What a photo pair can't tell you
- Whether the change was gradual or happened in one step.
- Whether it was still improving at day 14 or had already plateaued at day 6.
- How fast it reverts once you stop — the number that determines what this actually costs per year.
- Whether the lighting changed between the two frames. Usually it did.
Why the middle is the useful part
The shape of a curve tells you about a mechanism; two endpoints tell you about a claim.
If brightness climbs steadily for ten days and flattens, you've learned your ceiling. If it jumps on day 2 and holds, you removed surface film and never touched intrinsic colour. If it climbs and then decays over three weeks, you've learned your real reapplication interval. Same endpoints, three completely different products.
None of that is visible in a photo pair, and all of it is visible in a weekly reading.
The same problem, everywhere else
Gum appearance has the same structure — the interesting information is that it moves at week two, which is exactly the point at which someone who started flossing on week one has already given up.
Tongue cleanliness has it inverted: it moves so fast that a photo pair overstates the effect of whatever you happened to do that morning.
In both cases, the fix is the same. Measure at a fixed interval, keep the conditions constant, and read the slope instead of the endpoints.
What we built instead
ToothPal's whole design sits on this: a weekly reading, the same five photos, the same rubric, and a trend line rather than a verdict.
That produces a less impressive marketing asset and a considerably more useful one for the person actually doing the routine.