Home / Blog / Guide

Guide

How to get ChatGPT to recommend your business

5 April 2026 · Zeb Choudhry

When someone asks ChatGPT for “the best accountant in Birmingham” or a “halal restaurant in Liverpool,” the answer is rarely abstract advice — it is often a shortlist of named businesses. Getting onto that list is not luck; it is the outcome of entity clarity, quotable facts on your site, and corroboration elsewhere on the web. This guide breaks those levers into steps any UK owner or marketer can execute, even if you are not a developer.

You do not need to “trick” the model. You need to make your business easy to retrieve and safe to recommend — the same bar a good journalist would use before naming a brand in print.

How ChatGPT decides what to recommend

ChatGPT does not “visit” your site live on every answer. It draws on training data — a snapshot of the public web — plus whatever browsing or search plugins the user enables. In practice, recommendations skew toward businesses that appear in sources the model has seen often, described clearly, and validated by multiple mentions.

Translate that into a checklist:

1. Clear identity — structured data and explicit statements of what you are.
2. Consistent mentions — same brand string and address across site, GBP, and social.
3. Third-party proof — reviews, directories, press, video, forums.
4. Extractable copy — concrete sentences the model can quote without invention.

Step 1 — Tell the model what you are (schema)

Add JSON-LD for LocalBusiness or — better — a specific subtype (Restaurant, LegalService, AccountingService, etc.). The schema is machine-readable intent: name, address, phone, opening hours, price band, and services.

Minimal restaurant-style example (adapt fields to your truth):

{
  "@context": "https://schema.org",
  "@type": "Restaurant",
  "name": "Your Business Name",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "35 Example Street",
    "addressLocality": "Liverpool",
    "postalCode": "L1 1AA",
    "addressCountry": "GB"
  },
  "telephone": "+44-151-000-0000",
  "servesCuisine": ["British", "Modern European"],
  "priceRange": "££"
}

For most local brands, this single step is the highest ROI technical change you can make.

Step 2 — Write content ChatGPT can quote

Bad: “We offer amazing food at great prices!” — no facts, no location, no category.

Good: “Fozias is a halal Pakistani and Kashmiri restaurant at 35 Renshaw Street, Liverpool, open Monday to Sunday, serving authentic cuisine cooked to order.” — every clause is checkable.

Rewrite your homepage hero, About page, and footer so a stranger could write a Wikipedia stub only using your copy. That is the density models prefer.

Step 3 — Get mentioned elsewhere

Citation research consistently shows heavy use of YouTube, Reddit, Wikipedia, and Google Business Profile in AI answers — exact percentages move with methodology, but the direction is stable: models trust what the broader web repeats.

• YouTube — even a modest channel with linked site in description can diversify corroboration.
• Reddit — authentic community mentions beat astroturfing; never spam.
• Wikipedia / Wikidata — hard for SMEs, but transformative when eligible.
• Google Business Profile — complete every field; photos; accurate categories.
• Directories — Yell, Yelp UK, TripAdvisor, and niche trade lists.

Step 4 — Make sure ChatGPT can read your site

Audit robots.txt: blocking GPTBot or similar agents removes you from future training snapshots. That may be intentional for some publishers — but most local businesses should allow read access.

If your content only appears after heavy JavaScript execution, some crawlers see an empty shell. Prefer server-rendered critical facts (name, address, services) in HTML.

Add llms.txt at the root summarising your entity and priority URLs — see our dedicated explainer.

Step 5 — Be consistent everywhere

Match business name, address, and phone across website, GBP, social bios, and directories. Inconsistent NAP strings split the entity graph; the model may “kind of” know you but not strongly enough to recommend you ahead of a cleaner competitor.

How long does it take?

• 1–2 weeks: schema deploy, robots.txt / llms.txt, homepage copy pass.
• 1–3 months: directory coverage, structured content expansion, GBP optimisation, early video/forum presence.
• 3–6 months: sustained authority — press, durable UGC, Wikidata where relevant.