Szajna Snap was a tablet app for the eyeglass lens industry. The job sounds small but it was not: opticians needed precise measurements to order advanced lenses, and the old way meant doing them by hand, then transcribing them into an order. That is slow, and every manual step is a place for a small error to creep in.
The app let them take a photo and get the measurements from image processing, then push the order electronically in the same flow. In the first month it hit 40% of its annual KPI, which honestly surprised even us.
Find the step people dread
The interesting part was not the computer vision, it was deciding which step to remove. The measurement was the part opticians dreaded and the part most likely to go wrong. So that is where automation paid off the most. The electronic ordering on top of it just made sure the good measurement did not get lost in a re-typed form.
The win was not a fancier ruler. It was deleting the ruler from the workflow entirely.
Design so people trust it
Automating a measurement is only useful if the person ordering trusts the number. So the app had to make it easy to see what it had measured and feel confident before sending. I think that is what drove the adoption more than the raw accuracy - it fit the way opticians already worked, and it took less effort than the old way rather than more.
A few lessons that stuck with me:
- Automate the step people already dread, not the one that is easy to automate.
- Show your work. A measurement the user cannot sanity-check is one they will not trust.
- Connect the automated step straight to the action, so a good result does not get lost in manual re-entry.
- Adoption follows effort. If the new way is genuinely less work, people switch on their own.
The image processing was the headline, but the real product was removing a tedious, error-prone step and earning enough trust that people stopped reaching for the ruler.