One-tap dose logging is Unfair's signature fast-entry path — a single tap from the lock-screen notification, Home Screen widget, or the stack row on the Today tab that confirms a planned dose landed on schedule. The tap writes a complete dose event using the plan's expected values: the planned product, the planned serving size, the current timestamp, and the current dose window. It exists because adherence collapses the moment logging takes more than a few seconds, and an adherence log that is mostly empty is not a log at all.
Why it matters
Tracking apps fail on the same failure mode — logging a dose requires opening the app, finding the right product, confirming the serving, tapping save, and dismissing a confirmation. Total time, 12–20 seconds. Do that four times a day for two weeks and most users stop. One tap drops that cost to roughly one second and moves the work that used to happen at log time into setup time, where the user has attention to give. This is the practical shape of the fast entry paths principle.
What a single tap actually captures
A one-tap confirmation is not a thin event. It captures everything the plan already knows, defers a small set of fields to the background, and leaves a smaller set for the user to correct later only if they actually changed something.
| Field | How it is captured on a one-tap | Example value |
|---|---|---|
| Product identity | Auto from the scheduled plan row | Thorne Magnesium Bisglycinate 120 mg |
| Serving size | Auto from plan (user's confirmed default) | 2 capsules, 240 mg total |
| Timestamp | Auto at tap time | 2026-04-21 22:14 local |
| Dose window label | Auto from current time vs plan | Bedtime |
| Adherence state | Auto — marked taken on schedule | On-time |
| Source surface | Auto — which UI fired the tap | Lock-screen notification |
| Route and formulation | Inherited from product metadata | Oral capsule |
| Meal context | Deferred — inherited from plan default | "With food" if configured |
| Side-effect tag | Deferred — asked only if side-effect tracking is on | None unless user adds one |
| Serving correction | Manual — only if user opens the dose row | e.g., "1 capsule, late" |
| Notes | Manual — only if user opens the dose row | "Half dose, feeling jittery" |
The split matters. Auto fields are high-adherence signal and reasonable-precision dose signal. Deferred fields become available without blocking the tap. Manual fields are left alone until the user actively flags that today was different.
When one-tap is the right choice
A single tap is the right tool when the plan is correct and the user is executing it. It fits most mature stacks most of the time — a steady magnesium bedtime routine, a morning multivitamin, a pre-workout block that does not change week to week. It is a poor tool when dose size, product, or route is actually changing.
- Use one-tap when the plan row matches reality, the serving is the default, and nothing about the day is unusual.
- Use the manual sheet when serving, timing, or product differs from the plan, when a new product is inside its first seven days, or when an adverse effect is under investigation.
- Use a correction when the tap already landed but the details were wrong — edit the event rather than adding a duplicate.
A reasonable default is to pin new or recently changed compounds to the manual sheet for the first 48–72 hours, then release them to one-tap once the dose is stable. Unfair applies this rule automatically.
Confidence weighting inside the model
The ranking model does not treat all dose events equally. A one-tap event contributes high adherence weight and medium dose-precision weight. A manual event contributes medium adherence weight (it arrived late) and high dose-precision weight. The model is explicit about that trade-off so that a stack scored mostly from one-tap events carries appropriate uncertainty, and a stack with manual serving detail gets credit for it when an adverse effect needs to be isolated. Tags captured through tag-based logging attach to the dose event regardless of which path created it.
Edge cases the tap handles cleanly
- Split packs and partial doses. If the user planned 400 mg and took 200 mg, one-tap is the wrong path. Open the dose row and edit the serving; the event keeps the same timestamp and window.
- Late taps. A tap outside the planned window is marked late rather than on-time, and the delta (e.g., "+42 min") is stored for the feedback loop.
- Missed doses. A window that closes without a tap is marked skipped-dose, not interpolated.
- Double taps. A second tap inside a short deduplication window is treated as a confirmation, not a second dose.
A numeric example
A user on an eight-product morning stack spends roughly 95 seconds logging by the manual flow (12 seconds × 8 products, plus navigation). One-tap brings that to about 10 seconds across the same eight products, with the same plan metadata attached. Over 60 days that is about 85 minutes of recovered time, and the observed cadence impact in our internal beta was a jump from 62% to 91% on-time completion — the difference between a reviewable cycle and a cycle the model refuses to score.
How this appears in Unfair
One-tap entry points are wired into the lock-screen notification, the Today tab stack row, the Home Screen widget, and the Apple Watch complication. Every tap writes a complete dose event and streamlines your tracking without asking the user to leave whatever they were doing. Corrections remain available on the dose row for 72 hours, and the daily check-in stays a separate surface so adherence logging never competes with outcome logging.
Clinical safety note
One-tap is a path of least resistance, which is exactly the wrong path when symptoms are changing quickly. During a new trial, a suspected adverse effect, a dose titration, or any clinician-guided change, switch to manual logging and capture exact serving, timing, and context. The goal is precise evidence for the clinician, not a clean streak.