How much AI search optimization costs a small business, and what the money pays for

The short answer

AI search optimization is sold three ways: as software you run yourself, as project or retainer work from an agency, and as a flat monthly service that does the work and measures it. What you pay depends less on the label than on what is included: whether anyone fixes your data, whether anyone publishes the facts the engines quote, and whether anyone measures the answers every week.

Our own price is two thousand dollars a month, month to month, with no contract and no setup fee. It is on our pricing page with everything it includes. We are not going to quote other firms' prices here. They vary, they change, and we have no checked source for them. What we can do is show you what the work is, so you can judge any quote, including ours.

What the work actually is

Getting a local business named by ChatGPT, Gemini, Perplexity, and Claude is mostly technical work, and it falls into four parts. Each one costs time, and a quote that leaves one out is a quote for less work, not a cheaper version of the same work.

Measurement. Someone has to ask the engines the questions your customers ask, the same way every week, and record which businesses each answer names, cites, and recommends. That is not one query. Our Phoenix window and door index, in its newest weekly run, was built from 86¹ answers, and they named or cited 11² different businesses. That is one trade and two engines so far, Claude and Gemini, with ChatGPT and Perplexity to be added. Every engine added and every repetition adds API cost and answers to read.

Consistent business data. An engine has to work out that your website, your Google Business Profile, your Yelp page, and your BBB listing describe one business. That means one schema.org JSON-LD entity on your site with a stable identifier and links to your profiles, and the same name, address, and phone on every directory the engines read. BrightLocal found Yelp used as a source in 33%³ of its local test searches across ChatGPT Search, Gemini, Google AI Mode, and Perplexity, so the directories are part of the job, not an extra.

Access. Your robots.txt and your host or CDN must let each engine's search crawler in: OAI-SearchBot for ChatGPT, Claude-SearchBot for Claude, PerplexityBot for Perplexity, and Googlebot for Google's AI features. This is cheap to check and expensive to miss. Our guide to AI crawler access covers it.

Citable facts. The engines quote what they can read without guessing: price ranges, specs, service areas, install timelines. Publishing those as structured data and as dated tables, under a named author from your team, is the slowest part of the work, because the numbers have to come from your own jobs.

And then all of it has to be kept true. Listings drift, a plugin update duplicates your schema, a firewall rule starts blocking a crawler. Maintenance is where much of the ongoing cost sits.

The three ways it is sold

Software you run yourself

A growing set of tools track how brands appear in AI answers. They usually charge a subscription, often priced by how many prompts, engines, or brands you track. They are good at the measurement part, and some go further into content or crawler analytics.

What you are buying is a dashboard. Someone still has to decide which questions to track, read the results, and do the fixing: the schema, the listings, the published facts. For a business with a marketing person who has time for that, software can be the lowest-cost route. For an owner who is also running crews, the dashboard tends to become one more login.

Agency work

SEO and marketing agencies increasingly offer AI search work, under names like generative engine optimization or answer engine optimization. It is sold as a one-time project (an audit and a cleanup), as a line on an existing retainer, or as a new retainer.

The range here is wide, because "AI search" can mean anything from a real technical program to the same blog posts with a new heading. Ask three questions of any quote:

• What will you fix, and where? Name the schema changes, the directories, the crawler checks.
• What will you publish, and whose numbers are in it? Generic posts are what many agencies already ship. Engines have little reason to cite them.
• How will you measure it? Which questions, which engines, how often, and whether you see the raw answers.

A flat monthly service

The third model bundles the work and the measurement into one monthly price. That is what we sell. For two thousand dollars a month, month to month, a client gets the audit and plan, one entity graph on the site, consistent listings, crawler access monitored, prices and specs published as structured data and datasets, every update pushed out through the sitemap, llms.txt, IndexNow, and Bing Webmaster Tools, and a weekly share-of-voice report across the four engines.

The advantage of a flat price is that it covers the unglamorous maintenance, which per-project pricing tends to leave out. The disadvantage is that you pay monthly whether or not a given month is busy. That is why we think "no contract" matters: if the work stops being worth it, you stop paying.

What drives the cost up or down

Whatever model you choose, a few things decide how much work there is:

• How messy your data is. A business with duplicate schema, several phone numbers, and old addresses on directories needs more cleanup at the start. Before calling a second listing or phone a mistake, though, check whether it belongs to a separate line of business. Sometimes it is deliberate, and the job is to keep the two clearly distinct.
• How many services and cities. Each line of business and each city you want work in adds questions to measure and facts to publish.
• How many engines are measured, and how often. Weekly measurement across four engines, with repetitions, costs more than a one-off check of one engine, and tells you far more. See how to track whether ChatGPT mentions your business.
• Whether you have numbers to publish. Prices, install times, and specs from your own jobs are the most citable thing you own. If you can share them, the content work goes faster.

What no price buys

No one can sell you a ranking in ChatGPT. The engines decide what they say, and a single answer varies from one asking to the next. Be wary of any quote that promises a position. What you can buy is measured output: citations tracked, listings fixed, content published, datasets live, and a weekly number that shows whether the engines are naming you more often.

What this means for an owner

Start with what it costs you to find out where you stand, which should be nothing. Our free AI visibility audit crawls your site the way the AI crawlers do, checks crawler access, structured data, and whether your name, address, and phone agree, and emails you a score and the findings that matter most. Then compare quotes on what they include, not what they are called. If a quote covers measurement, data, access, citable facts, and maintenance, it is quoting the whole job.