How a Portsmouth business shows up in AI search (or doesn't).

The new question your customers ask

A would-be customer in Portsmouth used to type "plumber Portsmouth NH" into Google and get ten blue links. Now a meaningful share of them are typing "who is the most reliable plumber in Portsmouth, NH for an old house with original cast iron" into ChatGPT, or asking Gemini "where should I take my parents to dinner near Strawbery Banke for an anniversary," or pasting an inn's URL into Claude with the question "is this place actually walkable to the waterfront, or are the reviews lying."

These are not edge cases. AI assistants now answer roughly 12 to 18 percent of informational queries in the United States, up from under 2 percent a year ago, and that share is on track to reach 30 to 40 percent by 2027. That trajectory is the entire reason RalliGEO exists.

The relevant question for a Seacoast business owner is not "do I rank on Google." It is "when a model is asked to recommend a business like mine, does it know I exist, and does it describe me accurately." Those are two different questions, and the second one fails far more often than owners assume.

What we actually tested

For each of the 21 businesses we ran the same five-query template against eight engines. We did not ask "tell me about [business name]," because every engine can do that. We asked the kind of category question a real customer asks when they do not already have a name in mind:

• "Recommend a [category] in Portsmouth NH for [a specific use case]"
• "What are the best [category] in Greater Portsmouth, ranked"
• "Compare [category] options near Portsmouth NH"
• "Is [business name] a good fit for [specific use case]"
• "What do reviews say about [business name]"

The eight engines:

Engine	Owner	Why it matters in the Seacoast
ChatGPT	OpenAI	Largest assistant by usage. Default for most under-40 customers.
Gemini	Google	Built into Pixel phones, Gmail, Android. Quiet but enormous footprint.
Google AI Overviews	Google	The generated answer above the blue links. Most users do not realize they are reading an AI summary.
Copilot	Microsoft	Default on Windows 11 and inside Edge. Strong among 45 to 65 customers.
Claude	Anthropic	Heavy use in professional services. Different retrieval bias than ChatGPT.
Perplexity	Perplexity AI	Used by buyers who want sources. Disproportionately cites the businesses with clean structure.
Meta AI	Meta	Inside Instagram, WhatsApp, Facebook search. Reaches customers who never type a URL.
Grok	xAI	Inside X. Smaller, but cites very differently from the other seven.

What we found

Three patterns showed up over and over, in roughly this order of frequency.

1. The business does not exist to the engine

Asked to recommend a [category] in Portsmouth NH, the engine returns four to seven competitors and never mentions the business at all. This happened in at least one engine for all 21 sites we tested. It happened in four or more engines for 16 of 21. Owners reading this are almost always surprised, because they are easy to find on Google. Being easy to find on Google does not mean being retrieved by an AI assistant. The retrieval paths are not the same paths.

2. The engine knows you exist, but describes you wrong

This is the most quietly damaging pattern. The model retrieves your business, but assigns you the wrong category, the wrong specialty, the wrong location, or merges you with a similarly named business in another state. A Portsmouth inn confidently described as being in Portsmouth, Virginia. A restaurant described by a five-year-old menu that no longer reflects who they are. A plumber described as a heating company because most of their site copy is about boilers.

Customers act on what the model tells them. If the model tells a wedding planner that an inn is "primarily a business-traveler hotel," that planner does not pick up the phone.

3. The engine knows you, describes you, and never cites you

Some engines (Perplexity, ChatGPT in search mode, Google AI Overviews) attach citations to their answers. Many businesses we tested show up in the prose of the answer but are never the linked source. The engine learned about them from third-party listings, directories, or a single review aggregator. That means three things that matter to an owner:

• The traffic from that answer goes to whoever is the cited source, not to your site.
• You have no control over how your business is summarized.
• If the third-party listing is out of date, you carry the consequences.

Why this happens to good businesses

None of the patterns above are caused by the business being bad at what it does. The 21 businesses we tested include some of the best operators in the Seacoast. The patterns are caused by something more boring: the website was built for human readers and Google's crawlers, on the assumption that those were the only two audiences. AI engines retrieve and summarize differently. They reward different signals.

The structural causes we see most often, in rough order:

• Thin or missing entity definition. The site never declares plainly what the business is, in a form a model can quote back. Tagline copy looks great on a homepage and reads like marketing to a model.
• Service or menu information buried behind imagery. What is on the page as a photo is invisible to most retrieval; what is on the page as crawlable text is what gets quoted.
• No structured data, or structured data that disagrees with the visible content. Many sites have schema markup left over from a 2019 build that says they are a different category from what they actually do today.
• Inconsistent name, address, phone across the web. One spelling on the site, another on Google Business, another on TripAdvisor, another on the chamber of commerce listing. Models read all of these and average them, badly.
• Bot blocking on the homepage. A surprising number of small business sites are unintentionally blocking the user agents that AI engines crawl with, while still letting Googlebot through. The result is invisibility to ChatGPT and Claude while looking fine on Google.

None of these are catastrophic individually. Most can be repaired without rebuilding the site. The reason most Seacoast businesses are still invisible is not that the fix is hard. It is that nobody has told them the problem exists yet.

What changes when you fix it

We have not been running long enough to publish 90-day outcome data. We will, soon. What we can say from the audit work itself:

• Going from invisible in 5 of 8 engines to cited in 6 of 8 is achievable in most cases within one engagement. The lift is mostly structural, not content.
• Once a business is consistently retrieved, its conversion profile changes. AI-referred visitors arrive with intent that is closer to a referral than to a cold search click. They have already been told you are the right answer.
• The competitors who are already optimized show up in the answer above you. That is the cost of the gap, every day, compounding.

What an owner should do this week

You do not need to hire us to do step one. Open ChatGPT, Gemini, Claude, and Perplexity in four browser tabs. Ask each one the question your best customer would ask if they did not already know you. Do not name yourself. Read the answer. Note which engines mention you, which mention you accurately, and which cite you as the source rather than someone else.

That is your starting line. Most Portsmouth owners who run that exercise are unsettled by what they read. The good news is that the gap is mechanical. The work is real, but it is finite. There is a version of every business in this report that is fully retrievable. We have not yet seen one where it is not technically possible.