All workSimeon Daniel
Health

Acany: understanding your heart, one reading at a time

People with high blood pressure are told to check it at home, but a number on its own means nothing to most of them. Acany explains every reading in plain language, spots patterns across the week, and earns money at the moment the readings matter most: the next doctor’s visit.

My role
Product design, end to end
Platform
iOS mobile app
Focus
First reading logged, weekly return rate, Plus trial starts
Tools
Figma

01The business problem

Home blood pressure monitoring only helps if people keep doing it and understand what they see. Most apps store numbers in a list and leave the meaning to the user.

When the numbers don’t mean anything, people stop logging within a couple of weeks. For a subscription health app, that drop-off is the business problem: no readings, no value, no reason to pay.

02The metric

One number to move, and one I refuse to hurt.

PrimaryPlus trial starts after the first ten readings
SupportingFirst reading logged on day one, and readings logged per week
GuardrailNo rise in people ignoring the urgent-care prompt

03What people actually do

Blood pressure changes through the day. A single reading taken after rushing, talking, or drinking coffee can look worse than it is, and a single good reading can falsely reassure.

People respond to a number with emotion first: worry or relief. If the app doesn’t say what the number means and what to do next, they either panic or stop checking.

Doctors make better decisions from a pattern than from one reading taken in a busy clinic. That gives home readings a clear moment of value: the appointment.

04Hypotheses

Untested. Each one is written so it can be proven wrong.

H1

Explain the first reading

More people log a second reading, because the first one felt useful instead of just stored.

H2

Compare morning and evening

Higher weekly return rate, because patterns are more interesting and more useful than single numbers.

H3

Tie the paid feature to the appointment

More Plus trial starts, because a doctor-ready report solves a real, dated problem.

H4

Offer Plus after ten readings

Better trial-to-paid conversion than asking at sign-up, because there’s finally enough data to be worth sharing.

05Design decisions

Activate

One plain promise

What I did
Understand every reading. There’s no sign-up wall before the first log, so people reach the useful moment before they’re asked for anything.
What I chose not to do
Asking for an account and a full health profile first. Every question before the first reading is another place to quit.
Activate

A number with a meaning

What I did
The reading is labelled, explained by AI in plain words, and followed by what to do next. A safety line for dangerous readings is always visible.
What I chose not to do
Letting the AI diagnose or suggest medication changes. Its job is context and next steps; anything urgent goes straight to a person.
Retain

Patterns, not numbers

What I did
Morning and evening readings sit on one chart, and AI turns the gap between them into one small thing to try this week.
What I chose not to do
A table of every reading. It’s complete, but it makes people do the analysis themselves.
Monetize

A paid feature with a deadline

What I did
The doctor report is the Plus feature. Page one is visible so people can see exactly what they’d get; only the detail behind it is locked.
What I chose not to do
Locking basic logging or the safety guidance. Paywalling anything that keeps people safe would be wrong, and it would destroy trust.
Monetize

Ask when there’s enough to share

What I did
Plus appears after ten readings, opens with what the person has already logged, and sells one outcome: a better conversation with their doctor.
What I chose not to do
A feature-list paywall on day one. At that point there’s nothing to share, so the report has no value yet.

06How AI is used

AI explains readings and spots patterns, but it never diagnoses or changes treatment. Every explanation says what a reading means and what to do next, and anything that might need urgent care points straight to a person, not to more app.

Health data is sensitive, so a real product would need clear consent for every report shared, readings stored securely, and wording reviewed with clinicians.

07How I’d test it

Events I’d track from day one

  • first_reading_logged
  • reading_explained_viewed
  • second_reading_logged
  • weekly_readings
  • pattern_viewed
  • report_preview_viewed
  • paywall_viewed
  • trial_started
  • urgent_prompt_shown
  • A/B test the first-reading explanation against a plain stored number, measuring whether people log a second reading within 48 hours.
  • A/B test paywall timing: after ten readings versus at sign-up, measuring trial starts and trial-to-paid conversion.
What success looks like. People who see their first reading explained log again within 48 hours, and Plus trials started after ten readings convert better than those started at sign-up.

08What I’d do next

  • Connect Bluetooth blood pressure monitors so readings log themselves and typing errors disappear.
  • Work with a clinician to review every AI explanation and the thresholds behind the safety prompt.