It's 1943. World War II.
The British Royal Air Force is losing bombers at an alarming rate and nobody can figure out why. The military does what makes sense and they study the planes that return from missions, mapping out bullet holes. The wing tips, the fuselage and the tail section.
If they could reinforce the areas getting hit the most, more planes would survive the missions.
Then a statistician named Abraham Wald says, hang on. The planes they're studying are the ones that survived. The bullet holes on those aircraft show exactly where a bomber can take a hit and still make it home. Maybe the areas with no damage aren't undamaged because they weren't hit, but because the planes that were hit there never even made it back.
The RAF had been looking at survivors and mistaking their story for the whole picture.
Wald's fix was simple, but counter-intuitive because he had no data to back it up. Armour the clean spots.
When someone tells me they don't really have users with disabilities, I keep thinking of this RAF story. I point out they're not seeing these users not because they don't exist, but because they likely abandon their product altogether. They're those missing planes.
The absence of complaints isn't proof that everything's fine and the accessibility data you don't have is still important data you can't afford to ignore.