Naltrexone Usage Guide + Goals: Patient Insights

Cohort: 2,449 patients who completed the naltrexone usage guide (non-draft naltrexoneUsageGuide submission). Source: de-identified aggregate query over form_submissions (usage-guide answers) + patient_goals (goals set at onboarding), 2026-09-29. Caveat: nearly all of this is the web funnel (the native guide only just merged). Goals were set at onboarding, not in the usage guide, so the two describe the same people at two moments.

TL;DR

• The typical patient is an evening drinker (60%) aiming for moderation (56%). The single most common profile: drinks in the evening, wants to moderate, doesn't have a hard workout.
• Half of patients (52%) do work out, which matters for dose timing (take after the workout).
• Habit-stacking is contextual and works: patients anchor their dose to a cue that lands before their drinking window (evening drinkers → "leaving work" / lunch; afternoon → morning snack / 2nd coffee).
• The weekly-limit default (4) is over-chosen (25% of everyone who set a number picked exactly 4), and only ~36% of moderation patients set a number at all. Two collection issues below.
• 4 potential bugs / gaps found (see §4), the most important being a native-vs-web habit-stack mismatch to fix before the native guide reaches everyone.

1. Drinking archetypes

When drinking starts (n=2,449)

Workout / natural endorphins (n=2,449)

Habit-stack anchor (n=2,449)

Anchor by drinking window (the archetypes)

The chosen anchor almost always lands before the drinking window, which is exactly the intended dosing logic (take it 1 to 2 hours before you'd normally start).

Read: the flagship persona is the evening unwinder aiming to moderate, who pegs their dose to the end of the workday. The morning-drinker slice (11%) is small but clinically the highest-acuity group and worth watching.

2. Goals (set at onboarding, same patients)

Primary drinking direction (% of the 2,449 cohort)

*moderation (7), special_occasions (4), no_alcohol (3), less_than_now (2) are old-taxonomy values (see §4).

Grouped: ~60% moderation-leaning, ~35% abstinence-leaning. Naltrexone's moderation story matches what most patients actually want.

Also on file: tracking_commitment 246, weekly_max_drinks 598, not_sure_yet (weekly skip) 23.

"What success feels like" (top selections)

(then calm 148, balance 145, relationships 140, confidence 140, stability 120, freedom 115, pride 109)

"Other focus areas" (top selections)

Read: control, health, sleep, clarity, energy dominate both lists. The payoff patients care about is feeling clear and functional, not abstract sobriety. Good copy/insight fuel: "sleep better, think clearer, feel in control."

3. Weekly max drinks (the number they chose)

598 patients set a specific weekly limit.

• The default (4) is the runaway mode: 148 patients = ~25% of everyone who set a number picked exactly 4, 1.7x the next value. Strong default-anchoring (people accept what the picker starts on).
• "Per day" round numbers cluster: 7 (1/day), 14 (2/day), 21 (3/day) together = 168 patients. People think in drinks-per-day and multiply.
• 12 patients entered values above 28 (30, 35, 36, 40, 42, 44, 50), which exceeds the current picker cap of 28. Legacy / uncapped entries (see §4).

4. Data quality & potential bugs

1. Native usage guide will not match web on habit stacks (fix before full rollout). The live anchors here (leaving work, morning/afternoon snack, second coffee, breakfast, dinner) are the web funnel's dynamic, contextual suggestions (from the habit-suggestions endpoint). The new native guide I built renders the static fallback option set (morning_coffee, supplements, brushing_teeth, lunch, evening_routine, other) — it's missing the contextual anchors patients actually pick, and includes evening_routine, which zero patients chose. When the native guide reaches everyone it will give worse suggestions and fragment this data. Fix: point the native habit-stack step at the same suggestions endpoint the web uses.

2. 10.1% of completers have a blank habit-stack anchor (248 patients). Far higher than the blank rate on the other two questions (0.2% and 0.4%). Worth checking whether the anchor step is skippable, was added after some submissions, or is dropping the value on save.

3. Drinking-goal taxonomy is fragmented. Current codes coexist with legacy/orphan values (moderation, special_occasions, no_alcohol, less_than_now) and a fifth current option (drinking_goal_mostly_abstinent, 204 patients) that isn't in the mobile Edit Goals 3-option list. Two actions: (a) normalize the legacy codes in any reporting; (b) verify mobile Edit Goals covers abstinence + mostly-abstinent (864 patients). If those options aren't selectable there, editing goals could silently re-map them (Edit Goals deletes and recreates the primary on save), corrupting a big chunk of patients' stated direction.

4. Weekly limit is under-collected and default-anchored.

   • Only 598 numbers + 23 skips (621) out of 1,668 non-abstinence patients (37%) engaged with a weekly limit at all. Most moderation patients have no target on file.
   • This is partly the Edit Goals "can't save 4" bug (fixed 2026-09-29): when no weekly row existed, the baseline was fabricated as 4, so saving the default did nothing. That silently blocked exactly the most common value.
   • Values above the 28 cap (12 patients) confirm older, uncapped entries in the mix.

Suggested next reads

• Cross drinkingStartTime x primary goal (is the morning-drinker slice more abstinence-leaning?).
• Pull the "other" free-text habit anchors (22%) to see what real anchors we're missing from the suggestion set.
• Once the native guide is live, re-run this to confirm anchors match web and the blank rate drops.