The best marketing tool I used this week wasn't a marketing tool

Everyone's asking whether AI can make more stuff. That's the wrong question.

For a day, I let Google Merchant Center make me feel stupid.

My product feed got rejected. No Shopping ads. For a store that does real revenue through Google, that is not a fix-it-whenever problem. It is a you-are-currently-lighting-money-on-fire problem.

So I did what you're supposed to do. I opened the policy docs. I read the rejection notice, written in the native dialect of a company that has never once wanted you to understand a rejection notice. I checked the footer. I audited product pages. I made a guess. I resubmitted. I waited. I guessed again.

This is the work, right? The job? No. This is the tax you pay before you get to the job.

Here's what I actually did next: I pointed Fable 5 at the site and told it to audit the thing.

It came back with the real culprit in about a minute: my refund policy directly contradicted my own sitewide guarantee. Every product page promised a 30-day money-back guarantee with no hoops. But the actual policy page was a stock Shopify template that said returns had to be unworn, with tags, in original packaging, and explicitly listed beauty products as non-returnable.

Read literally, my legal page said my own products could not be returned at all, while the store promised unconditional refunds. Google's crawler compares promotional claims against policy pages, and that exact mismatch is a textbook misrepresentation flag.

Fable 5 turned that into specific fixes, along with a missing shipping page, social-proof claims with no visible reviews behind them, and a contact page with no findable email or address. I read every one, kept the right ones, pushed them live, and resubmitted.

AI did not do the job. It removed the drag.

The lazy read is that AI replaced a marketer. It didn't. I still had to know that a Merchant Center rejection was quietly bleeding revenue. I still had to know which proposed fixes were right and which were confidently wrong. I still had to own pushing them live.

AI did not do my job. It did the part of my job I hate.

And that part is the part nobody posts about. Everyone's asking whether AI can write their blog posts. Everyone's asking whether AI can spin up their images. Everyone's asking whether AI can crank out a hundred more product descriptions.

Wrong question. Every one of them. That's asking AI to help you make more noise, for a feed that is already 90% noise.

The better question is uglier

What annoying, revenue-blocking, I'll-get-to-it-later problem can you hand off and have fixed before lunch?

• Approvals.
• Feed errors.
• The checkout edge case that only breaks on mobile.
• The QA pass you keep skipping.
• The report you rebuild by hand every Monday.
• The broken workflow everyone routes around because owning it is not anybody's job.

None of that makes a good AI-is-changing-everything post. It just quietly hands you back the hours you were spending as a human spell-checker for a machine.

Most brands are going to use AI to produce more. The operators who win are going to use it to remove friction.

Yesterday's side quest took twenty minutes. That's the whole post.