Getting into AI answers when you have no brand recognition

None of that is a trick, and none of it happens in a fortnight.

The frustration is real, though, so let me name it plainly. You ask an assistant who the best provider in your city is, and it names four businesses that have been around for fifteen years and one that closed. You are better than two of them. You are not in the answer.

That is not a mystery and it is not a conspiracy. It is a retrieval system doing what retrieval systems do, which is favor the well-documented.

Why does recognition matter if this is not a popularity contest?

Because recognition, in this context, means documentation.

A system assembling an answer about your category needs sources that describe businesses in that category. A fifteen-year-old competitor has a long tail of local news mentions, forum threads, review text, association pages, old blog roundups and directory records that all agree with each other. That is not fame. It is a large, consistent evidence base.

Your business, six months old, might be excellent, and there is almost nothing to retrieve. The system cannot cite what does not exist.

You are not being outranked. You are being out-documented. The distinction matters because it points at what to do.

There is a measurable version of this problem, and it is worth knowing before you spend anything. In the University of St. Gallen study of AI search visibility, citation concentration across engines averaged a Gini coefficient of 0.715, meaning a small set of sources absorbs most of the citations. That is the mathematical shape of "the same five businesses keep getting named."

How big is the gap, actually?

Count it rather than feeling it. The arithmetic takes an hour and it changes what you do next.

Your competitor. Fifteen years trading. 340 reviews. 6 local press mentions. 4 association or chamber pages. 3 forum threads where someone recommended them by name. Call that 353 retrievable sources that describe the business.

You. 22 reviews. 0 press mentions. 1 directory record. Call it 23.

The gap is 330 sources, and you are not closing it this year. If you earned one new third-party mention a month, which is a good rate for a small business doing real outreach, you would need twenty seven years.

Now split the gap into the part that is a stock and the part that is a rate. Their 353 sources are a stock, accumulated, and you cannot catch it. Recency is a rate, and a rate resets every month for everybody.

Run the rate. At 5 new reviews a month you add 60 in a year. Your competitor at the same pace adds 60 to a base of 340, which moves them from 340 to 400. You move from 22 to 82. In percentage terms your review base almost quadrupled and theirs grew by less than a fifth. On the recency and sustained-influx signals, which Whitespark's 2026 survey of 47 local search experts scores separately from review count at 164 and 154 points, you are competing on equal ground from the first month.

And the last column is the winnable one. Count the specific questions where their 353 sources say nothing. Saturday hours. A particular insurance plan. A private room for twelve. A gluten-free fryer. If you find eight, you have eight queries where the documentation gap is zero, because nobody documented it.

What can an unknown business realistically win?

The narrow question. Every time.

Broad queries in an established category are decided by the weight of existing evidence, and you will not accumulate fifteen years of it this quarter. Specific queries are decided by whether any source addresses the specifics, and frequently none does.

So the winnable ground is where you are genuinely more specific than everyone else. Not "best dentist in San Jose." The Saturday-hours dentist who takes a particular plan and sees anxious adults. Not "Korean restaurant Honolulu." The one with a private room for twelve, open late, that handles a gluten-free request without drama.

Those questions get asked constantly and answered badly, because the businesses that could answer them do not say so anywhere. Your site describes you the way every competitor's site describes them, so retrieval has nothing to distinguish you with.

Write the specifics down, in plain sentences, on pages that are indexed. That is most of the work, and it is free.

Then check your expectations against reality, because AI answers change every time you check them and appearing once is not a win, nor is missing once a loss.

Does the way you write the page actually matter?

A little, in a way that has been tested once, and the honest version of that test is narrower than the pitch built on it.

The KDD 2024 paper that introduced the term generative engine optimization tested nine content tactics against a benchmark of 10,000 queries. Against a do-nothing baseline of 19.3, the winners were adding direct quotations at 27.2, adding verifiable statistics at 25.2, and adding inline citations to credible sources at 24.6.

Two results from the same paper matter more for an unknown brand than the winners do. Keyword stuffing scored 17.7, below the baseline, meaning it actively made things worse. And writing in a more authoritative tone produced no significant improvement at all. You cannot talk your way in.

The limits are real and you should state them before anyone sells you a package. The experiment fed cleaned text from five fixed Google results to an older model. Effectiveness of the same tactic ranged from 19.1% to 35.8% depending on subject category. And nothing in the benchmark was a local commercial query, so it has never been validated for the thing you are trying to do.

What it supports is modest and useful: cite your sources, quote real people, publish checkable numbers, and stop stuffing. That is a description of good writing, which is the pattern in this whole subject.

How do you become an entity the systems recognize?

By making it impossible to think you are two businesses, or a business that closed.

One legal name used consistently. One address written the same way everywhere. One phone number, current, everywhere. One category set that describes what you actually do. One website that is indexed. A Business Profile that is claimed, complete and active. Completeness of the profile is itself a scored factor at sixteen on the 2026 list.

That sounds like a checklist from 2015 because it is. It also still governs, and entity consistency across the web is the piece most new businesses skip while chasing something more exciting.

Small things carry disproportionate weight here. Hours are the field that goes stale fastest and the field assistants get asked about most, and wrong holiday hours cost you rankings and customers at the same time. "Business is open at time of search" entered the 2026 ranking factors survey at number five with 189 points. A business that is closed when the system says it is open is the failure mode these products are most sensitive to.

Then get reviews flowing, immediately, and keep them flowing. Whitespark's practitioner testing of AI Mode local answers in May 2026 found reviews and unstructured citations, meaning blogs, news and community pages, to be the heaviest inputs. Darren Shaw's blunter version, on rankings generally: the moment you stop getting new reviews, your local rankings start to slip. Recency matters more than count, which is unusually good news for a business with twelve reviews and a pulse.

Which surface should you try to win first?

Not the one everybody names, and this is the most actionable finding in the research.

Citation concentration differs sharply by engine. In the St. Gallen data, Google AI Mode was the most concentrated at 0.782, meaning the smallest set of sources takes the largest share. Perplexity was the least concentrated at 0.671, spreading citations across more sources than any other engine measured.

For a business with no documentation base, that ordering is your priority list. The engine that hands citations to a wider set of sources is the one where a new entrant can appear at all. The engine everyone asks about is the one where the incumbents are most entrenched.

Two cautions. This is a measurement of citation distribution, not a promise about your business, and engine behaviour changes without notice. And Perplexity was also the least accurate engine in a separate controlled test of an unfamiliar brand, misidentifying it outright, so appearing there is not the same as being described correctly. Measure there because the signal is readable, not because it is the market.

How long does this take?

Longer than a campaign and shorter than you fear, and I am not going to put a number on it because it depends on your category's density.

What I can tell you is the order things move. Profile and entity fixes register fastest. Review flow shows up next. Third-party mentions accumulate slowly and then compound, because each one makes the next easier to earn.

Where this advice breaks down: in a category with very few competitors and very little published material, an unknown business can appear in answers almost immediately, because the bar is low and nobody documented anything. In a dense urban category with four fifteen-year incumbents, the narrow-question strategy is the only one available and it will take several quarters. Count the sources before you set an expectation.

What does not work, and Google names these directly: special files such as llms.txt, artificial content chunking, rewriting content for AI systems, and seeking inauthentic mentions. That last one is the specific temptation for an unknown brand. Buying mentions to simulate documentation is exactly the shortcut being sold, and it is on the list of things that do not work. It also carries a subreddit ban and, in regulated categories, a federal rule you do not want to meet.

What to do this week

1. Count the gap. Their retrievable sources against yours, using the arithmetic above. Write both numbers down. 2. List the eight specific questions where their sources say nothing. That list is your content plan. 3. Write ten sentences stating what is specifically true about your business that competitors cannot say. Concrete facts, not adjectives. 4. Put those sentences on indexed pages, in plain language, near the top. 5. Complete every field on your Business Profile, including services, hours and holiday hours. 6. Ask every customer for a review starting today. Not a campaign. A permanent habit tied to a moment in your service. 7. Find one local publication, forum or association that already covers your category and get into it legitimately this quarter.

If a client or partner asks you how any of this works, what to tell someone who asks about AI search will save you an awkward meeting.

Want to know what is recognition and what is fixable?

Send your business name and the name of your best-known local competitor to eric@seod.com and I will count both documentation bases, then tell you which parts of their advantage you cannot shortcut and which parts are simple gaps you could close in a month. No charge, no pitch.

I offer this because the split between those two is the whole strategy for a newer business, and almost nobody separates them before spending money.

Otherwise, there is more on AI search and AEO here.