Skip to content

Is your skincare routine actually working?

Is it working?Run a fair test.

Take a baseline photo, use one product for one concern, and check back weeks later. ProofLoop tells you whether the score moved further than normal measurement variation, or says plainly that it can't tell yet.

No photo to hand?

Hydration, example trial

Baseline 69, normal variation ±5

Illustration
Example dial: baseline 69, normal variation plus or minus 5. Use the slider below to move the example check-in.

69

Baseline
69
Check-in
69
Change
0
  • Hydration
  • Oil balance
  • Redness
  • Texture
  • Pores
  • Blemishes
  • Radiance
  • Under-eye
  • Measured by YouCam
  • Compared over weeks

The problem

The mirror is a bad instrument.

Three weeks into a new serum, you look and hope. Different light, a good night's sleep, a new camera — any of them can look like progress. Without a baseline and a sense of normal variation, a small change is a guess.

A woman checking the skin under her eyes in an oval mirror by a bright window.
Editorial photo by Pavel Danilyuk on Pexels. Not a ProofLoop user or result.
  • The light changes

    Window light at 8am and a bathroom bulb at 11pm make the same skin look different.

  • Scans wobble

    Two analyses of unchanged skin rarely match exactly. Some wobble is just the instrument.

  • A match isn't proof

    A product recommended for your skin is a hypothesis, not evidence it's working.

  • Small numbers mislead

    Hydration up 3 can be real or can be noise. Without a band, you can't tell which.

  • New phone, new skin?

    Pores and texture react to sharpness and compression, so a different camera can move them.

  • When do you stop?

    Quit too early and you'll never know. Keep going forever and you keep paying.

Why a fair test

A change only counts if it's bigger than the noise.

Before trusting a number, we measured how much the instrument wobbles on its own. On 11 October 2026, ten live analyses of one face gave these results.

How much scores moved when the skin didn't

  1. 0

    Same photo, analysed twice

    The live API gave identical scores.

  2. ≤2

    Reframed, shifted or 18% darker

    The skin hadn't changed; scores moved by at most 2.

  3. 5

    Same photo re-encoded in the browser

    Pores and texture moved most: fine detail is sensitive to sharpness.

So a check-in has to move further than this

  • Radiance, under-eye

    ±3 measured

  • Hydration

    ±5 measured

  • Pores, texture

    ±7 measured

  • Blemishes, oil balance, redness

    ±7 assumed

One face, ten analyses: a floor for real-world variation, not a validation study. Concerns that scored 99–100 on the test face couldn't be measured, so they get the widest band, ±7, labelled “assumed”. An optional instrument check can widen your own band before anything counts — never narrow it.

Read the method and its limits

How it works

Six steps from a hunch to a fair answer.

The test is set up before you know the outcome, so the result can't be talked into anything.

1 of 6 · MeasureDay 0
  1. Day 0

    Step 1: Take one clear photo.

    Consent first, then a framing oval so your face fills the frame the way YouCam needs. Eight concerns come back in about ten seconds, with progress straight from our server.

  2. Day 0

    Step 2: Choose one concern to test.

    Each focus is tied to one measured score, so a later check-in can say whether it moved. One focus keeps the test fair.

  3. Day 0

    Step 3: Pick one product, see every reason.

    Matches come from the catalogue by explicit rules: a tier, the reasons, cautions drawn from your own scores, and anything your filters set aside.

  4. Day 0

    Step 4: Save the baseline.

    Your scan becomes the reference point, together with the normal-variation band for your focus and the scores that would count as up or down. They're fixed before you start.

  5. Weeks later

    Step 5: Check in under the same conditions.

    Same spot, similar light, same camera. ProofLoop compares the lighting of both photos before any units are spent, refuses the baseline photo itself, and tells you when a comparison isn't fair.

  6. Planned day

    Step 6: Read an honest verdict.

    A measured change, no clear change, or inconclusive — with the band, the numbers and every concern for context. Never a claim about why.

The verdict

Three honest answers. None of them guaranteed.

Every check-in ends in one of three readings, judged against the plan you fixed on day 0. Pick one to see how the dial reads it.

Illustration

Within normal variation.

The difference is inside the band we see between scans of unchanged skin. That's an honest answer, not a failure — many products take longer, and some don't change what a camera can measure.

Example: hydration 69 → 72, band ±5.

No clear change example: baseline 69, check-in 72, a change of +3. Normal variation is plus or minus 5, so this is within normal variation.

+3

vs band ±5

  • Upright needle: your baseline
  • Berry arc: normal variation, ±5
  • Red needle: the check-in, dashed while inside the band

Inside the app

Every screen is built around the evidence.

Each number says where it came from: a live YouCam scan, the recorded sample, or the labelled synthetic demo.

Scores you can read

Eight YouCam concerns on one scale, a summary written from your numbers, and YouCam's own detection masks over your photo, on your device only.

Reasons, not percentages

Tiers from explicit rules, with cautions drawn from your own scores and anything your filters set aside.

A plan fixed in advance

The baseline, the band and the scores that would count as up or down are saved before the first check-in.

A verdict that shows its working

Before and after, the band, and every concern for context, with only your focus deciding the answer.

A real YouCam Skin Simulation, only when you ask. Rendered from your own photo for supported concerns, it shows what a goal looks like, never what a product will do. There's no example here because it needs your photo.

Trust

What we promise, and what we don't.

A measuring tool is only as useful as its stated limits. Here are ours.

What ProofLoop does

  • YouCam does the measuring

    Scores come from the YouCam AI Skin Analysis API. ProofLoop shows them as returned and never fills in a missing one.

  • The rules are fixed before the check-in

    The band and the scores that would count as up or down are saved when the trial starts and don't move afterwards.

  • The baseline photo can't be reused

    A one-way fingerprint of each cleaned photo lets a check-in refuse the baseline photo, which would score identically.

What it can't tell you

  • Say why a score changed

    A measured change after starting a product isn't proof the product caused it. ProofLoop says so every time.

  • Predict a product's result

    Skin Simulation visualises a goal. It is not a prediction of what any product will do.

  • Diagnose or treat

    It measures how skin looks in a photo. It isn't a medical device.

Your photo is stripped of metadata, sent to YouCam and discarded. ProofLoop never stores it.

Scores and trials sit under an anonymous session. One button on My trials deletes everything.

How it works & privacy

Still wondering if it's working?

Give your next skincare trial a baseline, a fair comparison, and a clearer answer. It takes one photo to start.

Your baseline and its normal-variation band: baseline 69, normal variation plus or minus 5 points.

Day 0

your baseline, before anything changes