On July 24 Google started rolling AI Max into Standard Shopping campaigns.
If you run ecom, that's the one that matters. Search was the warm-up. Shopping is where your money actually lives.
Google's number is a 5% lift in conversions at a similar CPA. I believe it. I've also watched what the same trade does to accounts that weren't ready, so let's do both halves.
THE TRADE
What you hand over, what you get back
Standard Shopping was the last campaign type where you still held the wheel. You picked the products, the feed decided eligibility, and the structure was yours. That was the whole appeal after Performance Max swallowed everything else.
This is the surface AI Max now decides your placement on, query by query.
AI Max changes three things at once.
What changes | What you gain | What you lose |
|---|---|---|
Query matching | Long-tail and conversational searches you never bid on | Control over which searches you appear for |
Ad text generation | Copy written per query, at a scale no human matches | Sign-off on what gets said about your products |
Landing page selection | Google routes each click to its best-guess page | Certainty the click lands somewhere that can sell |
The first row is mostly good. Conversational search is real, it's growing, and your keyword list was never going to keep up with it.
The third row is where accounts bleed. Same story as Search, and it's the setting almost nobody turns off in time.
THE REAL VARIABLE
It's not the setting. It's your feed.
The model is only as good as the product data you hand it. Give it a clean, complete, well-structured feed and it will find demand you'd never have written a keyword for. Give it what most brands actually have and it will confidently sell the wrong thing to the wrong person.
The feed as it actually looks in most accounts. Every empty cell is a decision handed to the model.
I pulled a Merchant Center account last month that looked healthy. Green status, products approved, no suspension, nobody worried.
2,041 items missing a product highlight. 1,284 missing shipping weight. 312 with a mismatched GTIN.
None of it stops the products serving. All of it decides which ones Google chooses to scale.
We fixed the shipping weight field. One field, across the feed, in an afternoon.
Conversion rate on Shopping went from 1.1% to 2.2% inside a week. Clicks barely moved. Same traffic, twice the buyers, because the shipping cost stopped being a surprise at checkout for half the catalogue.
The same week in Shopify. Nothing changed on the store, only what the feed told Google.
The feed isn't admin. It's the only part of Shopping the algorithm can't guess at, and it's the part everyone leaves to whoever set up the store.
Revenue went from $117K to $234K a month, roughly $1.4M a year, from one field. No new budget, no new creative.
THE HELPFUL PART
Merchant Center will now tell you
Google shipped something useful the same day. Merchant Center now has AI Summary Insights, a Gemini panel that surfaces what's actually hurting you rather than making you go looking.
It'll tell you clicks dropped on twelve products after a price change. It'll tell you your shipping speed is slower than comparable retailers. It'll tell you which 2,041 items need product highlights.
Use it. It's the first Google feature in a while that shortens the weekly routine instead of adding to it. It won't fix anything for you, but it stops the diagnostic taking an hour.
DO THIS BEFORE YOU ENABLE IT
Four checks, in order
Run Merchant Center diagnostics and clear every item issue. Not the ones blocking approval. All of them. AI Max reads what you give it.
Custom labels 0 through 4. Margin band, stock depth, seasonality, whatever routes budget for you. If they're empty, the model has no idea which products you want scaled.
Landing page selection. Decide deliberately. If you turn it on, pull the landing page report weekly for the first month, without exception.
Baseline first. Screenshot your Shopping conversion rate, CPA and blended ROAS in Triple Whale or Northbeam before you flip anything. You cannot measure a 5% lift against a memory.
That last one catches people out. Enable it, feel good, and three weeks later nobody can prove whether it helped.
THE HONEST TAKE
Turn it on. After the feed work.
I'm not anti-AI Max. On a clean account it's a genuine lift and the conversational demand it picks up is demand you currently have no way to reach.
But it multiplies whatever you feed it. Clean feed, clear labels, honest conversion signal, and it compounds. Broken feed and it scales the broken thing faster than you can catch it, with a dashboard that stays green the whole time.
Do the feed work first. It's two weeks of unglamorous effort and it's worth more than the setting.
Free. Every field AI Max reads, and what happens when each one is wrong.
Want me to look at your feed before you enable it? Reply with your Merchant Center diagnostics screenshot and I'll tell you what I'd fix first. Just hit reply.
Talk soon,
Patrick