We've been going 15 years. Why does ChatGPT recommend newer firms?
The short answer
Usually because AI can only repeat what's written down in the places it reads, and a reputation built on referrals and word of mouth leaves very little in writing. Newer firms often publish plain pages that answer the exact questions customers ask, with prices, and get onto the lists and directories AI reads. You can close the gap by turning what your firm has learned over 15 years into pages, reviews and listings that describe the problems you fix in your customers' words.
Key takeaways
- Your years in business and your reputation only count with AI to the extent they're written down where it reads: your site, search results, reviews, directories and articles.
- Referrals happen in conversations, meetings and emails, which AI never sees, so a firm that has lived on them can be almost invisible in writing.
- Newer firms often win on writing: pages that answer customer questions directly, with prices, and listings on the round-ups and directories AI cites.
- Older sites tend to describe the firm by its service name and history, and old names, addresses and services elsewhere on the web can muddle AI's picture of you.
- Your experience is the advantage a newer firm can't copy. Write down the problems you've seen and how you handle them, and get reviews and contacts to describe you in the same words.
1. Why 15 years of reputation doesn't reach AI on its own
It's frustrating to be the firm everyone in your field knows, with years of happy clients, and then watch ChatGPT suggest a business that started the year before last. It helps to know how the answer gets made.
When someone asks ChatGPT, Google AI Mode or Perplexity which firm can help with a problem, the app usually searches the web, reads what comes back and writes its answer from those pages. OpenAI's help page on searching the web says that "ChatGPT search typically rewrites your query into one or more targeted queries that it sends those providers", meaning the search providers it works with. Google says AI Overviews and AI Mode may use a technique of "issuing multiple related searches across subtopics and data sources", and that to be shown as a supporting link "a page must be indexed and eligible to be shown in Google Search with a snippet".
So the answer is built from pages. How long you've traded and how much your clients like you only come into it where a page the app finds says so. Neither OpenAI nor Google publishes exactly how it chooses which firms to name, so treat the reasons below as the likely ones, not fixed rules. Some will apply to you and some won't, which is why each comes with a way to tell. If you suspect something more basic, such as AI not being able to read your site at all, our guide on why your business isn't showing up on ChatGPT covers the general causes.
2. Your reputation lives in conversations, not on pages
Fifteen years of referrals means a great many good conversations: introductions by email, a word at an event, a client telling a colleague over lunch. Almost none of that is public text. A newer firm without a referral network has to win work online from the start, so it writes things down because it has to.
How to tell: search your firm's name on Google and Bing, logged out. Count the results that aren't your own site or social accounts and that say what you actually do. Then ask ChatGPT "What is [your firm's name]?" and look at the sources it cites. If it cites only your homepage, or says it doesn't know much about you, AI has very little to go on.
What to do: get your experience into writing, on your own site and on other sites. Section 8 goes through how.
3. Your reviews are in one place, or don't mention the problem
Long-standing firms often have their reviews in one place, perhaps a Google Business Profile, a testimonials page or a trade body's site, or have few at all because referred clients were never asked. The reviews that do exist tend to praise the people ("brilliant service, highly recommend") without saying what the problem was. When someone describes a problem to AI, it seems likely that a review saying what the client was struggling with, in their words, is easier to match to that question than general praise.
How to tell: read your last twenty reviews and count how many say what the client came to you with. Then ask Perplexity, which shows its sources clearly, the question a customer would ask, and see which review sites it cites for the firms it names. If those aren't where your reviews are, that's a gap.
What to do: keep the reviews you have. When you ask your next clients for one, ask them to say what the situation was before they came to you and what changed, and point some of them to the review sites that came up in your check. Let them use their own words, and don't offer anything in return.
4. Newer firms answer the question; older sites describe the firm
Many older websites were built around Home, About, Services and Contact, and open with something like "Established in 2011, we provide a full range of..." followed by the name of the service. That makes sense to someone who already knows what they need. It doesn't match what a customer types when they only know their problem.
Newer firms are more likely to have pages headed with the questions customers ask, such as what something costs, how long it takes or what happens if you leave it, with a plain answer in the first paragraph and a price or range. When AI searches for the customer's question, those pages are the ones that match. Our guide on how customers describe the problem rather than your service explains why this matters so much for specialist firms.
How to tell: write down five questions people asked in first calls this year. For each one, is there a page on your site whose first paragraph answers it? Do you publish any price, "from" price or range anywhere?
What to do: write one page for each of the main problems you solve, starting with the question in your customers' words. Section 8 covers what to put on them.
5. Newer firms are on the lists AI reads
When you ask AI who to use for something, look at the sources it cites. They're often round-up articles ("the best firms for..."), comparison pages, directories and review platforms. A newer firm trying to get noticed will often have made a point of getting onto those. A firm that has been busy with referrals for years may never have needed to.
How to tell: ask ChatGPT and Perplexity the "who can help with..." question your customers ask, open each cited source, and note which ones name other firms and not you.
What to do: work through those sources and get onto the genuine, relevant ones: your trade bodies' member directories, the directories for your field, and round-ups that list firms on merit, where you can ask to be considered and send the facts they need. Check whether a list only includes firms that pay, and decide whether that's worth it to you.
6. Old details are still out there
Over 15 years most firms change something: a name, an address, a phone number, a partner who has left, a service you stopped offering, an old website address. Directories copy details from each other and rarely update them, so the old versions can stay online for years. When AI finds different descriptions of the same firm, it may use an out-of-date one or fail to connect them. The general fix for mismatched details is in our guide on why your business isn't showing up on ChatGPT; the long-standing firm's version is that there are usually more old details to find.
How to tell: ask ChatGPT, Gemini and Perplexity what your firm does, where it is and who runs it. Then search your old name, old address and old phone number and see what comes up.
What to do: list every name, address, phone number and website you've used, then update or remove the listings that still show them. If you changed your name, saying so on your about page ("formerly...") can help AI and people connect the two, and an old website address should redirect to the current one.
7. Your pages look out of date
A copyright line from years ago, a news page last updated before the pandemic, prices that are no longer right: these tell a reader the site isn't looked after. OpenAI's own advice to ChatGPT users is to "review when it was published or updated" before relying on a source. Neither OpenAI nor Google says how much a page's age counts when AI chooses what to cite, but for questions about prices or current options it seems likely that a recently updated page is the safer one to repeat, for AI and for the person reading.
How to tell: look at the dates on your main pages, your blog or news page and your prices. Would a stranger think the firm is still active?
What to do: review your main pages properly, update prices and details, and show a "last updated" date where you've genuinely checked a page. Changing the date without changing anything helps nobody, and readers notice.
8. Turn 15 years of experience into things AI can read
A newer firm can write tidy pages, but it can't write about fifteen years of seeing the same problems in different forms. That experience is the material for everything below, and most of it is already in your head and your files.
- Problem pages built on what you've learned. For each problem you solve, describe the situation clients are in when they call you, the fixes they've usually tried first, what you check, the options and how long it tends to take. Write it the way you'd explain it in a first meeting.
- Prices or ranges. You know what work costs better than anyone. A "from" price, a typical range and what makes it go up or down answers one of the first things customers ask AI, and a page that answers it gives AI something to repeat.
- How you've handled cases, without the names. If your work is confidential, describe the kind of situation, what you did and what changed, leaving out anything that identifies the client. Ask the client to read it first if there's any doubt.
- Your history as plain facts. The year you started, the kinds of problems you've worked on, the qualifications and memberships you hold. Put them on your about page in sentences AI can lift, not in a timeline graphic.
- Reviews that mention the problem. As in section 3: ask clients to say what they came with and what changed, in their own words.
- The listings that matter. Long-standing firms are often members of trade bodies whose directories they've never filled in properly. Complete those entries with the same description you use on your site, then add the directories that came up in your checks.
- Mentions from people who already know you. Partners who refer you might list you on a page about the firms they work with. Associations, events you've spoken at and publications you've written for may have pages about you that are thin or out of date. Ask them to describe what you fix in plain words, where it makes sense for their readers.
Our guide on how to get recommended by ChatGPT goes through writing these pages and building outside mentions step by step.
9. What to do first
You don't need to do all of this at once. This order gets the quick, sure fixes done before the slower work:
- Check what AI says about you by name. Fix any old names, addresses or services at the source, on your site and in the listings that still show them.
- Ask the questions your customers ask, logged out, on ChatGPT, Google AI Mode and Perplexity, and write down which firms come up and which sources are cited.
- Write one page about the problem behind your most valuable enquiries, with the answer first, a price or range and the area you cover.
- Ask your next few happy clients for a review that says what they came to you with.
- Complete your trade body and directory listings, starting with the ones that appeared in the sources.
- Then build out: more problem pages, anonymised cases and mentions from long-standing contacts.
If you're weighing up doing this in-house or getting help, see whether to do it yourself or hire an agency and how much it costs to get recommended by AI. Whoever you talk to, be wary of anyone who promises a place in ChatGPT's answers; our guide Is AI SEO a scam? covers the warning signs.
10. How to check it's working
Keep a short list of the questions you asked in step 2 and ask them again once a month, the same way: logged out or in a private window, on the same apps, and twice each, because answers change from one go to the next. Note whether you're named, how you're described next to the newer firms and which sources are cited. Alongside that, ask every new enquiry how they found you, with "asked ChatGPT or another AI app" as one of the options.
Changes take time to show up. Google says that crawling a page again "can take anywhere from several days to several months", and other apps have their own timings. In practice, corrections to how AI describes you by name tend to come first, then your pages being cited for the problem questions, and being recommended by name after that, often a few months in. If you work with us, we agree how enquiries are counted and a starting level at the beginning, and if your enquiries haven't gone up by the end of month three, month four is free. The details are on our pricing page.
Sources
- OpenAI Help Center, Searching the web with ChatGPT, read 16 November 2026.
- Google Search Central, AI features and your website, last updated 10 December 2025, read 16 November 2026.
Questions owners ask
Does being established for 15 years count for anything with AI?
Should we redesign our website so it looks newer?
Our reviews are all on one site. Should we start again somewhere else?
Our work is confidential. Can we still write about it?
Should we ask our long-standing contacts to mention us online?
How long before AI starts recommending us instead of newer firms?
Want to know why newer firms come up instead of you?
We'll ask ChatGPT, Google AI Mode, Gemini, Claude and Perplexity the questions your customers ask, and what they know about your firm by name, and show you what comes back with dated screenshots, the sources they read and the first things we'd change. It's free, takes 48 hours, and involves no call or pitch. If you want more, a one-off audit and plan is £399 and monthly work starts from £799, rolling monthly. See pricing.
Get your free check