Emergency Brand Audit: What AI Says About Your Locations

A 3.3-star automotive wash simply gained the AI reply.

When Annie Jackson, Director of Income Operations and Development at GatherUp, requested Google for a no-touch automotive wash that matches an SUV in Norfolk, VA, Google returned a 3.3-star enterprise and answered the clearance and hours questions above the star score itself. Question match outranked score.

That instance frames the session Jackson offered with Jason Wertham, Vice President of Overview Protection Operations at GatherUp: AI instruments assemble their description of every location from opinions, listings, and public net mentions, then repeat that description to prospects who by no means attain your web site.

Jackson and Wertham walked by means of the four-prompt emergency audit they run to floor what ChatGPT, Google AI Overviews, and Ask Maps say a few multi-location model right now, then the build, manage, defend rollout that modifications the reply.

The recap under covers the findings. The prompts, the rollout, and the audit handout reside within the full session. Watch it on demand.

How Do Prospects Use AI To Discover Native Companies Now?

They ask a full query and settle for the summarized reply. In GatherUp shopper information collected in fall 2025, 55% of customers had consulted Google or Bing AI summaries, 48% had requested ChatGPT a few native enterprise, and 31% had requested a number of instances.

Jackson’s automotive wash question reveals what that appears like in follow. As an alternative of “automotive wash close to me,” the search was “no-touch automotive wash for my SUV in Norfolk, VA.” Google pulled from its 300 million locations and 500 million evaluation contributors to return one enterprise, with clearance peak and 24/7 hours answered inline.

“Google answered my questions, however this enterprise is definitely exhibiting up as a 3.3 star,” Jackson stated. “It’s surfaced the context of my question above the star score.”

Wertham added that these instruments now consider who you’re and if you ask. “The time of day if you’re truly doing this question in Google Maps might affect which companies are getting returned in these outcomes,” he stated. An LLM that is aware of you personal an SUV or a big canine applies that context to each future native question, whether or not you restate it or not.

The session’s audit begins from this conduct shift: 4 prompts, from a brand-name question to a location-by-location spot check, that present you the reply your prospects already see. See the four audit prompts.

Do Google Opinions Feed AI Solutions?

Not from the itemizing itself: Google, Yelp, and the opposite main directories block LLM crawlers from studying evaluation content material on enterprise profiles. Opinions nonetheless help native rankings and conversion on the itemizing; they solely enter AI solutions by means of surfaces LLMs can crawl.

“The main listing service suppliers, Google, Yelp, and others, they don’t permit LLM instruments like ChatGPT and Claude to scrape or crawl the evaluation information on the enterprise itemizing,” Wertham stated. “You’ll discover they’re not citing particular opinions from these platforms.”

The identical opinions change into crawlable the second you republish them. Publish them to public social channels or embed them in evaluation widgets by yourself website and, in Wertham’s phrases, “now they’re truthful recreation for the LLM instruments to be pulling in.”

That determines which queries you possibly can win. When a buyer asks for “well-liked” or “extremely reviewed” companies, the LLM searches for evaluation textual content it could entry. Overview content material confined to the listing contributes nothing to that reply.

“In case you’re counting on the evaluation platforms to try this for you, it’s not going to be sufficient,” Wertham stated.

Motion merchandise: Republish your opinions the place AI can crawl them. The session covers which widget and social placements make review content readable, together with methods to carry the enterprise reply with the evaluation.

Wertham additionally broke down first-party review capture, the survey responses that by no means attain Google, and why one buyer’s 11,000 of them mattered. That segment lives in the on-demand recording.

Does Your Star Ranking Nonetheless Matter In AI Search?

It issues lower than evaluation recency and velocity. No AI reply in the session’s audit examples cited a mean star score; each one cited evaluation content material.

The patron information factors the identical route: 45% of customers prioritize evaluation recency over star score, 60% belief detailed written opinions over rating-only opinions, and 70% choose a evaluation request inside 72 hours of the transaction.

Wertham famous that buyers routinely override Google’s default “most related” evaluation type and change to “latest,” as a result of the newest evaluation predicts the expertise they are going to get. A excessive common constructed on years-old opinions carries much less weight than a present, regular stream.

“I’d slightly go to a enterprise with 1,000 opinions and a 3.9 or 4.2 than 30 opinions and a 5.0,” Wertham stated.

The session organizes the response into three workstreams, construct, handle, and defend: constructing constant listings and evaluation quantity, managing responses and monitoring inside a 72-hour response window, and defending the score you earned in opposition to policy-violating opinions and burial ways like review smothering. Walk through the build, manage, defend rollout.

Why Does AI Give A Completely different Reply About Your Enterprise Each Time?

As a result of LLM solutions behave like a slot machine, one question isn’t a dependable learn. Jackson cited SparkToro analysis by which completely different folks requested LLMs the identical query throughout units and accounts, and the outcomes by no means returned in the identical order.

“Asking AI a query is sort of like a slot machine,” Jackson stated. “It’s going to be giving again related information, however every time it’s going to look just a little in another way.”

Place is the unsuitable metric for AI visibility. Whole citations, the breadth of sources feeding the reply, predicts whether or not your model seems in any respect. Your model can miss one system’s reply solely and lead the subsequent.

Google revealed its personal information to optimizing for generative AI and up to date it this month. Wertham flagged one change with tooth: the AI slop penalty. Google now detects low-value AI-generated content material and, in his phrases, is “primarily penalizing companies” for it. Generic AI weblog posts and glorified FAQ scraping targets now value you slightly than getting ignored.

Motion merchandise: Run your audit prompts in incognito or temporary-chat mode so saved context stops shaping your outcomes, and re-run them on a schedule. The session reveals the monthly re-run method for measuring whether your visibility work is moving the answer.

Q&A: Most Useful Questions From The Webinar

Q: What’s the quickest factor I can do that week to alter what AI says about my firm?

“Deal with your listings. Make certain your listings are all appropriate and all constant, no matter platforms you’re on. After which just remember to are evangelizing your opinions off of the third-party listing the place you’re receiving them. Publish them to your social media platform, put up them to a bit of your web site.” Jason Wertham, 49:23 in the on-demand recording.

“Be sure to have the fundamentals down. Get the fundamentals down, be sure these are set, after which you possibly can transfer into the extra elaborate issues.” Annie Jackson, 50:30 in the on-demand recording. Jackson illustrated the purpose with an area restaurant whose Fb web page listed the proprietor’s private cell quantity; he by no means knew why the calls saved coming.

Q: How lengthy earlier than content material modifications truly present up in AI solutions?

Annie answered: small info transfer quick, positioning strikes slowly. Retailer hours and telephone numbers replace rapidly, however “what you’ve been recognized for will take just a little bit longer,” typically two weeks to a month with an extended tail past that. She pointed to your individual web site because the quickest lever: a brand new providing has to look by yourself channels first, as a result of opinions is not going to announce it for you. Full answer at 54:19 in the on-demand recording.

Q: My weakest location has previous unhealthy opinions that maintain exhibiting up. Do I’ve to attend for them to age out?

Jason answered: age fades a evaluation’s relevancy naturally, however keyword-heavy opinions and opinions from Native Guides maintain rating longer, and emoji reactions maintain a evaluation from slipping even once they add no upward momentum. Coverage-violating opinions keep disputable at any age; his review defense group removes opinions greater than ten years previous frequently. The dependable repair is quantity and velocity, since recency outweighs content material in long-run relevancy. Full answer at 47:33 in the on-demand recording.

Q: How ought to franchisors deal with this when every franchisee controls their very own profile?

Jason answered: franchise status administration fails on the consistency hole, as a result of every franchisee owns their itemizing whereas the model absorbs the AI reply. Set up finest practices, supply white-labeled or accomplice instruments franchisees will use, and put a playbook of their fingers. He additionally advised operating the audit prompts on behalf of franchisees and training them on the outcomes, since one location’s unsuitable AI reply prices the entire model. Full answer at 55:46 in the on-demand recording.

Watch The Full Webinar

The on-demand session accommodates the 4 emergency audit prompts and downloadable handout, the whole construct, handle, defend rollout, the month-to-month measurement technique, the evaluation protection walkthrough for disputing policy-violating opinions, and Jackson’s AI narrative audit supply. Watch the full webinar on demand.