ITHACA
Merlin Bird ID, developed by Cornell University's Lab of Ornithology, has accumulated 42 million downloads since launch in 2014, with 12 million coming in 2026 alone, according to Jessie Barry, director at the lab.
The free app identifies birds by appearance, song, or call. It now covers 11,000 species—nearly the full global bird taxonomy—up from 285 common U.S. species at launch. The expansion reflects a decade of machine learning improvements and a critical moat: crowdsourced audio and image data.
The Sound ID feature alone has logged 4.4 billion identifications. That volume of verified user submissions creates a compounding data advantage, continuously retraining models to reduce misidentifications and expand species coverage.
The business model is free-to-user, funded by Cornell's endowment and grants. But the data economics are real. Each identification contributes labeled training examples; each user submission improves the underlying classifier. This is the classic network effect of a research tool—more users generate better models, which attract more users.
Competitors are seeing movement too. Rival app Birda recorded 31 percent more downloads this year, per Sensor Tower, suggesting the category is growing faster than any single player.
Demographics matter here. Barry confirmed the lab observes a generational shift: younger audiences and first-time birders—even people in their 30s picking up the hobby for the first time. Gen Z's reported surge in nature reserve engagement tracks with this trend. The app's dual-input design (audio or visual) lowers the friction to entry; you don't need binoculars or field expertise to start.
But accuracy limits persist. Seasoned birders, experts, and conservation groups caution against using Merlin as the sole authority for official sightings. The app misidentifies in edge cases, which matters when records feed citizen science databases or regulatory compliance.
