Run both, but don’t let them share a brief. Search is where you control intent. Performance Max (PMax) is where you hand Google a goal and a product feed and let it hunt across every surface. Treat PMax as a replacement for Search and you’ll often pay twice for customers who were already typing your name.
That is the whole argument. The rest of this piece is the mechanism, the numbers we can actually source, and a setup you can copy this week.
Key takeaways
- Search wins the overlap. In an Adalysis study published December 2024, Search converted better than PMax 84.18% of the time on overlapping search terms. Adalysis reports 84.62% for conversion value.
- Search only gets priority on an exact match. Per Google Ads Help, Search takes precedence when a query matches an active exact match keyword. Otherwise Ad Rank decides.
- PMax’s brand results are often borrowed. If you haven’t applied brand exclusions, a slice of your PMax return is brand traffic that Search would have won more cheaply.
- PMax earns its place on product-feed objectives and broad discovery, with guardrails and a clean boundary.
- Measure incrementality, not platform ROAS. Platform reports reward whoever claims the conversion, not whoever caused it.
Table of Contents
- What each campaign type does
- Why PMax and Search collide
- What the data says about cannibalisation
- The brand-traffic problem
- When Search should lead
- When PMax earns its budget
- A split that works for most Australian ecommerce accounts
- How to tell whether PMax is adding anything
- Our view on PMax vs Search
- If you want a second pair of eyes
- FAQ
- Sources
What each campaign type does
Search campaigns show your ads to people typing queries that match keywords you chose. You pick the words, write the ads, set negatives and read a full search terms report. The trade is scale: you can only buy the queries you thought of, or that match loosely enough to reach you.
PMax is a goal-based campaign type that accesses all Google Ads inventory from one campaign, including YouTube, Display, Search, Discover, Gmail and Maps. It has no keywords. You supply assets, audience signals and a conversion goal. Google’s system decides where, when and to whom.
Google positions it as a complement to keyword-based Search campaigns, not a substitute. Read that line twice. Most of the confusion in this debate comes from people treating a complement as a competitor.
| Search | Performance Max | |
|---|---|---|
| Targeting | Keywords you choose | Goals, assets, audience signals |
| Inventory | Google Search (plus optional partners) | Search, Shopping, YouTube, Display, Discover, Gmail, Maps |
| Search terms visibility | Full report | Limited, harder to audit |
| Control | High | Low to moderate |
| Best at | Capturing known demand | Extending reach, feed-driven product ads |
| Main risk | Ceiling on volume | Claiming conversions you’d have won anyway |
Why PMax and Search collide
Both campaign types can bid on the same query. So who wins?
Google’s documentation is specific. Search is prioritised over PMax only when the query exactly matches, including spell-corrected variants, an active exact match keyword in a Search campaign. In every other case Google falls back on Ad Rank, a mix of ad relevance, landing page quality and expected return.
Sit with that. A phrase match keyword doesn’t guarantee priority. A broad match keyword doesn’t either. If the algorithm predicts PMax will return more on that auction, PMax can take it. That is not a bug. It is the rule.
The second mechanism is the one people miss. PMax reports a conversion if it served any part of the path. When it serves on a query Search would have won, the sale moves from one bucket to the other and PMax looks like a hero. Nothing new was created. The sale just changed address.
What the data says about cannibalisation
What the data shows on overlap is thin but consistent.
Adalysis looked at search terms that appeared in both campaign types. On those overlapping terms, Search campaigns had higher conversion rates 84.18% of the time, and higher conversion value 84.62% of the time (Adalysis, December 2024). PMax often claimed more impressions on those terms, but they skewed to lower-intent traffic.
Brad Geddes of Adalysis makes a sharper point. The line that PMax finds new audiences is skewed by brand terms. Strip the brand-search influence out, and PMax often isn’t adding net-new demand. It is moving existing demand at a higher cost.
Optmyzr ran its own study across hundreds of accounts. Its headline is that PMax often cannibalises Search, even where exact match keywords exist (Optmyzr, July 2025). The study is worth reading for how it measured the overlap.
Caveats, because you’ll want them. These are vendor studies of accounts using those vendors’ tools, so they skew toward advertisers who audit their accounts. Overlap is not proof that PMax added nothing. Some of those searchers might have clicked a competitor if PMax hadn’t shown. Nobody outside Google can measure that cleanly from search term data alone. Anyone quoting you an exact cannibalisation rate for your account, without a holdout test, made it up.
The brand-traffic problem
Brand search is where the damage concentrates. Someone types your store name. Search would win at a few cents of intent-heavy click. PMax serves instead, books the sale, and your blended PMax ROAS looks magnificent.
The fix is brand exclusions at the campaign level in PMax settings. A few honest notes:
- Apply the brand list. If you have never done it, assume a good slice of your PMax result is brand traffic in a trench coat.
- Check that the list covers misspellings, product names and the sub-brands customers actually type.
- Keep a dedicated brand Search campaign. Local Australian Shopify operators have been writing about this problem for ecommerce founders (Ecommerce Circle, July 2026), and the same advice holds across platforms: brand exclusions are only as good as their configuration.
- Re-check the PMax search terms insights monthly. Lists drift as you launch products.
A warning on the usual shortcut. “Add negative keywords to block PMax” is a half-measure. Search Engine Land’s coverage of Brad Geddes at SMX Next (January 2025) argues for a surgical approach instead: core high-intent terms stay in dedicated Search campaigns for control, PMax handles broad discovery or product-feed goals, and campaign-level negative lists and brand exclusions enforce the line. Blanket negatives can starve PMax of the signals it learns from.
When Search should lead
Pick Search first when any of these is true:
- You sell on known demand. People already search for the product category or your brand. Search captures that at the intent level and lets you write the ad around the exact query.
- Lead quality matters more than lead volume. PMax can fill a CRM with junk. Search Engine Land’s February 2026 piece says that when volume is high and quality is poor, the system is likely optimising for clicks rather than profit, and the fix is negative keyword guardrails plus tighter location and time-of-day limits.
- Your conversion data is thin. PMax needs conversions to learn. A new account with a handful of sales a month gives it little to work with. Search with exact and phrase match gives you control while the data builds.
- You need to read the search terms. If product or messaging decisions depend on what people actually typed, Search shows you. PMax shows you much less.
When PMax earns its budget
PMax is not the villain. It does three jobs well enough to deserve a place.
Product-feed ecommerce. If you have a clean product feed with decent margins data, PMax can place those products across Shopping, YouTube and Display from one campaign. That is hard to rebuild by hand.
Reach beyond your keyword list. Search can’t buy a query you never thought of. PMax can find some of them, and YouTube and Display placements you wouldn’t have bought separately.
Lean teams. One campaign with strong creative is less work than five. If you have a small in-house team, that matters.
The condition for all three: the brand boundary is set, conversion tracking reflects profit rather than just a click or add-to-cart, and someone looks at the account weekly.
A split that works for most Australian ecommerce accounts
There is no universal ratio, and anyone handing you one is guessing. What we can give you is a structure:
- Brand Search campaign. Exact match on your brand, with its own budget. Cheap, high intent, fully visible.
- Non-brand Search on exact and phrase match. Your proven money keywords, where you want control over the message and the bid.
- PMax with brand exclusions on. Fed by your product feed, with asset groups split by product category or margin band rather than one pile of everything.
- Weekly checks. Search terms overlap, brand leakage and the gap between platform-reported and actual revenue.
Why exact match matters here: it is the only match type Google documents as taking priority over PMax. If a keyword is your bread and butter, put it in exact match and let the rule work for you.
One trap on budgets. If you launch PMax and cut Search “to give it room”, you may see total conversions hold flat while cost per acquisition climbs. That is the pattern the overlap data predicts: the same buyers, a different campaign taking credit. Move budget in steps and watch blended revenue, not the PMax column.
How to tell whether PMax is adding anything
Platform ROAS can’t answer this. It counts who claimed the sale, not who caused it. Three checks, in rising order of rigour:
- Brand share check. Pull PMax search term insights. If most converting categories are your name and variants, PMax is mostly harvesting.
- Before and after on blended numbers. Compare total revenue and new customer share over matched periods when you switch PMax on or off. Messy, but directional. Call it directional, because seasonality and promotions pollute it.
- A holdout test. Run a geo or time split where one group sees no PMax. This is the only version that comes near causation. We’ve written about the logic in our incrementality testing guide for Meta, and the same thinking applies to Google.
If you want a wider lens on whether paid campaigns create demand or just collect it, this comparison of SEO and paid ads covers how the two channels interact, though it is written in Portuguese.
Before you change any bids off the back of a report, check which attribution columns you are reading. We cover that in Google Ads attribution: check the columns before you change bids.
Our view on PMax vs Search
We don’t pick sides. We pick boundaries.
Most agencies hand Google Ads to one person, creative to another and landing pages to a third. Each hand-off leaks information. The PMax debate is a symptom: the media buyer sees a PMax ROAS, the creative team never learns which assets drove it, and nobody asks whether the landing page converts the traffic once it arrives.
We run strategy, creative, paid media and conversion optimisation as one team, which changes how we handle this choice. Search tells us the language buyers use. That language goes into PMax assets and ad creative. The PMax asset reports tell us which angles work. Conversion work on the landing page lifts both campaign types at once, because the click is the same cost whichever campaign bought it.
Our rules of thumb, for ecommerce brands:
- Judge on revenue and new customers, not on the in-platform ROAS column alone.
- Never launch PMax without brand exclusions and a separate brand Search campaign.
- Put proven keywords in exact match so Google’s own priority rule protects them.
- Fix the landing page before you raise the budget. A bad page taxes every dollar equally.
The best split depends on your margins, your feed quality and how much of your demand is already branded. If someone gives you a ratio without asking those three things, they sold you a template.
— Harry
If you want a second pair of eyes
If your PMax and Search campaigns are overlapping and you can’t tell who is claiming what, our Google Ads team audits the account structure, sets the brand boundary and ties results back to revenue. We also work across conversion optimisation and creative, so the fix doesn’t stop at the ad. You can see the full range on the Thrive homepage. Happy to walk through your search terms report if it would help.
FAQ
Is PMax better than Search?
Neither wins everywhere. Search gives you keyword control and full search term data, and it won the overlapping-term comparison in Adalysis’s December 2024 study 84.18% of the time on conversion rate. PMax gives you reach across Google inventory and works well with a product feed. Most accounts do best with both, separated cleanly.
Does PMax cannibalise Search campaigns?
It can. Google says Search has priority only on an exact match to an active exact match keyword, and otherwise Ad Rank decides. Brand terms are the usual leak. Apply brand exclusions in PMax and keep a dedicated brand Search campaign.
Is Google PMax worth it for a small ecommerce store?
Often, if the store has a clean product feed and enough conversions for the system to learn from. With very few sales a month, Search on exact and phrase match gives you more control while the data builds. Whichever you choose, set conversion tracking to reflect real revenue first.
How do I stop PMax bidding on my brand?
Apply a brand list as a brand exclusion in the PMax campaign settings, then audit the search term insights to confirm it holds. Don’t lean only on blanket negative keywords: Search Engine Land’s write-up of Brad Geddes’s SMX Next talk notes that approach can disrupt PMax’s learning. Re-check monthly, since new products and misspellings slip through.
Sources
- Data study: Is Performance Max cannibalizing your Search ads? (Adalysis)
- How Performance Max interacts with other campaigns in your account (Google Ads Help)
- About Performance Max campaigns (Google Ads Help)
- Is Performance Max cannibalizing your Search campaigns? (Optmyzr)
- How to make search and PMax campaigns complement each other (Search Engine Land)
- How to reduce low-quality leads from Performance Max campaigns (Search Engine Land)
- The Brand Search Defence Playbook (Ecommerce Circle)
