Zapier Keeps Integration Usage Data

Diving deeper into

Zapier: The $7B Netflix of Productivity

Document
while Zapier collected usage data from its third-party partners, it did not share any of that data with its partners.
Analyzed 6 sources

This shows the real fault line in integrations, who owns the user relationship also owns the learning loop. Zapier could see which partner apps were actually used together and in what patterns, but partners could not use that data to decide which native integrations to build next. That mattered because the first 10 to 15 integrations usually capture most real demand, and those are the ones that make a product feel complete inside its own UI.

  • Former partners describe a one way relationship. They were asked to build and market their Zapier connection, but were not given hard usage stats back. That meant Zapier kept the map of demand while partners supplied the endpoints and customer acquisition surface.
  • For a SaaS company, this data is product planning gold. It tells the team whether users mostly want Slack alerts, Mailchimp sync, Salesforce updates, or something else. Without it, the company has to guess which integrations should be built natively and which can stay outsourced to Zapier.
  • That gap is exactly what embedded integration vendors sell against. Tray.io and Paragon make money by helping software companies ship integrations inside their own product, keep setup in one interface, and retain the behavioral data that can feed future product decisions. Tray reached about $70M ARR in 2023, while Paragon was at about $4M ARR, showing both the size and early stage of this shift.

Going forward, the winners in integrations will increasingly be the companies that treat connectivity as a core product surface, not a bolt on marketplace listing. Zapier remains valuable for long tail coverage, but the highest value workflows will keep moving into native and embedded experiences where software companies can control onboarding, reduce churn, and learn directly from usage.