Answer: No. Age is not what the research measured. One study checked 4,776 restaurants and bars, then asked the AI apps ChatGPT, Perplexity, Gemini, and Claude where to go. The ones that got named were more likely to have a website, show prices, and be written about on other sites (Source: Pitenin, 2026). None of that takes years.
A new owner usually arrives at this question from search engines, where a brand new website really can sit and wait. The worry then gets carried across to ChatGPT, Perplexity, Gemini, and Claude, where the published research points somewhere else. Not one study below recorded how long a business had been open. Every one of them sorted businesses by how well known they already were, which often travels with age and sometimes does not. What follows is what was actually measured, the one place where being brand new genuinely costs something, and the short list of what to do about it.
Does AI care how old a business is?
No, not in anything anybody has measured. The large public studies sorted businesses by how well known they already were, meaning how much had been written about them where anybody could read it. Not one of them recorded a founding date.
The first audit put 215 business questions, covering 19 trades and industries, through about 37,000 runs. It scored the answers against a list of 533 brands, sorted into five levels by how widely known each one was. Between 48% and 52% of the specialist and regional brands never came back in a single run (Source: Jack, Lehman, Maloney, and Xu, 2026). The paper is explicit about what its levels measure: how known a brand is inside its own field, not its revenue and not its market share.
The second tracked more than 100,000 answers about more than 100 brands between March and May 2026. On a brand's first run, household names came back in 73% of the answers where they belonged, established mid sized and regional brands in 44%, and niche or small brands in 11% (Source: Kumar, 2026). That one was published by a company that sells visibility tracking, which is worth knowing while reading it.
So the thing being sorted on, every single time, was how well known a business already was. Age often travels alongside that, and sometimes it does not, and nobody has pulled the two apart.
Is being new the same as being unknown?
No, and the gap between those two is the useful part. A business can be twenty years old and known to nobody outside its own street. Another can be three months old and written about all over the web. The second one is easier for an AI app to find, and its age had nothing to do with that.
There is one place where being brand new genuinely costs something. Part of what an AI app answers from is material it learned while it was being built, and that material stops at a date. Anthropic, the company behind the Claude app, publishes that date for its current models: June 2026, listed as the training data cutoff and as the date through which the model's knowledge is "most extensive and reliable" (Source: Anthropic, 2026). A business that did not exist then cannot be in that part of the answer. Anthropic is one company of four here, and the makers of ChatGPT, Perplexity, and Gemini each set their own dates.
So a very new business leans on the other half, which is the live look at the web, and that half is not always switched on. Semrush, a search data company, read more than a billion lines of US browsing records collected between October 2024 and February 2026. By February 2026, ChatGPT was turning its web search on for 34.5% of questions, down from 46% in late 2024 (Source: Semrush, 2026). The rest of the time the answer came from what it already held.
Put those two together and the reading is the same either way. Being written about in other places, by other people, is doing work that a tidy website of your own cannot do on its own.
What went with getting named, if not age?
Four plain things, and three of them a new business can start on today. In the study of 4,776 cafes, restaurants, and bars, having your own website came with about 1.92 times the odds of being named, close to double. Showing prices came with about 1.54 times. Being written about on other websites, about 1.44 times. Having plenty of ratings came with about 1.64 times, and that is the one that takes customers (Source: Pitenin, 2026).
Two things did nothing at all. A high star rating did not help a place get into an answer. Being listed in a big public places database did not help either, which is worth knowing, because that idea gets sold (Source: Pitenin, 2026).
The headline of that study is the part a worried new owner should sit with. Across 2,208 answers from ChatGPT, Perplexity, Gemini, and Claude, 85.6% of the 4,776 places were never named by any of them. Among the long established places, the ones carrying fifty or more ratings, 72.6% were never named either (Source: Pitenin, 2026). Years of trading protected nobody.
Its limits are real and worth stating. It covered food and drink in two towns in Bali, over seven days, using 96 questions written as a real traveller might ask them. Those four findings are things that went together, not proof that one causes the other. What makes it worth reading anyway is where it started: a complete list of every cafe, restaurant, and bar in those towns, rather than a sample of the ones already doing well.
Can a business nobody has heard of get into an answer?
Yes, when the question is narrow enough. "Who can fix a cracked phone screen tonight" has far fewer honest answers than "best phone repair shop". One study put questions to six AI models across five kinds of product. Broad questions left plenty of known brands out. Questions that spelled out what the buyer needed brought back a different set (Source: Malthouse, Lee, Yang, Pal, and Feng, 2026).
The paper's own wording for the opening that leaves is that brands "omitted from ordinary recommendations can remain conditionally retrievable when distinctive cues are supplied". In plain words: a business missing from the general answer can still turn up when the question says exactly what somebody wants. The word doing the work there is "conditionally". The cue has to be there to be found, which means the specific thing a business does has to be written down somewhere findable.
The same study found that how prominently a brand turned up went more with how much it was being searched for and talked about online than with how big a name it was. That is the opening a new business has. It is a narrow one rather than a shortcut.
Why do well known names keep coming back?
Because an AI app leans on names it already knows, and that pull is strong. In one test every product carried the same written specifications, and the well known brands took 100% of the picks. A rating edge of less than a tenth of a star was enough to break that hold (Source: Chu and Hou, 2026).
Both halves of that matter, and the second half is the one that gets left out. The pull towards familiar names is real. The gap that broke it was tiny.
Read it as a direction rather than a measurement of your own odds. The test used skincare products, three AI models, and a setup the authors built themselves rather than the apps a customer types into.
What can nobody promise a new business?
A date. Nobody can give one, and anybody who does is making it up. No study on this page recorded how long a business had been open, so none of them can say how long a new one waits. One review read 45 papers on getting picked by AI and found no trick that keeps working over time and across different apps (Source: Martinez, 2026).
That review was published in July 2026 by Olivier Martinez, who read the research from 2023 onward. Its conclusion is that no reviewed technique shows "a stable, longitudinal, cross-platform causal effect on organic discoverability" (Source: Martinez, 2026). In plain words, nothing in that body of work has been shown to keep working over time and across different AI apps.
That is a finding about proof, not a reason to sit still. The four things in the section above are the ones that went together with getting named in a real count, which is a weaker claim than caused, and it is also the best evidence anybody has.
It cuts both ways too, and that is better news for a new business than for an old one. Nobody can tell a new owner to wait six months, because nobody has measured it. Nobody can tell a twenty year old business it is safe either.
What should a new business do first?
Get findable, then count how often you are named. Put the words a customer would really use on your own website, show your prices, and get mentioned by name somewhere that is not your own site, such as a local news page, a supplier's site, or a trade group list. Those three all went with getting named in the study of 4,776 places (Source: Pitenin, 2026).
Then write down the questions a customer would type, such as who can fix a cracked phone screen today, and keep that list the same every month. Put each question to ChatGPT, Perplexity, Gemini, and Claude at least three times, because one reply settles nothing. Count how many replies name the business, app by app, and write the date beside each count.
A count of zero is information, not a verdict. It is where almost every new business starts, and it is where 85.6% of the places in that Bali study sat as well, years of trading or not. The job then is getting named at all, which the related answers listed with this page cover.
It is worth the trouble because customers really do ask this way now. In a survey of 1,002 US adults, 45% said they had used AI to find a local business in the past year, up from 6% a year earlier in the same survey (Source: BrightLocal, 2026).
Where does that leave a brand new business?
Better placed than the worry suggests, with work to do rather than time to wait. Age was not the thing the research measured, and in the one study that began from a complete list of every business in its category, being long established protected nobody. What counted was being easy to find and easy to check. That is the work AI Says Who, the service behind this page, does.
We research the questions customers really use, freeze that list so later months stay comparable, put them to ChatGPT, Perplexity, Gemini, and Claude several times each, and write the date beside every count. No single blended score, no position in a list, and no claim that any of it is owed to how old a business is. Gemini is asked with no live web search, so it reports what it already knows about a business and never shows a link. The first check is free, and the button below starts it.
Free. No card, no call.