The Privacy Rule a Fitness App Skipped, and the 60-Day Fix It Chose

Archi4 min read

privacycompliancedata

Contents
  1. The Decision: Delete or Ask Again
  2. 1. Map what you actually collect, not what the notice says
  3. 2. Pick the lawful basis for each field and write it down
  4. 3. Decide your retention window before you touch the data
  5. 4. Build the deletion path before the notice ships
  6. 5. Test the re-consent prompt like a product feature
  7. 6. Write the notice in plain sentences and date it
  8. 7. Brief support before the prompt goes live
  9. 8. Log the decision and who owns each step

The Decision: Delete or Ask Again#

A mid-sized fitness app (call it the company, about 40 employees) had been storing workout histories, heart-rate logs, and location traces under a privacy notice written in 2019. By early 2026 the notice no longer matched what the product did. The app had added sleep tracking, a coach marketplace, and a data-sharing deal with two wearable makers. None of those were described accurately.

The legal team framed two options at a Tuesday meeting. Option A: purge everything older than 12 months and start fresh under a new notice. Option B: send a re-consent prompt to every active user and keep the history.

The numbers drove the choice. About 400,000 accounts had activity in the prior year. Purging a year of history would wipe the streak data that keeps people coming back, and support tickets would spike. Re-consent meant interrupting 400,000 sessions with a modal that some large share would dismiss or abandon.

The company picked a hybrid: purge data older than 18 months with no active session, re-consent on next login for everyone else, and freeze new collection of location data until the notice went live. The checklist below is the version its legal and engineering leads wrote up afterward. It is a template, not legal advice; the rules vary by jurisdiction and by what data you touch.

1. Map what you actually collect, not what the notice says#

Start with the data inventory. Pull it from the code, not the policy. The fitness app found three SDKs writing device identifiers and one analytics vendor collecting precise coordinates that nobody had signed off on. The notice described only "approximate location." That gap was the whole problem.

Ask each team to name the fields it writes and where they land. Compare against the notice line by line.

2. Pick the lawful basis for each field and write it down#

Consent is the easiest to explain and the hardest to operate. If you rely on consent, you need a way to record it, prove it, and withdraw it. If you rely on contract necessity for core features, say so explicitly in the notice. The fitness app moved workout logs to contract necessity and kept sleep and location on consent, which let it shrink the re-consent prompt to two items instead of six.

3. Decide your retention window before you touch the data#

A retention period you cannot enforce is a liability. The company set 18 months for inactive accounts because its storage costs and its legal exposure both pointed the same way. That is rare. Usually retention decisions are a trade between product value and risk. Write the number down, put it in the notice, and build a job that enforces it.

4. Build the deletion path before the notice ships#

Deleting from the primary database is the easy part. Check the search index, the data warehouse, the backup rotation, and every vendor that received a copy. The fitness app's warehouse held a second copy of location traces that its deletion script did not touch. That would have been an embarrassing answer to a user request.

Measure how many users complete it, how many dismiss it, and what happens to their account when they dismiss. Decide in advance: does dismissal mean no access to the feature, no access to the app, or a grace period? The company chose feature-level blocking, which kept most users in the app and cost less churn than a hard gate.

6. Write the notice in plain sentences and date it#

One paragraph per data category. What you collect, why, how long, who receives it. No cross-references to other documents. Put a version number and an effective date on it, and keep every prior version. If a regulator asks what users saw in 2024, you need the answer.

7. Brief support before the prompt goes live#

Support tickets are where privacy changes get judged. The fitness app wrote a two-page script with the five most likely questions and the approved answers. It also gave agents authority to delete an account on request without escalation, which cut the queue.

8. Log the decision and who owns each step#

The checklist ends with a simple table: decision, date, owner, evidence. The company's re-consent rollout took 60 days from the first meeting to the prompt going live. Six weeks later, completion sat above 80 percent and deletion requests were handled inside a day.

The part worth copying is not the outcome. It is that the company wrote down why it chose 18 months, who signed off on the vendor list, and what it would do if the completion rate came in low. Privacy work tends to fail quietly when those answers live only in someone's head.

More articles