Does ChatGPT recommend your brand, or the competition?

This article shows why the AI recommendation has become a competitive channel of its own, how to check yourself whether ChatGPT names your brand, and what to do when only your competitors appear there.

Why the question “does ChatGPT recommend my brand?” suddenly matters

For years the answer to “where does my customer do research?” was simple: Google, Instagram, perhaps a comparison site. Those channels still exist, but a new one has appeared next to them. Millions of people now phrase their buying intent as a conversation with an AI system, and accept its selection as a pre-filter.

The difference to classic search is fundamental. A Google search returns ten blue links, and the user decides. An AI answer returns a recommendation, and the user usually follows it. A list of twelve options becomes a selection of three. Whoever is not among those three effectively does not exist, no matter how well their own website ranks.

For e-commerce brands in German-speaking markets this matters twice over. First, your competitors are the same as offline, but the AI system decides faster and more definitively who gets named. Second, very few shops in the German-speaking market measure this at all. Whoever takes it seriously first gains a head start that cannot be caught up overnight.

How an AI system decides which brand to name

An AI system such as ChatGPT does not invent its recommendations. It draws them from what it has learned about a brand: from texts on the open web, from comparison articles, from forums, from product descriptions, from the way people write about a brand. Put simply: the more clearly, more often and in more fitting contexts a brand is described on the web, the more likely it appears in the answer.

That has consequences which surprise many people. A brand with a poor Google ranking can still be recommended by ChatGPT because it is named in many comparison texts and communities. The other way round, a shop at position one on Google can be missing from the AI answer because its content is optimised for search engines but hard for a language model to use: too promotional, too thin, with no clear statement of what the brand actually stands for.

Three levers follow from this, and they decide whether an AI system names you:

• Clarity. Is it stated anywhere unambiguously what your brand stands for, for which category, for which audience, with which difference to the competition? Models prefer brands with a sharp profile.
• Frequency in the right context. Is your brand mentioned in exactly the contexts customers ask about, that is, in comparisons, guides and reviews of your category?
• Consistency. Do different sources describe your brand in similar terms? Contradictory or outdated information weakens the chance of a mention.

Step by step: check whether ChatGPT recommends you

You do not have to buy a tool to get a first impression. The manual sample takes half an hour and is soberingly honest.

1. Write down the questions your customers ask

Do not start at the finish line, start at the beginning of the buying journey. Write down five to ten buying questions a prospect would ask who does not know your brand yet. Not “what do you think of [your brand]?”, but: “which shops do you recommend for [category] in Germany?” or “what are good alternatives to [well-known competitor]?”. The trick lies in the open character: you want to see who the AI system suggests unprompted.

2. Put the questions to several AI systems

Do not ask only ChatGPT, ask Perplexity as well. The systems draw on different sources and answer differently. It may well be that Perplexity names you and ChatGPT does not, or the other way round. Only the overall picture shows where you stand.

3. Note who is mentioned, and in which place

Keep a plain table: question, system asked, brands mentioned, order. Pay attention to two things in particular. First: do your competitors appear while you do not? That is the hardest but most important finding. Second: are you mentioned but described incorrectly, with an outdated range, the wrong positioning, a confused brand? That costs sales too.

4. Repeat the measurement

A single answer is a snapshot. AI systems do not answer identically every time, and their knowledge base changes. Only repetition over weeks shows whether something is moving, and whether your work has an effect.

An example from practice

Take a mid-sized direct-to-consumer shop for natural cosmetics from the German-speaking market. The owner is convinced she is clearly visible: position two on Google for her main category, solid Instagram presence, loyal regular customers. She puts the question her customers would ask to ChatGPT: “which natural cosmetics brands from Germany can you recommend?”

The result: three brands are named. Two of them are her direct competitors, one is a brand she had barely perceived as competition. Her own brand does not appear. Perplexity shows the same picture, only in a different order. The good Google position helps nothing here, because the customer who starts with ChatGPT never sees Google.

What she makes of it: she looks at why the three named brands are mentioned and finds that all three appear in several independent comparison articles and reviews which state clearly what they stand for. Her own brand is present on the web mainly through her own promotional website. That is exactly the gap she has to close.

Why does ChatGPT recommend my competitor and not me?

The short answer: because the model found more unambiguous material about your competitors than about you. An AI system does not prefer a brand on purpose, it reproduces what its sources support most safely. If a competitor is described more often and more consistently in comparison articles, directories, press coverage and forums, it is the safer recommendation for the model, regardless of whose product is better.

Three typical reasons in practice: the competitor has a clearer category assignment (the formula “X is a tool for Y” appears in identical wording in many places), third-party sources quote it more often, or it has simply been present for longer and is therefore anchored more firmly in the training material. All three points can be measured and changed. Nothing about this is fast, but the steps in the next section start exactly there. A deeper guide with seven concrete steps is in How to increase AI visibility.

What to do when your brand is not named

The reflex “then we will optimise for ChatGPT” falls short, because there is no switch to flip. What helps is systematic work on the three levers above:

• Sharpen the profile. Make sure that several places, your website, directories and comparison pages, state clearly and consistently what your brand stands for. Language models reward unambiguity.
• Appear in the right context. Aim for mentions in exactly the comparisons and guides your customers read. One honest assessment in an industry article weighs more than ten promotional posts.
• Correct false statements. When an AI system describes your brand incorrectly, find the source of the outdated information and bring it up to date.

This approach has a name: Generative Engine Optimization (GEO), the deliberate work of showing up in the answers of generative AI systems, just as SEO is the work on search engine rankings. Closely related is Answer Engine Optimization (AEO), the focus on direct answer formats. What unites the two: they can be measured, provided somebody measures them.

From a one-off test to continuous measurement

The manual sample is the right first step, and for a first picture it is enough. Its limit shows as soon as you want to run this seriously: you would have to put the same questions to the same systems week after week, document the answers, compare them with where your competitors stand and spot the changes. By hand that quickly becomes busywork that gets left undone.

This is where Rivly comes in. You enter your domain and your competitors once. Rivly puts defined buying questions to four AI systems regularly (ChatGPT, Perplexity, Claude, Gemini), evaluates whether and in which place your brand is mentioned, and sets the result against your competitors in a brand scorecard. The one-off sample becomes a development curve, and your AI visibility sits next to your advertising, your Google positions and your press coverage in one picture.

How the AI recommendation interacts with your other channels is covered in What are GEO and AEO?. If you want to know why good Google rankings alone are no longer enough, the answer is in AI visibility for online shops.

Check where your brand stands, the first results arrive in about 30 minutes, no credit card, 14 days free.