← Back to Blog

How to Find Out If AI Chatbots Like ChatGPT and Gemini Recommend Your Product to Buyers

More buyers are skipping Google and asking AI directly: "What is the best tool for X?" If your product is not in that answer, you are losing deals you never knew existed.

Here is how to check what AI says about your product and what to do if the answer is not good.

Step 1: Ask the Right Questions

Do not just search your product name. That only tells you if AI has heard of you. The questions that matter are the ones buyers ask:

  • "What is the best [your category] tool?"
  • "Compare [your product] vs [competitor]"
  • "How do I solve [problem your product solves]?"
  • "Best alternative to [competitor]"

Step 2: Check Each Model Separately

Each AI model works differently:

  • ChatGPT uses training data from months ago. If you published something last week, ChatGPT does not know about it yet.
  • Perplexity searches the live web for every query. New content can show up within days.
  • Claude uses training data and does not search the web. Similar to ChatGPT but with different training sources.
  • Gemini combines training data with Google search results.

A product can score well on Perplexity and be completely invisible on ChatGPT. Checking only one model gives you an incomplete picture.

Step 3: Look for These Problems

When you check each model, you might find:

  • Absent: Your product is not mentioned at all. AI recommends competitors instead.
  • Misclassified: AI describes your product but puts it in the wrong category.
  • Conflated: AI confuses your product with a competitor and describes you using their features.
  • Generic: AI mentions you but with no specific details about what makes you different.

Each of these problems needs a different fix.

Step 4: Fix What Is Wrong

The content that moves the needle is specific and structured:

  • For absent gaps: Publish comparison pages that explicitly name your product alongside competitors. "X is designed for [audience] while Y focuses on [different audience]" gives the model a clean frame.
  • For misclassification: Publish clear category-defining content on your docs or blog. "Bersyn is an AI visibility scanner" not "Bersyn is a marketing tool."
  • For conflation: Publish differentiation content that draws explicit boundaries between you and the competitor.

Automate the Process

Checking manually works but does not scale. AI models change their responses over time as they update training data and retrieval sources. What worked last month might not work today.

Bersyn does this systematically. It puts real buyer questions to ChatGPT, Claude, Perplexity and Gemini, several times each, and keeps every answer as evidence. A market can be measured again later, and the difference between two measurements is reported as a difference — Bersyn does not claim to know what caused it.

Bersyn does not run a self-service scan today. Measurement covers a market — one category, one country, one language — and runs for agencies and design partners: bersyn.com/pricing.

What Bersyn measures

Recorded answers, not a score

Bersyn asks the buyer questions people actually ask, across ChatGPT, Claude, Gemini and Perplexity, several times each — then reports how many of those recorded answers named each company, which other companies were named, and what the assistants cited. Every count keeps its denominator.

Bersyn is being rebuilt around verified measurement, and self-service purchases are closed while that work is in progress.

See where Bersyn is now

Get Bersyn Weekly in your inbox

Every Wednesday: 5 public scans of B2B SaaS companies, plus the patterns we see across AI Surfaces. Free, one-click unsubscribe.

We send one email per week. Unsubscribe with one click.