What Do AI Tools Say About Your Products?

Buyers can now ask an AI tool which product fits their application, how two parts compare, or who sells a replacement, and get an answer without ever seeing your website. Most companies have no idea what those answers say.

Software vendors have noticed. In June, Sitecore acquired Scrunch, a platform built to show brands where they appear, are missing or are misrepresented in AI-driven discovery. Dedicated tools can be useful, but you don't need one to get started. You can get a first look in an afternoon with a spreadsheet.

Step 1: Write Your Questions

Write 15 to 20 questions the way a real buyer would ask them, not the way your marketing team would. Cover five types:

• Category: "Who makes the best [product type] for [application]?"
• Specification: "What is the [rating, dimension, material] of [your part number]?"
• Comparison: "How does [your product] compare to [competitor's product]?"
• Replacement: "What can I use to replace [discontinued or competitor part]?"
• Buying: "Where can I buy [your product]?"

Include your actual part numbers and a few of your hardest products. Those are where a wrong answer costs the most.

Step 2: Ask Several Tools, Several Times

Run each question through the AI tools your customers are most likely to use. If you aren't sure which ones those are, ask your sales team and a few customers. Run each question more than once, ideally on different days, because the same question can produce different answers. You're looking for patterns, not a ranking.

Step 3: Log What You Find

Keep it simple. For each answer, record:

• The question and the tool
• Whether you were mentioned at all
• Whether the facts were right, checked against your own data
• Which sources it cited, if any
• Who it recommended instead

Step 4: Sort the Problems

Most of what you find will fall into one of these groups, and each has a different fix:

• You're missing. You sell the right product but don't come up. Check whether your application and category content actually says what you do, in the words buyers use.
• The facts are wrong. Look at where the answer came from. It may be an old PDF, a distributor listing, or your own site contradicting itself.
• The facts are outdated. Discontinued products and old specs can live on across the web. Replacement paths need to be published clearly.
• Someone else is the source. A distributor or competitor is explaining your product better than you do.
• It can't find specifics. The information exists, but it may be locked in a PDF or behind a login, or be missing altogether.

Step 5: Fix the Sources

You can't edit an AI tool's answer, but you can improve what it has to work with, which is good for your human buyers anyway:

• Publish key specs as text on product pages, not only inside PDFs and images.
• Use one version of the facts. Your site, your data sheets and your distributors' listings should agree.
• Fill in the blanks. Missing attributes can't be found by anyone, person or machine.
• Write for real questions. Application guides, comparisons and replacement information answer what buyers actually ask.
• Keep it clean over time. A one-time cleanup drifts. Put rules and ownership around the data that matters most.

Two Commergenix modules help with the data side. Product Sources finds missing values in your suppliers' documents and hands them to your team to approve, and Data Auditor checks your catalog against your rules on a schedule.

Step 6: Do It Again

Rerun the same questions every quarter and compare. These fixes won't guarantee what any AI tool says, and it can take time for changes to show up. But you'll have a baseline, a list of specific problems and a way to tell whether things are improving.

Closing Perspective

A lot of the talk about AI and commerce is about the future. This part is already here: someone may be asking about your product today, and the answer is built from whatever information about it is out there.

It's worth an afternoon to find out what they're hearing.