How to Improve Your Brand Sentiment in ChatGPT and AI Search

Getting mentioned by ChatGPT is only half the job. The other half is what it says about you. When someone asks for the best running shoes, does ChatGPT recommend your brand, or does it warn people that you are overpriced? That is brand sentiment in AI search: whether an AI assistant talks about your brand positively or negatively. This guide walks through a step-by-step process to analyze it and improve it, using Nike as an example.

What positive and negative AI sentiment looks like

Take a simple prompt: "Give me a list of running shoes to run a marathon under $100." In ChatGPT's answer, two brands get very different treatment:

• Adidas Adizero Evo SL: described as "my first choice if you are looking to run a faster marathon" and labeled best overall. That is a very positive sentiment.
• Nike Vomero 18: "appealing for runners who prioritize comfort over race day speed" and "best for slower marathon paces". That is mixed, somewhere between positive and negative.

Both brands are mentioned. Only one is truly recommended. If you only track mentions, both look like wins.

Why brand mentions alone are not enough

Being mentioned in a prompt is the top of the iceberg. What matters is whether the AI pushes your brand or pushes a competitor. A mention that says "avoid this brand" or "too expensive" can cost you the sale. Analyzing sentiment manually in ChatGPT does not scale. You need a GEO tool that runs your prompts repeatedly and scores how each engine talks about you. In Qwairy , the sentiment view shows the global sentiment for your brand, per engine, and the prompts where you are mentioned negatively.

Step 1: find the prompts where you are mentioned negatively

Start with one engine, for example ChatGPT on Qwairy. Filter your tracked prompts by sentiment and isolate the answers where your brand appears in a negative light. Overall sentiment can be very positive while a handful of prompts drag it down. Those are your priorities.

Step 2: understand why the AI says it

On Nike's data, one ChatGPT answer stands out. The prompt is a buying guide on affordable running clothes, and the answer says:

Choosing the right athletic apparel for running doesn't mean spending a fortune on expensive brands like Nike.

The answer even suggests looking at clearance sections to buy Nike at a discount. Nike is mentioned, but as the expensive option. The next move is to open the sources behind the answer and see who is used as a reference:

• A page on budget-friendly sportswear by Decathlon.
• A running guide from REI.

These pages shape the narrative. They tell you exactly where the "expensive" perception comes from.

Step 3: fix sentiment with content and third-party sources

There are two levers.

Create your own content on the topic. If AI says Nike is expensive while Nike does sell affordable products, Nike should publish a dedicated page on budget-friendly running gear. Its brand promise is to have a product for every runner, so the content exists to be written.

Be present on the sources AI uses. Check whether you appear on the third-party pages cited in the answer, and work on being featured on them or on similar sources.

Step 4: repeat on Google AI Overviews and AI Mode (or any other LLM)

Sentiment differs by engine, so check each one separately. On Google AI Overviews, Nike's sentiment shows no issue to fix. On AI Mode, however, Nike is listed as a top performer but associated with greenwashing allegations: it frequently lands on top-performing lists while facing scrutiny and litigation over sustainability claims. The first fix for this kind of issue is real: if the criticism is about practices, change the practices. Then look at the sources. One cited source is a ranking of the world's most sustainable companies of 2026. If a brand wants to be recognized as sustainable, being present on that kind of ranking is a concrete target.

Step 5: fact-check what AI says about your brand

Sentiment is not the whole picture. AI assistants also state facts about your products, and some are wrong. Qwairy's fact-checking feature extracts the claims made about your brand and groups them by category. Examples from Nike's data:

• Price: a Vomero model described as costing $180, another model retailing around $80.
• Performance: "Air Jordan 40 offers 85% energy return".

For each claim, you can see which prompt it came from, then check the sources to confirm the right price or the right figure is there.

Step 6: correct misinformation at the source

When a claim is wrong, you have two options:

• Contact the source that published the inaccurate information and ask for a correction.
• Publish the right information on new sources so that AI assistants start using them instead.

Make every team speak the same way

The most important lesson: when you launch a product, every part of the company must describe it consistently. Press relations, marketing, sales, partners: everything they publish can surface in AI answers. Nissan applies exactly this approach, with a shared product reference used by all teams. Read How Nissan Turned AI Search into a Reputation Economy with Qwairy .

Want to know how AI assistants really talk about your brand? Try Qwairy for free  with 120 credits to track sentiment, sources and claims across ChatGPT, Google AI Overviews, AI Mode, Claude and more.