Cache Canvas Core

Diagnostic work

Retention diagnostic

A two-to-three-week reading of how people in a habit product come back after the first completed ritual — not after the install.

From RM 8,500 · Two to three weeks

Two colleagues reviewing printed notes and tables at a wooden table

A habit product does not retain people the way a shop or a news reader does. The person who opened your app on Tuesday and completed a ten-minute stretch is not the same kind of returner as the person who opened it, dismissed a reminder, and left. If those two events share a name in your log, every table you have been showing yourselves is mixed.

The retention diagnostic is the piece of work we do most often. It is a reading, not a dashboard handover. We take the ritual you actually ask of people — the recitation, the fast window, the physiotherapy circuit — and we ask when that ritual is completed, when it is skipped, and when someone comes back after a skip. Those three moments become the bones of the cohort tables.

We work from an export you prepare. We do not need a login. We do not install anything on your machines. If the file still carries emails or raw device identifiers, we stop and ask you to strip them. The diagnostic is meant to be something you can put on the table in a product meeting in Kuala Lumpur or a kitchen table in Ipoh, not a file that should never have left the building.

Malaysian and regional habit products have calendars that imported playbooks ignore. Ramadan changes fasting and prayer products. School terms change drill apps. A public holiday week is not a “seasonal dip” if your users are sitting exams. The diagnostic treats those weeks as named cohorts, not as noise to be averaged away.

You should not book this if you want someone to “look at the numbers” for an afternoon. The first canvas and the first honest table usually take the better part of a week, and the working session is useless if you have not read the draft.

Who it is for
Independent makers and small product teams shipping a daily or weekly ritual — meditation minutes, language drills, fasting windows, physiotherapy exercises, reading streaks, or similar.
What you leave with
A written diagnostic that names which cohorts actually describe your habit, where return collapses, and which event definitions are inflating or hiding that collapse.
Scope
One live product (or one clearly bounded beta). We reconstruct activation as first completed ritual, build return tables by habit cadence, and sit with you for a working session in Ipoh or on a call in Malaysia time.
Who does the work
Led by Aina Rahman, with cohort tables prepared by Wei Jun Tan. Priya Menon joins when the product is classroom- or clinic-adjacent.
Where
Remote across Malaysia and neighbouring time zones. In-person sessions by arrangement at the Ipoh studio.
What you prepare
A one-page description of the ritual you ask people to repeat; a glossary of event names; and an export covering at least six weeks, with user-level identifiers stripped or hashed.
Limits
We do not start until the export is free of obvious personal identifiers. Products younger than four weeks of real use rarely have enough return to read; we will say so rather than invent a story.
Fees
Most diagnostics for a single habit product sit between RM 8,500 and RM 14,000, depending on how messy the event glossary is and whether a second ritual (for example a social share on top of a private session) must be read separately.

Included

  • Written intake on your ritual, cadence, and current event names
  • Review of an export you prepare — aggregated or pseudonymised, never a live production login
  • Cohort tables cut by first completed ritual, install week, and at least one lapse-and-return group
  • A marked habit canvas showing cue, routine, and reward against the events you actually fire
  • A 90-minute working session to walk the tables and agree what to change in logging or in the ritual itself
  • A short addendum after the session capturing decisions

Not included

  • Building or hosting analytics software for you
  • Running paid acquisition tests
  • Accessing production databases or advertising accounts
  • Legal advice on personal data (we will refuse an export that still contains names, emails, or device identifiers we cannot justify)

How the work proceeds

  1. You send the ritual description, cadence, and a sample week of event counts. We confirm whether the export is fit to read.
  2. We draw the habit canvas before any cohort is cut, so the tables follow the ritual rather than the calendar month.
  3. Draft tables and a first written reading go back to you. You mark what feels wrong.
  4. We meet for 90 minutes. You leave with named next changes — usually to event names, to the first-week ritual, or to how you forgive a broken streak.

Write to the studio with the name of the ritual and the week people tend to disappear. We will say whether a diagnostic is the right piece of work. Write to the studio · Rates in ringgit