:

• Keep a funded Search foundation so query diagnostics still exist somewhere in the account
• Treat CRM and call transcripts as first-class search-term substitutes
• Prefer family negatives over waiting for a perfect export

Teams that wait for “full PMax query transparency like 2018 Search” never ship exclusions. Ship the controls you have.

Anti-cannibalization test (run this once)

Week A (baseline): PMax live, brand exclusions OFF or incomplete, branded Search funded. Note PMax ROAS, branded Search clicks/CPA, MER, CRM quality.

Week B: Brand exclusions ON; branded Search budget held constant. Re-measure.

Interpret:

• PMax ROAS drops hard, MER and booked jobs stable → previous PMax ROAS was brand theatre. Keep exclusions.
• Branded Search volume collapses and MER falls → brand Search was underfunded and PMax was covering brand; fund brand Search properly, then re-test exclusions.
• Nothing changes → brand was not the issue; look at Soft Primary and geo junk next.

Do not change tracking, landers, and exclusions in the same week if you want a clean read.

Logging the test so finance believes you

Write four numbers into the account notes or a shared sheet before Week B:

1. PMax ROAS (or CPA) with the same conversion columns you always use
2. Branded Search clicks + CPA
3. MER (or total ad spend ÷ revenue / booked revenue proxy)
4. CRM qualified rate or booked jobs

When Week B ends, paste the same four. Finance arguments die when the sheet shows MER flat while platform ROAS “crashed” — that crash was honesty (attribution).

Ecommerce vs lead-gen nuances

Lead gen / services: Brand exclusions almost always on. Negatives heavy on DIY/jobs. Pair with call floors and OCI (home services).

Product / Shopping-influenced PMax: Negatives for returns, wrong SKU language, freebie hunters; brand exclusions still usually on if brand Search exists. Feed quality still dominates (Shopping vs PMax).

AI Max on Search is a different lever — do not confuse Search AI expansion with PMax brand lists (AI Max).

Multi-market and franchise notes

If you run market-specific PMax campaigns (UK vs US, or franchise territories):

• Brand lists should still include the master brand; add local trading names per market
• Shared lists can be global for DIY/jobs; geo junk belongs on campaign or market-shared lists
• Do not one-list-to-rule-them-all across countries with different service footprints

A negative that correctly blocks “Sydney” for a Melbourne-only offer will incorrectly starve a national brand later if someone clones the list blindly.

Weekly / monthly ops checklist

Weekly (first 60 days):

1. Brand exclusions still attached after any campaign rebuild.
2. Review available search term insights; add junk families.
3. Spot-check branded Search is not limited into invisibility.
4. CRM notes for “wrong intent” → negatives.

Monthly:

1. Shared list audit — remove blockers that now match products you sell.
2. Cannibalization glance: branded Search CPA vs PMax; MER.
3. Competitor / geo negative refresh.

Name one owner. “The team watches PMax search terms” means nobody does.

30-minute monthly shared-list ritual

1. Export or open the shared list; skim for tokens matching products you now sell.
2. Pull last month’s CRM “wrong intent” tags; add missing families.
3. Check one Search campaign and one PMax still link the list.
4. Note any exact negatives that should become phrase (or vice versa).

Thirty minutes beats a quarterly “we should clean negatives” that never happens.

Failure modes

• No brand exclusions + celebrating PMax ROAS.
• Negatives never reviewed after launch week.
• Shared list blocks a new profitable service.
• Expecting Search-level query control from PMax.
• Negating Soft Primary converters instead of fixing the conversion.
• Rebuilding the PMax campaign and forgetting to re-apply brand lists.
• Starving branded Search so PMax “has to” buy brand.
• Treating search themes as if they were negatives (themes invite; negatives refuse).
• Cloning another industry’s negative pack without localization.
• Celebrating Shopping-influenced PMax ROAS while brand Shopping and brand Search still overlap unchecked.

When to escalate beyond self-serve hygiene

Escalate to a structured Google Ads audit when:

• Brand exclusions are on, Soft Primary is fixed, and MER still collapses when PMax scales
• Search term / category insights show commercial language you cannot capture in Search because structure is broken (intent structure)
• Multiple PMax campaigns fight each other without clear brand / geo / offer fences
• Leadership insists on PMax-only and refuses branded Search budget — that is a strategy conflict, not a negative-list gap

Query control cannot fix a conversion taxonomy that rewards spam or an account with no high-intent Search foundation (playbook, PMax Readiness).

FAQ

How do Performance Max brand exclusions work?

They prevent PMax from targeting your listed brand terms so branded Search can own that demand. Use when brand Search already works; rare to leave off intentionally.

How many negative keywords can PMax have?

Far more than early PMax eras — enough for real family hygiene. Exact caps change in the UI; design for living lists, not folklore limits.

Can I use shared negative lists with PMax?

Yes — and you should for universals. Still keep campaign-specific negatives for theme economics.

How do I stop Performance Max brand cannibalization?

Brand exclusions + funded branded Search + the before/after test above. Then judge MER, not PMax ROAS alone (attribution).

Do negatives replace search themes?

No. Themes steer; negatives block. Full steering OS: PMax asset groups & search themes. Both matter — one invites commercial language, the other refuses junk and protected brand.

Should Search and PMax share the same negatives?

Share universals. Keep campaign-specific lists for theme and geo economics. Blind 1:1 cloning overblocks or underprotects.

What if search terms are barely visible?

Use category insights, CRM, and call notes. Strengthen Search foundations for query diagnostics. Transparency limits are a reason for Search + exclusions — not an excuse for zero hygiene.

Will negatives hurt Smart Bidding learning?

Huge chaotic list dumps plus bid changes the same day can. Steady family hygiene helps learning by removing junk. Avoid weekly “reset everything.”

Related frameworks

• Performance Max SMB playbook · PMax launch checklist · PMax Readiness
• PMax attribution inflation · PMax vs Search
• Search terms & negatives (Search campaigns)
• PMax lead quality UK · US
• Shopping vs PMax · AI Max
• Free Google Ads audit

Next steps

1. Confirm brand exclusions on every live PMax campaign; fund branded Search.
2. Seed shared + campaign negatives (DIY, jobs, salary, geo junk).
3. Run the anti-cannibalization test for 1–2 weeks.
4. Put search term / CRM junk review on the weekly calendar.
5. Score PMax Readiness; request a Google Ads audit if ROAS looks elite and MER does not.

Performance Max query control is not “set keywords.” It is brand exclusions + living negatives + Search foundations + honesty about the tinted window. When brand is protected and junk is blocked, PMax can chase incremental demand. When it is not, every budget step buys prettier cannibalization.