> If 25% store their credit card details in autofill, and 4% would decide it's not worth the bother, you're still ignoring the ones for whom autofill works as expected and inputs the correct data. I can pull a similar number from nowhere, but even if it's 50%, the result is significantly off of what this "researcher" estimated.
Fairly sure you've entirely missed the point here, and possibly not even read the article. But I'm feeling nice, so I'll quote the relevant paragraph here:
"Based on Chipotle's publicly reported average order value of $16-$17 and assuming that fixing autofill would increase transactions by half a percentage point, Grigsby estimates that Chipotle could clear an extra $4.4m in sales annually by eliminating this bug"
The question isn't around how many people successfully navigate the process. Instead, it's about how many give up when they hit this stumbling block.
If only half a percent of their customer base (aka: one in two hundred, alternative-measurement-scale fans) hit this problem and walk away, that's the equivalent of $4.4 million to Chipotle's bottom line.
To be fair, this suggests their overall revenue is around $880 million, and at that point, the odd million here or there perhaps isn't that much of an issue - though I'd love to find even a fraction of this sitting down the back of the sofa!
However, it's also money they can effectively get for free - after all, the cost of getting a fix through dev/QA is unlikely to be more than a few thousand dollars...