Google AI now writes your Charleston ad copy
Google documented AI review summaries inside Search ads. It pulls reviews from across the web, writes them into your ad, and you cannot edit a word of it.
Google published a help page this week that decides who writes your ad text, and the answer is no longer you.
The page is titled About review summaries. The opening line does not hedge. Review summaries are "AI-powered summaries of online reviews that appear directly as part of your Search ads." Google's AI reads reviews, writes a line of copy, and puts it inside the ad you are paying for.
If you run Google Ads for a Charleston service business, that sentence is your whole week.
What the document actually says
I read the help page rather than the coverage of it, because the coverage is already drifting. Four lines matter.
On where the material comes from: "Google AI automatically aggregates online reviews from across the web to generate review summaries." Across the web. Not from your Google Business Profile. Not from a list of approved partners. Google does not name a single source site on that page.
On tone: "These summaries are positive only, and represent customer experiences and key insights about your business."
On control: "Because review summaries represent user feedback, advertisers cannot directly edit or customize the generated text."
On reach: Search campaigns, all positions on Google.com, in the US, Australia, the UK and India, in English, on all devices, across every advertiser vertical except sensitive categories such as pharmaceuticals. Every Charleston roofer, plumber, HVAC shop, dentist and personal injury firm running Search ads sits inside that description.
What the page does not contain is just as loud. No opt-out instructions. No reporting column. No minimum review count, no recency rule, and nothing about how the system behaves when a business has mostly rough reviews. The only adjacent link points to Google's guide on account-level automated assets, and the page never confirms that the toggle there applies here.
This was a small experiment four months ago
The timeline is the part I keep turning over.
On July 1, Darcy Burk posted a screenshot on X of an AI-written summary sitting under a Google ad description, and she was not happy about it. Barry Schwartz picked it up at Search Engine Roundtable the same day. Google's comment then: "This is a small experiment to see if adding AI-generated context to Search ads helps people make more informed decisions." The test even carried a disclaimer in the ad saying Google AI responses are generated independently and can make mistakes.
By September a "What customers love" block was showing up, labeled as AI-generated from store rating reviews. This week it has a help document and four live markets.
Small experiment to documented product in under four months. (I have stopped reading the word experiment as a reason to wait. It is a release schedule.)
The two words that should change your plan
Across the web.
Every Charleston agency, mine included, has spent two years telling service businesses that Google reviews are the ones that count. For map pack ranking, that is still basically true, and I wrote the long version of that argument in why fresh Google reviews now beat review count.
For paid Search ads, as of this week, it is wrong.
Google will not say which sites it pulls from, which means you have to assume the full footprint: Yelp, Angi, the Better Business Bureau, Facebook, Nextdoor, Houzz, Thumbtack, Birdeye, whatever industry directory your trade association runs. The Angi profile you abandoned in 2023 because the leads were garbage is still live, still collecting the occasional review, and may now be feeding sentences into an ad you are paying four dollars a click for.
That reframes a line item. Cleaning up off-Google review profiles used to be housekeeping you did when you had a slow month. It is now ad creative work, and it belongs in the same conversation as your Charleston digital marketing budget rather than in a someday folder.
At this point a lot of people are thinking: positive only, who cares
Fair objection, and it was my first reaction too. If Google is only going to say nice things about me, let it.
Here is where that breaks. Positive is not the same as accurate, and it is definitely not the same as on-strategy.
Picture a Mount Pleasant HVAC company that spent this year moving away from small repair calls and toward full system installs, because installs are where the margin is. Five years of reviews praise them for being cheap and fast on a Saturday afternoon capacitor swap. Google's AI reads that history, writes "customers praise their quick, affordable repairs," and staples it to an ad bidding on air conditioning installation. Every word is positive. Every word is also pulling the wrong customer.
Same shape for a roofer whose older reviews celebrate patch jobs while the business now wants full replacements, or a dental practice whose reviews love the hygienist while the ads are selling implants.
And you cannot edit it, cannot see a reporting column for it, and have no documented switch. That is three different kinds of blind. Advertisers already lost control of headline assembly, then of asset creation. Now the social proof writes itself. This is the same pattern I flagged when Local Services Ads started billing Charleston pros for calls they never answered: the platform adds an automated layer, the invoice stays yours, and the controls arrive a year later if they arrive at all.
What I would do this week
Ninety minutes, maybe two hours if your review footprint is messy.
- Inventory every review profile with your name on it. Search your business name plus Charleston, plus your name plus the nearby towns you serve. Yelp, Angi, BBB, Facebook, Nextdoor, Houzz, Thumbtack, trade directories. Write down the URL and the star rating of each one. Most owners I talk to are on four or five platforms they have not logged into since setup.
- Read the newest ten reviews on each, not the ratings. You are looking for what the words say you are good at. If the recurring praise does not match the service you are currently bidding on, you have a mismatch that Google is now free to publish.
- Claim and correct the dead ones. A profile you do not control is a profile you cannot improve. Claiming an abandoned Angi or Yelp listing takes about ten minutes each.
- Point your review requests at the service you want more of. Ask the install customer for the review, not just the easy repair customer. Over a quarter that shifts what any summary of your reviews can honestly say.
- Check your automated assets settings in Google Ads anyway. Google has not confirmed the account-level toggle governs this, but look at what is on, and screenshot it. If the control shows up later, you will want to know what the default was.
One more thing worth watching. Barry Schwartz also noted Google testing a scrolling marquee treatment for ratings and reviews in ads. Review content is not a footnote in paid search anymore, it is becoming the creative.
Start with step one this afternoon. Open a notes file, search your own name, and list every site with reviews about you on it. That list is now your ad copy, whether or not you have ever looked at it.