Loading Motifs…
Put the one personalisation question inside onboarding, after the account exists, and say what the answer will improve. Asked in the middle of signup, the same question cost Pinterest signups.
−30% signupsPinterest's Google signups when a gender question sat between Google sign-in and registration, even as activation rose about 7%. Moved into onboarding with a stated reason, completion rose 11%. Company-reported; relative or absolute not stated.
Product screens: Pinterest via Banani ↗. Screens illustrate the product; the number is sourced in the evidence below.
Built from published experiments and company reports across several products. Nobody has checked the screens on a dated day, so read each number with the caveat printed beside it.
Put the one personalisation question inside onboarding, after the account exists, and say what the answer will improve. Asked in the middle of signup, the same question cost Pinterest signups.
Recommenders need personal signals that many new users have not given, and people often feel uncomfortable giving them; at Pinterest, gender was missing for a good share of people who signed up with Google (Sophia Feng, Pinterest).
This play is not tied to a full playbook. Use it for this decision alone.
Asking between Google sign-in and registration cut Pinterest's Google signups.
COMPANY-REPORTEDWhat it does not show Whether the percentages are relative or absolute, the comparison, sample size and period are not stated. The post does not say whether this version explained why gender was asked. Measured drop in signups; the net effect on activated users is not reported.
Exploring effective user signals Sophia Feng · 2018
The same question inside onboarding, with a stated reason, raised completion.
COMPANY-REPORTEDWhat it does not show Whether the 11 percent is relative or absolute is not stated, so it is kept out of structured results. Sample size, baseline, period, significance and randomization are not described; the post calls it an experiment. Position, design and the stated reason changed together, so none is isolated. No activation or engagement figures are given for this version.
Exploring effective user signals Sophia Feng · 2018
Completion rose even for users whose answer Pinterest already had.
COMPANY-REPORTEDUse the play for this decision alone, or combine it with an existing product strategy after checking for conflicts.
A stated reason can read as a warning rather than a benefit.
Asking for a sensitive attribute such as gender without offering to decline, clear answer options, or a way to change it later.
If the answer does not visibly change what the user sees, the stated reason stops being true.
Plus everything above, as plain text for your coding agent, with a short instruction on top: check the fit, say what the evidence doesn’t support, and adapt it rather than copy it. Paste it and you get an answer, no prompt to write.
Pinterest's engineering post on experiments from late 2017, and one controlled experiment on disclosure · evidence reviewed 25 September 2026.
Explore company-inspired strategies that connect your product, experience, and growth.
Browse the playbooks ↗What it does not show Whether the eight percent is relative or absolute, the comparison, sample size and period are not stated. The post calls it an experiment and does not describe its design.
Exploring effective user signals Sophia Feng · 2018
In a controlled experiment, explaining why did not increase disclosure.
SUPPORTINGWhat it does not show A mock-up; participants were told they would receive no recommendations, so they could not react to real results. Every one of 31 requests carried a justification, unlike a single onboarding question. Effects are reported as structural-equation coefficients in standard-deviation units and a figure; no disclosure rates by condition are given in the text.
Making Decisions about Privacy: Information Disclosure in Context-Aware Recommender Systems Knijnenburg and Kobsa · 2013