Guide · 7 min read · 2026-10-12
How ChatGPT chooses which software to recommend
What drives product recommendations in ChatGPT-style answers: retrieval, source consensus, category fit and recency, and what SaaS teams can influence.
Two paths to a mention
An assistant can name a product from what it learned in training, or from pages it retrieves when web search is on. Incumbents benefit from both. Newer products depend almost entirely on retrieval.
What retrieval favours
- Pages that directly answer the question: "best X for Y" lists, comparison tables, alternatives pages.
- Sources with strong reputations: G2, Capterra, major tech publications, established blogs, active community threads.
- Recency: recently updated roundups and threads tend to be retrieved for "2026" style prompts.
- Agreement: a product named in several retrieved sources is far more likely to make the final list.
Category fit decides eligibility
The assistant first decides what kind of product the buyer wants. If your site and listings describe you inconsistently, as a CRM in one place and a sales platform in another, you can be excluded before ranking even starts. Use one category phrase everywhere.
Signals you control
- Your title tag, H1, meta description and first paragraph: state the category and audience.
- Public pricing: answers often compare costs, and products without prices get skipped.
- Comparison and alternatives pages written honestly.
- Structured data (SoftwareApplication, FAQPage) and an llms.txt summary.
Signals you influence
- Reviews on the grids cited in your category.
- Inclusion in the roundups that get retrieved, through outreach to their authors.
- Helpful, disclosed participation in cited community threads.