How to get your business recommended by ChatGPT

A five-step method: check where you stand, find the sources ChatGPT reads, fix your site, work on reviews and comparisons, then keep tracking.

Christophe Tran Founder of iPand Oct 23, 2026 · 6 min

How does ChatGPT decide to recommend a business?

ChatGPT recommends the businesses it finds in the pages it reads while answering, and that match the question closely. For anything current or purchase-related, it searches the web before writing.

OpenAI explains that ChatGPT search leverages third-party search providers and content provided by its partners, and that any website or publisher can choose to appear in it^[1]. Recommendations are not for sale: OpenAI, which announced ad tests in the US in January 2026, states that ads do not influence answers and are always kept separate^[2].

A recommended business is therefore one that shows up, consistently, in the sources a search engine returns for the question: its own site, but also comparisons, articles, reviews and discussions. The method below works through those sources in order.

Step	What you do	Where	When
Check where you stand	Ask your customers’ real questions	ChatGPT and other assistants	Once, then at every review
Find the sources	List the sites cited in answers	ChatGPT’s Sources button	At every check
Fix your site	Open access, answer the questions	Your website and robots.txt	Right after the first check
Work on reviews and comparisons	Get onto the pages that get cited	Review sites, media, comparisons	Ongoing
Track over time	Re-run the same questions	A tracking tool, or a spreadsheet	Weekly or monthly

Step 1: check where you stand

Ask ChatGPT the questions your customers ask before they buy, and note whether you are cited, where, and next to whom. That is your starting point, and the only honest baseline for measuring progress.

Start from real questions: the ones your sales team, your support desk and your reviews already hear. Mix intents: “which invoicing software for contractors?”, “compare X and Y”, “cheaper alternative to Y”. Take a fictional invoicing software company: it might find that it gets named on comparison questions but never on discovery questions, the ones where the buyer does not know any name yet.

Do not trust a single answer. SparkToro showed, across nearly three thousand answers, that the list of recommended brands changes almost every time the same question is asked again^[3]. Ask each question several times, or use a tool that does: the free test gives you a first snapshot.

Step 2: find the sources ChatGPT reads

Under each answer, open the Sources button: it lists the pages ChatGPT read^[1]. Those pages are your real target, more than your own website.

Group them by type: comparisons and “best of” lists, specialist media, forums and discussions, review platforms, competitors’ sites. A page that comes back across several questions weighs more than one cited once. If you are not on it, that is often where your absence comes from.

Google shows the same pattern: in Pew Research Center’s study of AI summaries, Wikipedia, YouTube and Reddit were the most linked sites^[4]. Third-party platforms you do not control matter as much as your own site.

The sources behind an answer

Comparisons and rankings
Specialist media
Forums and discussions
Review platforms
Competitors’ websites

A page that comes up on several questions weighs more than one cited once.

Group the cited pages by type to see where your absence is decided.

Step 3: fix your website

Your site needs to be readable by assistants’ crawlers and to answer your customers’ questions clearly. Start with access, then content.

Let OAI-SearchBot in: a site that opts out of it is not shown in ChatGPT search answers^[5]. Our article on robots.txt explains the setting.
Write one page per important question, answering it in the first sentence: who your offer is for, what it costs, and how it differs.
Publish your prices, or at least a range: it is a question your customers ask, and an assistant can only answer with what it finds.
Add structured data that matches the visible text, as Google asks for its own AI features^[6].

Also explain clearly who you are. An About page that gives your exact name, what you do, where you are based, who runs the company and links to your official profiles helps a search engine tie every mention of your brand to a single entity. Without it, two companies with similar names can be mixed up, and the recommendation can go to the other one. Keep that name identical everywhere, from your footer to directories.

Step 4: work on reviews and comparisons

Get known by the sources you identified: comparisons, media, review platforms and communities. It is the longest step, and the one that moves answers the most.

Offer your product to the authors of comparisons that keep showing up in answers, with verifiable information: pricing, features, customers. Encourage customers to leave reviews on your industry’s platforms, with no incentive. Take part in discussions about your field, and say who you are when you do.

For Wikipedia, the rule is clear: editors with a conflict of interest are strongly discouraged from editing affected articles directly and can propose changes on the talk page instead, and anyone editing for pay must disclose it^[7]. A page written by the company itself is unlikely to last.

Step 5: track over time

Re-run the same questions at regular intervals and look at the trend rather than a single answer. It is the only way to know whether your actions moved anything.

Track at least three things: the share of answers that cite you, the competitors cited instead of you, and the sources that keep coming up. Write down the date of each action, such as a new page, a new review or a press article, to link a change to its cause. Our article on six metrics explains what to read, and our comparison of tracking tools covers the options on the market, iPand included.

What does not work

Paying to be recommended, publishing pages in bulk or buying reviews will not get you recommended for long. At best it is useless; at worst it exposes you.

Promises of “first place in ChatGPT”: nobody controls the answer, and the order of brands changes from one run to the next^[3].
Mass-produced pages with no precise answer: they give the assistant no reason to cite you.
Fake or incentivised reviews: banned by platforms, and in the US targeted by the FTC.
Keyword stuffing: an assistant reads sentences, it does not count occurrences.

app.ipand.io/marche · fictional example

iPand

MJ Meeple Junction meeplejunction.example

SM Sarah Mitchell Meeple Junction LLC

Market

Your ranking against 6 competitors in Denver.

Market United States

Period 30 days

Type Brands

Next collection run in 4h 12m

#1 Tablewright Games

7.2% share of voice

Visibility 41% · Cited by 3 engines mentions

#2 Card Canyon

5.9% share of voice

Visibility 33% · Cited by 3 engines mentions

#3 Owlbrook Games

5.3% share of voice

Visibility 30% · Cited by 2 engines mentions

RankPlayerVisibilityShare of voiceCited byRelationship

#1 Tablewright Games 41% 7.2% 3 engines Competitor

#2 Card Canyon 33% 5.9% 3 engines Competitor

#3 Owlbrook Games 30% 5.3% 2 engines Other

#4 Mesa Peak Games 24% 4.4% 2 engines Competitor

#5 The Rolling Mug 21% 3.8% 2 engines Competitor

#6 Meeple Junction 18% 3.3% 2 engines My brand

Excerpt from the Market screen · fictional example

In the app, “Google AI Overviews” shows as “AI Overviews” when space is short.

Key takeaways

ChatGPT recommends the businesses it finds, consistently, in the pages it reads to answer.
Start from your real questions and from the sources cited in answers: those are what you need to work on.
Open your site to OAI-SearchBot and answer your customers’ questions clearly, prices included.
Recommendations cannot be bought: measure over time, and be wary of promised positions.

Ads do not influence the answers ChatGPT gives you.

— OpenAI, “Our approach to advertising and expanding access to ChatGPT”

Frequently asked questions

Can I pay to be recommended by ChatGPT?

No. OpenAI is testing ads, shown separately and clearly labelled, and states that they do not influence answers. A recommendation inside the answer itself cannot be bought.

How long does it take to get recommended?

Nobody can promise that. A robots.txt change is picked up in about a day, according to OpenAI; a presence in comparisons and reviews builds over weeks or months. Measure before and after on the same questions.

Does ChatGPT rely on Bing?

OpenAI refers to third-party search providers without naming them in its announcement. Being well indexed by Bing as well as Google is a sensible precaution.

Do I need a Wikipedia page?

It is not required. If your company qualifies, let independent editors write the article: Wikipedia strongly discourages editing an article about yourself directly.