How to Run an AI Visibility Check on Your Range in Furniture Connect
Set up a product sample and shopper questions, review the estimate, run the check and read the results without treating one run as proof.

If you've run the manual test in how to test whether your range is readable by AI, you know the shape of the job: shopper questions, several assistants, results per product and gaps to investigate.
Before you start
The check works from your catalog, so the products you want to track need to be in Furniture Connect as active records. A more complete record gives the report more catalog fields to assess.
It also spends credits. Each check asks the tracked questions across the assistants and regions you selected, shows an estimate before the run and asks for confirmation. Check the current estimate in the product before proceeding.
1. Set up the brand
On the AI Visibility page, confirm your brand name and website, so the check knows what counts as "you" in an answer. Add the competitors you want to compare and choose the regions you want checked. The first check also turns on weekly automatic checks; change or turn off the schedule if you'd rather run checks yourself.
2. Add shopper questions
These are the questions the check will ask, phrased the way a shopper would ask them, not the way a retailer would. "A sectional under 100 inches with a washable fabric for a small apartment", not "best modular sofa". Add the ones that matter for your range; if you've done the manual test, you already have the list, and your customer service team's top questions are the best source.
Track the questions you'll act on rather than everything you can think of. Each additional question increases the work in later checks; use the estimate shown before the run.
Use two kinds of question. Shopper questions that don't name your brand show whether your products appear without being requested. Named questions ("tell me about the Harlow sofa from [brand]") test what an assistant says when pointed at a particular product. Do not combine the two into one visibility figure; they answer different questions.
3. Track the products
Pick the products you want tracked. For a first run, use a mixed sample: best-sellers, a recent launch, some you suspect have thin records and some that compete directly with products you've seen recommended.
4. Run the check
Ask the agent to run the check. Review the credit estimate and confirm. The check then asks the tracked questions across ChatGPT, Claude, Gemini and Perplexity in the selected regions.
5. Read the report
The report covers observed brand recommendations, other brands named, linked sites, sales-channel results, tracked products, catalog gaps and individual questions. Three parts are useful to inspect first:
Who's recommended instead. This is the competitor list the assistants actually produce, which may differ from the one you entered. Open two of their pages and do the side-by-side in what the AI found on your competitor's page that it didn't find on yours.
Per-product gaps. For each tracked product, what's missing from its record: description, dimensions, materials, images and so on. This is the list that turns a score into work. Reading the AI visibility report goes through it.
The per-question breakdown. Which questions you appear for and which you don't. A question you never appear for, across several products, is worth checking against your record for a shared missing attribute. It can also reflect retrieval, relevance, availability, third-party coverage or ordinary variation between runs, so the report shows gaps to investigate, not causes.
What one check is
A sample. The assistants vary from run to run and change over time, so the report's figures are an indication of what happened on that run, not a precise measurement. Use it the way you'd use the manual test: as a baseline, then re-run after you've closed gaps and compare the direction.
What it doesn't do
It doesn't change your pages or your record. It tells you what's missing; closing the gaps is a separate step, through Autofill for fields and Studio for imagery, covered in closing the gaps. It doesn't tell you how many shoppers are asking these questions. And it doesn't make an assistant recommend you; it shows you gaps in your record and channel listings that are worth closing, alongside what the assistants actually answered. The gaps are work, not proof of why a product was left out.
Start with ten
Ten products, the shopper questions you'd most want to win, one check. Read the gaps, fix the ones that affect several products at once, re-run. For what to do in the first fortnight, see your first two weeks. For the whole argument, see being found by AI.




