Flagship · 6 weeks
Cohort Signal Studio
A supervised cycle for people who already ship apps and still feel their audience language is borrowed. Informational fee: £1,140 per seat. Enrolment is by conversation, not checkout.
What you should be able to do afterwards
- Name three audience slices with a screen, a week, and a behaviour — not a mood board.
- Write a habit window with a start condition and an honest end condition.
- Mark every product claim as observed, inferred, or hoped, and keep the hoped ones off the board pack.
- Explain to a sceptical finance partner why one metric is being retired.
Modules
- Slice before story You pick one live surface. We refuse “all users” as a unit of analysis. Homework is a one-page slice card with exclusions listed in plain English.
- Entry traces Install, invite, deep link, store listing. You annotate where the audience you wanted actually arrives, and who arrives instead.
- The pause Permission dialogs, empty states, and the first unpaid wait. This module is slow on purpose.
- Habit windows Not streaks. A window is a time-bounded pattern you can watch without inventing loyalty.
- Return without romance Who comes back, after what silence, and which message would be dishonest to send them.
- Claim court You present five sentences. Peers try to break them. The surviving sentences become your studio record.
Tutor
Amira Cole
Amira spent a decade in research ops for consumer apps in London and Manchester before opening the Swayfield desk. She marks hoped claims with a blunt pencil and still reads session notes the night before studio.
Informational fee
Signal Desk seating for this studio is listed at £1,140. Materials are included. Travel to Swayfield, if you choose an in-room week, is not. See fees for neighbouring seats. There is no payment form on this site.
Questions we actually get
Do I need SQL?
No. You need permission to look at a product and the patience to write. Exports in a spreadsheet are enough. If your only access is a locked vendor dashboard, say so in the first week so we can plan around it.
Can two people share a seat?
One named participant per seat. A silent observer may sit in if both of you write your own claim cards. Shared slides tend to hide who actually observed anything.
Where does this fall short?
The studio will not give you a growth model, a media mix, or a forecast you can paste into a fundraise. If your app has almost no returning people yet, the habit-window module will feel thin — we say that in week one rather than pretending the method scales to empty rooms.
Notes from the last two cycles
Claim court was uncomfortable in a useful way. I had been quoting “engaged weekly users” for a feature that three people touched. The limitation line on my final card is the part I still keep on my desk.
4 / 5 — dense, a bit writerly, and the Pause module changed how we talk about notification copy. I wanted more time on store listing traces than we got.