The 1-minute rule and the 7-day check-in

There are two design rules sitting underneath Precog that almost nobody asks about. They are also, in our view, the two rules that decide whether a behavior-tracking product survives the first year of an actual user's actual life.

This essay is about each one separately, then about how they interact.

Rule one: a minute a day, no more

The first decision was the daily-input ceiling. We capped it at one minute.

Not "we aspired to under a minute." Not "we made it as short as we could." We sat in a meeting and said: one minute a day. If anything pushes over, we cut it.

The math is simple and unforgiving:

• 60 seconds × 365 days = 6 hours per year. The cost the user pays.
• A user who pays 6 hours a year and gets 52 portraits is paying about 7 minutes per portrait. That's a fair trade for an artifact they're going to sit with on Sunday.
• A user who pays 30 hours a year — what most habit apps actually demand — and gets a streak counter is paying for nothing. They quit by month four.

Most habit apps under-respect daily input cost. The data entry creeps. A "quick log" turns into 90 seconds. Multiple slots add up. By month two, the cost is 5 minutes a day, the user is starting to dodge the app, and by month four they've quit.

Precog's slot-by-slot input cost is, deliberately:

• Sleep: ~8 seconds (alarm dismissal logs the wake; HealthKit fills the night)
• Exercise: ~10 seconds (HealthKit fills the workout; you tap "how it felt")
• Water: ~1 second per cup (just one tap)

Total daily ceiling: under 60 seconds for most users on most days. Often closer to 20 seconds.

The ceiling is the survival mechanism. You can't honor a behavior tool that costs you 5 minutes a day, every day. You can honor a tool that costs you under a minute. That difference, multiplied across 365 days, is the difference between Precog still being on your phone in two years and uninstalled.

Rule two: the check-in is weekly, not daily

The second decision was the cadence of the meaningful interaction.

Daily check-ins are what most apps default to. The "how did your day go" prompt at 9 PM. The streak count at the top. The day-shaped window into your data.

We don't think the day is the right unit for the things Precog measures.

Sleep — measured weekly. The body responds to the four-week running average more than to any single night. A user who looks at their sleep score nightly is misreading the data. A user who looks at the seven-day average has a much closer signal to what the body is actually producing.

Exercise — measured weekly. Frequency, not intensity, sustained over weeks, is the variable that matters for long-arc face/body effects.

Hydration — measured weekly. Eight cups one day and zero the next is worse than five cups every day. The week is the unit of consistency.

The day is too short to extract any real signal. The week is the smallest meaningful chunk. So that's our check-in unit.

How the two rules interact

The interesting thing is that daily one-minute input + weekly meaningful check-in are not unrelated. They reinforce each other in a specific way:

A daily input that demanded daily reflection would be exhausting. "How did today go?" prompts every night for 365 days are how habit apps wear out their users. The reflection cost compounds; by month three, opening the app feels like work.

A weekly check-in that demanded daily input would be vague. If we asked you to "review the week" without you having logged anything, you'd reconstruct from memory — and memory is unreliable for things like sleep duration and water intake. The data wouldn't be honest.

The combination — minimal cost across the daily flow, anchored by a meaningful viewing once a week — is the only structure that survives a year of real life. The daily cost is so small you don't resent it. The weekly viewing is meaningful enough that you actually wait for it.

Why "Sunday at 7 PM" specifically

Once you've decided on weekly, the question becomes: which day, what time. We tested a few:

• Friday evening: too much weekend ahead — the data feels half-cooked. The week isn't done.
• Saturday: still inside the weekend. The next-week-question doesn't have weight yet.
• Sunday morning: the day's energy hasn't accumulated. The reflection feels rushed.
• Sunday evening: post-dinner, pre-Monday. Quiet. Phones in hands. Almost everyone has 5 minutes.
• Monday morning: too late. Next week has already started. Reflection is theoretical, not actionable.

Sunday evening kept winning. Within Sunday evening, 7 PM is when most users said they had downtime. So we built the entire app around that single moment.

The portrait is delivered there. The Live Activity counts down to it. The Apple Watch surfaces it. Push notification at 7 PM local. Widget shows the freshly-revealed image until next Saturday's countdown begins.

Everything points at one moment per week. The user knows when it is. The user waits for it.

Why most apps get this wrong

Most behavior apps are designed by product managers who think more interaction is better — more streaks, more notifications, more daily prompts, more nudges. The metric they're optimizing is opens per week.

That metric is wrong for behaviors that compound on weekly or longer timescales. Optimizing for daily opens produces a tool that wears the user down before the underlying behavior change has time to compound.

The right metric, for a product like Precog, is consistent low-friction logging × meaningful weekly viewing × users still active at month 12. We are betting that the under-1-minute / weekly-anchored design produces month-12 retention that streak-based apps can't reach.

The bet won't be settled for a year. But the design rules are clean enough that we'll know early whether the cohort that picks up the cadence picks it up durably. Durability is the only metric that matters at this scale.

— Codeful