Status
Bug confirmed
Surface
us.47skin.com / quiz
Run
run_f4187600
Encountered by
2 of 10 participants
incident report · silent conversion failure

The form submitted. The page rendered. The content didn’t.

A teenage participant filled out the skin quiz with intent, hit submit, and was served literal template placeholder text where his product recommendations should have been. He left. No conversion event fired. No alert tripped. The funnel report shows a healthy quiz-start rate.

01 The evidence#

This is the page returned to the participant after submitting the find-your-product quiz. Not a screenshot of the source. Not a staging environment. The live us.47skin.com production response, captured during a noemica run on May 1, 2026.

47 Skin quiz results page rendering literal template tokens [Type of Routine] and [Skin concern] instead of recommended products
Quiz results page after submit· us.47skin.com · 2026-05-01screenshot 043 / step 39

The headline copy reads: “Because you chose for a [Type of Routine] that can help with [Skin concern]these products can help best.” Those bracketed tokens are template placeholders. They were never replacedwith the participant’s quiz answers. There are no products listed below the headline. The page is otherwise empty.

02 What the participant did#

Ryan, 16, moderate inflammatory acne, no skincare experience, on the site reluctantly because his mom asked him to look first. He spent five turns warming up to the brand: read the homepage, watched the before/after stories, found the Austin testimonial that looked exactly like his own forehead. Then he started the quiz.

Step 33action
Clicked into the Find Your Product quiz. “Ryan would definitely pick the keep-it-simple option, no way he’s committing to a full routine.”
Step 34action
Selected “Under 18” on the age step. “The quiz actually has an Under 18 option, they know teens use this. Ryan would pick that.”
Step 35-37action
Walked through skin type, primary concern, severity. Real input, real intent.
Step 38action
Body concerns: clicked None. Bypassed the email gate. Hit submit.
Step 39page render
Submit returned a results page. Headline placeholders were not interpolated. No products rendered. Empty white space below the fold.
Step 40reflection
Closed the quiz modal. Sentiment dropped from B to F in a single turn.

The participant’s verbatim reaction at breakage

moment of failure · turn 5
“I clicked through the quiz super fast, body concerns? none, obviously, but then the results page showed broken placeholder text instead of actual skin advice. That’s a dead end.
Ryan Matsuda, 16, run_f4187600, turn 5 reflection

Compare that to two turns earlier, when the same participant was warming up:

“OK so the quiz actually asked my age and skin type stuff, that’s kinda cool that they have an under 18 option, feels like they actually thought about people like me. Let’s see what they recommend.”Ryan Matsuda, run_f4187600, turn 4 reflection

03Why your monitoring won’t catch this#

The bug compiles. The page renders. No 500. No JS error.

The server returned a 200. The HTML parsed. Analytics fired the page-view event for the results URL. From the dashboard’s perspective, this is a healthy quiz completion.

What didn’t fire: any add-to-cart, any product-view, any downstream conversion event. Because there were no products on the page to trigger them.

Quiz-start to product-view drop-offs in your funnel get attributed to “the user wasn’t ready” or “the recommendations didn’t match.” The actual cause is that the recommendations were never rendered.

04 Encounter rate in this run#

10
Participants
2
Hit the bug
20%
Encounter rate
0
Triggered an alert

Ryan (16, acne) and Jake (28, cystic acne) both used the quiz the way a real motivated buyer would. Both hit the broken results page. Neither outcome surfaces in a session-level engagement metric, because both participants kept browsing the site after closing the modal.

05 Diagnosis#

What it looks like

String interpolation failure on the quiz results template. Likely a CMS field rename, a localization fallback mismatch, or a JS hydration error that leaves the unrendered template visible to the user.

Where to look

The “Your Products” results route on us.47skin.com. Replicate by completing the quiz on a clean session. The failure is non-deterministic across our two encounters, suggesting state-dependent rendering rather than a flat broken route.

Cost shape

Every motivated quiz-finisher who lands on the broken render is a lost lead. They had stated intent (age, skin type, concern) and got served a placeholder. That’s the highest-intent moment in the entire funnel.

Why testers miss it

QA fills in test data the same way every time.Real-shaped participants with varied attribute combinations and genuine completion intent produced this failure. A staging suite checking that the page returns 200 wouldn’t flag it.


verify · the full participant trace, the screenshot, and the verdict are public
noemica.io/studies/stu_c2ef4f0a?who=run_f4187600&step=39

This run is one of ten in a noemica study of 47 Skin’s orientation mechanisms. Ten participants spanning ages 16 to 52, varied skin concerns and skincare sophistication. Two of them encountered this bug. The remaining eight produced their own findings on what works and what doesn’t on us.47skin.com.

noemica catches this class of silent-failure bug, the ones that pass every uptime check and still cost real revenue, repeatedly across customers. This study took roughly 30 minutes of real-elapsed time, not days. The same approach runs against any environment where a participant can act, websites, static product lines, retail clones, internal tools.

If you want one pointed at your funnel, seb@noemica.io.

Take it with you
If you’d rather just write, seb@noemica.io.