Teardown: five CRM brands in ChatGPT
We asked ChatGPT 40 buying questions across five CRM brands. Three took almost every mention, review sites took the citations, and no vendor published schema.
- 3 of 5
- Brands named in any answer
- 61%
- Citations pointing at review aggregators
- 0 of 5
- Vendors publishing Organization schema
The method
We wrote 40 prompts a real buyer would type — “best CRM for a 20-person sales team”, “CRM with the easiest HubSpot migration”, “compare Pipedrive and Close” — and ran each three times in ChatGPT with web browsing enabled. We logged every brand mentioned, every source cited and every claim made.
This is a snapshot, not a ranking. Models are non-deterministic. But the pattern is stable enough to be useful.
Finding one: half the category is invisible
Two of the five brands never appeared in a single answer. Both have strong classic SEO — one ranks in the top three for its primary category term. Neither has a meaningful presence on the review and comparison sites that ChatGPT retrieves for these prompts.
The models were not ignoring them. They were never retrieved in the first place.
Finding two: review sites own the citations
61% of citations pointed to review aggregators and comparison blogs. Vendor pages accounted for 14%. Reddit accounted for 11%.
This has a direct strategic implication: your own content is necessary but not sufficient. Citation earning — being present and accurately described on the sources models already trust — carries more weight than another product page.
Finding three: the facts are wrong
For three brands, at least one retrieved source described the product incorrectly: outdated pricing, a discontinued feature, or an acquisition that never happened. Because the model synthesizes rather than links, those errors were presented as consensus.
Entity hygiene is not a branding exercise. It is a retrieval correctness problem.
What we would fix first
- Reconcile the fact sheet across every major profile and directory.
- Publish a comparison page that states the trade-offs honestly — models reward balanced, specific comparisons.
- Add Organization, Product and FAQ schema, and attach real author entities to category content.
- Earn placement in the two aggregators that appear in most answers for this category.
None of this requires a new content factory. It requires being legible.