Service
ChatGPT is checking your references.
When a customer asks an AI assistant who to hire, we'd expect the answer to draw on what the web says about you: your listings, your reviews, your own pages, and whoever else mentioned you. We find out what that adds up to and work on the parts that are wrong or missing.
It asks around, the way a neighbor would.
Think of an assistant as the neighbor with an opinion on every contractor on the street. It hasn't hired any of them. It has heard things. Some answers come from a live web search and some from what the model absorbed in training. OpenAI, for one, runs separate crawlers for its search results and for model training. How many of your customers ask, we can't say. We haven't seen a number we'd put our name on. Our read is that it's more than last year, and more for the roof than for lunch.
The cheap way to learn is to listen. Run the free AI visibility check, or open ChatGPT and type what your best customer would type. Read the answer as if someone said it about you at a barbecue. We wrote up the customer who never reaches your site, who hears three names, calls one, and never shows up in anybody's analytics.
Wrong is worse than missing.
Assume an assistant can tell a stranger your old address with complete confidence. We'd go looking for the service you dropped years ago, the partner who left, the town you don't serve. Missing costs you a mention. Wrong can cost you the customer who believed it and then found the door locked.
As far as we know, there's no editor at these companies to call. So we work backward. We read what each engine says, then look for where a sentence like that could have come from: a dead directory listing, an old news item, a stale Google Business Profile, a Yelp page nobody claimed. We correct what can be corrected at the source and ask again. Some of it moves. Some of it takes its time, and some of it may not budge.
Say one clear thing about yourself, in every place.
Start with the sentence you control. In our Phoenix-area sample of 487 business homepages, 14% had no search description on the homepage. That's one of the few lines a business writes for machines about what it is. Having one is only a start. "Quality service you can trust" doesn't tell an assistant whether you fix pools or teeth.
We write the plain version: who you are, what you do, which towns, and who you don't serve. Then it goes on the site, the listings and the social accounts, with pages that back it up. The aim is that a machine reading ten sources finds one business, and not three that look alike.
Reviews and mentions are other people talking.
Your site is your own testimony, and we'd expect a careful reader, human or otherwise, to discount it some. So we set up a routine for asking real customers for reviews, and we look for honest places to be mentioned: a supplier's dealer page, a trade association, the local paper when you actually did something. The shortcut is stuffing your listing with keywords. Google says unnecessary information in a business name could get the profile suspended, and we wrote about that.
We don't buy reviews, write fake ones, or plant comments on forums. The technical half of this work, crawler access and markup and measurement, lives on the GEO page. Both rest on ordinary SEO, and if that's missing we'd do it before either.
How it works
Listen, trace, correct, listen again.
Ask what customers ask
We put real buying questions to the major assistants and write down the answers word for word, the right ones and the wrong ones.
Trace the sentences
For each wrong or thin answer we look for a likely source: a listing, a directory, an old article, your own site. If we can't find one, we'll say so.
Correct the record
We fix what we can reach, request changes to what we can't, tidy your own description, and start the review routine with your team.
Ask again
The same questions, on a schedule. You see what changed and what didn't, and we drop whatever isn't helping.
They made the technical side feel clear.
We talked about goals first — then platforms, content, search, and automation.
They get it done.
We had no idea where to start; they explained the build process and were patient with our many questions. All expectations and timelines were met. Double thumbs up!
Price it in a minute
Price the work on what AI says about you.
Tap Reviews or Local marketing and write "AI search" in the notes. The starting price appears on screen: from $575 a month, depending on complexity, setup and budget. The developer who'd do the work follows up.
- Every price published on the pricing page
- 0% financing with 25% down. Financed projects are hosted on the DataCram platform for the length of the plan. Shopify stores can't be financed.
- One developer from first call to launch
What do you want help with?
Where should we focus? (pick any)
Locations
Rather talk? (480) 525-7582. 0% financing on any project, 25% down. See all pricing
GPT & AI Search: questions we hear a lot.
Are customers really asking ChatGPT who to hire?
We'd expect some are. We haven't measured how many and we'd be wary of anyone quoting you a neat figure for your town. You can test it for free: ask your next ten customers how they found you, and run the AI visibility check to hear what they'd be told.
How does an AI decide which businesses to recommend?
Nobody outside those companies knows the weights. Our read is that it draws on what it can find and read: your site, your reviews, other sites' mentions of you, and how consistently all of that describes you. That's what we work on, and we'd distrust a recipe with exact proportions.
ChatGPT says something false about my business. Can you fix it?
We can't edit a model. We can look for where the claim likely came from, correct that source, publish the right fact somewhere easy to read, and check back. We'd expect answers from assistants that search the live web to change sooner than answers from training data. We won't promise when, or that it will. If the source is your own broken page, that's a repair.
How is this different from your GEO page?
GEO is the machine side: can the crawler get in, can it parse the page, and what's the count over time. This page is the reputation side: what gets said about you, by you and by others, and whether it's true. Plenty of businesses need some of both.
I'm in a regulated field. Can you collect reviews for me?
What a law firm, a clinic or a financial firm may ask of clients or say about itself is a question for your counsel, licensing board or compliance officer, and we don't answer it. Get their word first and we'll work inside it.
What does it cost?
It's ongoing marketing work, which starts from $575 a month, depending on complexity, setup and budget. See the pricing page. It often pairs with SEO because the groundwork overlaps. If you only want the picture for now, the visibility check costs nothing and the Deep Report is $49.
Find out who the machines say you are.
Ask the question yourself, then look at what the work costs. If the answer's wrong, tell us which part and we'll tell you what we'd try first.