Methodology
A transparent explanation of what is curated, what comes from members, and how products earn a place in a recommendation.
Last updated September 28, 2026
Each catalog product has one canonical record, optional aliases, source URLs, and one or more functional categories. The primary category describes what the product mainly does. An AI capability or target audience does not override a stronger operational function.
The current engine uses four deterministic signals: reviewed alternative or complementary relationships, vendor-documented integrations, products observed together in active public member stacks, and products adopted by multiple members in the same industry. Results exclude products already active in the stack and explain their strongest qualifying signal.
Most used counts distinct members with a product active in a public member-created stack, so one person with several stacks contributes only once per product. Highest rated uses the 1–5 product rating and a Bayesian confidence adjustment toward the catalog-wide average. Most recommended uses only ratings where a member answered the recommendation question. Rating and recommendation lists require at least two eligible responses. Industry and profession lists count distinct members in each segment.
Archived users, stacks, stack entries, and software are excluded. MapMyApps Team examples and system accounts are excluded. Affiliate relationships never affect rank. Counts and averages are shown beside results so visitors can judge the strength of the evidence.
Stacks published by MapMyApps Team provide useful launch content. They are excluded from community ratings, adoption counts, co-occurrence, popularity, reactions, comments, notifications, and recommendation-learning signals. Community labels refer only to eligible member-created data.
Official vendor pricing is stored separately from member-reported prices and includes source and verification dates where available. Prices can change and may depend on seats, usage, region, taxes, or promotions. Public stack totals include only current prices a member chose to share and can normalize only supported monthly or annual amounts in the same currency.
When a product is missing, the importer safely reads its official public website, extracts structured candidate details, and may use an AI model for ambiguous fields. New records and taxonomy suggestions require member confirmation. The ordinary recommendation browser does not make AI requests.
Approved affiliate metadata is stored separately from catalog and ranking data. An eligible official-website link carries a disclosure and sponsored link attribute. Affiliate availability never changes rankings or recommendation scores. Member offers remain a planned benefit until a verified arrangement is active.