Service · Shopping & Performance Max
Shopping that sells to strangers, not to your own fans
Performance Max will happily spend your budget re-buying people who already searched your name. We run it the strict way: feed hygiene first, brand exclusions always, return judged against the 2.87 ecommerce ROAS average, and a carbon estimate per channel in every report.
Why does the feed come before the campaign?
Shopping auctions are entered by product data, never by keywords. A title that starts with your internal SKU name instead of "recycled wool jumper" simply misses the auctions that matter, whatever the bid. Week one of every engagement is feed work: titles rebuilt around real query language, categories and GTINs fixed, availability synced. On the refill-goods account in this write-up, feed fixes alone moved click-through before any bidding change: the rebuild took Shopping ROAS from 2.1 to 3.4 inside 90 days.
What do we let PMax do, and what do we refuse?
- Allowed: prospecting across Shopping, YouTube, Display and Gmail once search owns brand and proven non-brand queries
- Allowed: creative testing with your real product photography; AI-generated variants only with your written sign-off
- Refused: launch without brand exclusions, ever
- Refused: judging PMax on blended numbers that include your own brand demand
- Refused: URL expansion onto journal or careers pages that cannot convert
The refusals exist because PMax reporting flatters itself. Post-exclusion numbers start lower and mean more, and they are the only numbers we print.
What counts as a good return?
| Benchmark (sourced) | Value | Reading |
|---|---|---|
| Ecommerce ROAS average, 2025 | 2.87 | the middle of the pack |
| Ecommerce ROAS median, 2024 | 2.04 | half of brands sit below this |
| Common "healthy" bar | 4:1 | top-quartile territory, margin-dependent |
| Shopping/gifts CVR median, 2026 | 4.01% | the conversion context for DTC |
Your target comes from your contribution margin, not from a slogan: a brand at 50% margin can grow profitably at 2:1 while a 25%-margin brand needs nearly 4:1. We set the number during the paid-search onboarding, write it down, and report against it monthly, next to the carbon column explained in our carbon method.
Questions we actually get
Titles rebuilt front-loaded with what shoppers type, GTINs and product categories corrected, availability synced so ads never sell out-of-stock items, and images that meet Shopping's requirements. Feed quality decides which auctions you enter at all; bidding only decides how hard you compete once entered.
Because PMax without them re-buys people already searching your name, then reports those cheap conversions as its own performance. With exclusions, its reported return reflects new customers. Every PMax number we show you is post-exclusion, which is why our ROAS figures start lower and mean more.
The 2025 ecommerce average landed at 2.87, with the 2024 median at 2.04, so anything above 3 is genuinely good and 4-plus is top-quartile. Your margin decides what is profitable: 50% contribution margins can live at 2:1 while 25% margins need close to 4:1.
Per the Global Media Sustainability Framework, each channel gets an emissions estimate from impression volume, format, and device mix; a programmatic impression runs roughly 1 to 5 grams of CO2e. PMax's display and video placements weigh more than Shopping's product tiles, so the report shows the split by channel.
Feed check first
Send your store URL. The working session starts with what your feed is doing to your auctions right now.
Book a working session