Google PMax tROAS Settings When Finance Owns Blended Targets
A practical settings playbook for Performance Max when the company-wide goal is a blended MER number — tROAS staircase, brand exclusions, and asset-group splits that protect organic brand revenue.
Quick answer
Set PMax tROAS 15-25% above your channel target to compensate for brand-traffic inflation, add an account-level brand exclusion list, and split asset groups by margin tier. Review the exclusion list weekly — PMax keeps re-discovering brand queries that would have converted organically for free, quietly lifting channel ROAS while dragging blended MER down.
PMax tROAS settings under a blended MER target
Configuring Performance Max tROAS, exclusions, and asset groups so paid ROAS gains don't come from cannibalising organic brand revenue.
When finance owns a blended marketing efficiency ratio (MER) — total revenue divided by total ad spend — Performance Max creates a measurement trap. PMax will happily hit a 5x channel tROAS by absorbing brand search and remarketing traffic that would have converted organically, so blended MER drifts down even as Google Ads dashboards look healthier.
The fix is structural, not bid-by-bid: a tROAS set above the channel target by a known cannibalisation premium, an account-level brand exclusion list refreshed weekly, asset groups separated by margin tier, and a feedback loop between Finance's MER report and the campaign settings.
This page assumes you've already accepted that channel ROAS and blended ROAS will diverge, and that finance is the system of record. If that framing is new, start with the parent topic on bidding on channel ROAS while forecasting on blended ROAS — it explains why the two numbers move in opposite directions during PMax scaling.
Everything below is operational: settings, lists, and review cadences you can apply this week to a Shopify or Magento store running PMax with a Merchant Center feed.
Why PMax quietly eats organic brand revenue
PMax's Final URL Expansion and broad audience signals push it toward the cheapest conversions Google can find. For most stores, the cheapest conversions are people searching your brand name — high intent, low CPC, near-100% conversion rate.
Those sessions would have converted via organic search or direct traffic at zero media cost. Letting PMax buy them inflates campaign ROAS by 30-60% while reducing blended MER by 5-15%, depending on how strong your organic brand demand is.
The dashboard lie
If your PMax ROAS jumps from 4.2x to 6.1x in a week with no creative change, you're almost certainly cannibalising brand. Check branded search impression share and direct-traffic sessions in GA4 over the same window — both will have dropped.
Setting the tROAS staircase
A tROAS staircase is a stepwise progression of targets — not a single number. You start permissive enough for the algorithm to gather conversion data, then tighten in 25-50 bps increments once the campaign exits learning.
For a finance-owned MER of 3.0x, the channel tROAS sits higher than 3.0 by your measured cannibalisation premium. If brand exclusions are weak, that premium is 20-30%; with a strong exclusion list it shrinks to 10-15%. So a 3.0x MER target usually maps to a 3.5-3.9x PMax tROAS in week one, stepping toward 4.2-4.5x by week four.
Increase tROAS only when you have at least 50 conversions in the trailing 7 days and CPA hasn't drifted up. Cutting too fast collapses spend and resets learning; the staircase exists to avoid that.
Typical settings by store profile
PMax tROAS configuration anchored to a 3.0x blended MER target, by store profile
| Store profile | Channel tROAS | Brand exclusions | Asset groups | Final URL expansion |
|---|---|---|---|---|
| Apparel, strong organic brand (>40% of revenue) | 4.0-4.5x | Account-level, weekly refresh | By margin tier (3 groups) | Off |
| Beauty SKU, mixed demand | 3.6-4.0x | Account-level, biweekly refresh | By collection (2-4 groups) | Off, then test on |
| Electronics, low brand demand | 3.2-3.6x | Light brand list | By price band (3 groups) | On with URL contains rules |
| New DTC launch, <€2M revenue | 3.0-3.4x | Brand + competitor list | Single asset group | Off until 50+ conv/wk |
| Multi-market (Shopify Markets) | 3.5-4.2x by market | Per-market exclusion lists | Per market, by margin | Off |
The asset-group split matters more than most teams realise: a single asset group flattens margin signals, so PMax optimises for the cheapest conversion regardless of contribution. Splitting by margin tier lets each group carry its own tROAS — 3.2x for clearance, 4.5x for full-price hero SKUs.
Maintaining the brand exclusion list
Brand exclusions live at the account level in Google Ads (Tools → Account-level negative keywords for brand lists, plus the Brand Lists feature in PMax campaign settings). Add every spelling variant, misspelling, and brand+product compound a customer might type.
Refresh weekly. PMax re-discovers brand queries as Google rolls in new audience signals or as your organic rankings shift. Pull the Search Terms Insights report every Monday, flag any query containing your brand token, and add it. A neglected list leaks 5-10% of brand revenue into PMax within a month.
Closing the loop with finance
Once a week, reconcile three numbers: Google Ads reported PMax ROAS, GA4 last-non-direct revenue for the PMax campaign, and the blended MER from the finance P&L. If channel ROAS rises while MER falls, the cannibalisation premium has crept up — tighten the exclusion list before raising tROAS further.
Document the staircase decisions in a shared log: date, old tROAS, new tROAS, 7-day conversions before/after, MER before/after. Finance will ask, and without the log every weekly review starts from zero.
Frequently asked questions
On most accounts with a curated feed, yes — at least until you have 50+ conversions per week and a mature exclusion list. Expansion sends traffic to pages PMax chose, including help articles and category pages that convert poorly and distort the algorithm's view of which products win.
Because PMax inflates channel ROAS by absorbing cheap brand and remarketing conversions that would have happened anyway. The channel target has to sit above the blended target by your measured cannibalisation premium — typically 15-25% — or you'll hit the channel number while missing the company number.
Two layers. First, account-level negative keyword lists applied to all campaigns. Second, the PMax Brand Lists feature (Campaign settings → Brand exclusions) — this works on the AI-matched brand signal even when the query string is slightly different. Use both; they catch different cases.
Only if any single group falls below the ~30 conversions/month threshold PMax needs. For stores doing €1M+ revenue, three asset groups by margin tier is usually fine. New launches should stay single-group until volume justifies the split.
No more than once a week, and only after 50+ conversions in the trailing 7 days. Smaller changes (25-50 bps) more often beat large jumps — PMax re-enters learning above ~20% bid swings and you lose two weeks of stability for a number you could have reached in three steps.
Branded organic sessions drop, direct traffic drops, and PMax-attributed conversions rise by roughly the same total. Total revenue is flat or slightly up while ad spend rose meaningfully — the give-away is that incremental spend produced no incremental sessions, only re-attributed ones.
Yes, if brand search volume is meaningful. A dedicated Brand Search campaign with manual CPC or a high tROAS lets you decide whether to defend SERPs against competitors, while keeping that traffic out of PMax's optimisation. It also gives finance a clean number for defensive brand spend.
Run a geo holdout: pause PMax in two matched markets for two weeks, keep it running in two others. Compare incremental revenue against incremental spend. The delta between observed channel ROAS and the holdout-derived incremental ROAS is your cannibalisation premium.
The principles transfer, but Shopping gives you query-level control PMax doesn't. With standard Shopping you can negative-match brand queries directly. PMax requires the account-level lists and Brand Lists feature because there's no query-level bidding surface.
Recompute the channel tROAS using your current cannibalisation premium, then step toward the new number in 25-50 bps increments over 2-3 weeks. Avoid setting the final target immediately — PMax will collapse spend, you'll miss the quarter, and learning resets.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.