Defend a definition
Write, in one paragraph, who is in week-zero and who is out. A colleague who was not in the sitting should be able to reproduce the set.
Flagship programme
Retention Signal Studio
Six weeks, fourteen seats, one artefact of yours in the room. The sitting exists for teams who already plot N-day retention and still cannot agree what the cliff means.
You will not be taught which button to press in a vendor. You will be taught how to write the sentence that sits under a retention table, how to cut a cohort without fishing, and how to mark weather so that a store outage does not become a product myth.
Homework is public inside the cohort. If you cannot share even a redacted table, this sitting is the wrong room — consider Private Advisory instead.
Ask for the next sittingModules
01
Event grain and the first week
We settle the birth of the cohort. Install, first key verb, or first paid action: only one may be the origin. You rewrite the first-week event list in verbs a support agent would recognise.
02
Retention tables without the vanity overlay
Classic N-day views are kept, then paired with a ‘return to the core verb’ table. DAU is allowed in the appendix, not on the title slide.
03
Cohort cuts that survive a board meeting
We practise three cuts only: first platform, first successful task, and country of first session. Channel is introduced last, and only with a sample-size warning.
04
Session replay as annotation, not surveillance theatre
Qualitative pairing: five recordings against one unexplained cell in the table. The point is a margin note, not a highlight reel.
05
Critique weeks
Each participant presents. The room may ask about time zones, about deleted users, and about whether a holiday was labelled. Polished decks are discouraged.
06
The spoken readout
A ninety-second recording. If you need a slide to finish the sentence, the sentence is not finished.
What you should be able to do afterwards
Mark weather
Keep a dated list of incidents that belong in the margin of a retention chart, including those that flattered you.
Refuse a single number
Answer a board request for ‘the retention figure’ with a short climate report instead of a false precision.
Instructor
Dr Amira Pell
Amira directed analytics at a UK payments company for six years, then left to stop pretending that a vendor demo is a pedagogy. She holds a doctorate in human geography — maps, not machine learning — and still marks student tables in the margin with a pencil. She teaches this sitting herself. Guest counsel from privacy law attends week four.
She will not migrate your stack. She will not praise a chart because it is colourful. She will ask what would have to be true for the cliff to be a real change in habit.
Fee (informational)
£1,890 per seat, excluding VAT where VAT applies. The fee covers six live sessions, written notes from Amira on your artefact, and twelve months on the alumni list. It does not include vendor licences or a certificate of mastery. Payment, if you proceed, is arranged by invoice after the desk confirms a seat — never on this website.
Studio Membership holders may join this sitting without the flagship fee when the calendar allows; seats are still capped at fourteen.
Questions we are asked twice
Do I need a particular analytics vendor?
No. Bring whatever you use. The limitation is real: this programme does not teach a specific vendor’s UI in depth. Mixpanel, Amplitude, and PostHog are interchangeable canvases. If you need a click-path certification for one tool, this is the wrong room.
Can two people from the same company share a seat?
No. A seat is one voice in the critique. A second colleague may observe silently if you say so at booking; they do not present.
What if our data is under NDA?
Redact. If redaction would still expose a strategy, take Private Advisory instead of the open sitting.
Is there a test at the end?
Only the spoken readout. We do not publish scores. Attendance letters are available on request.
From people who sat this programme
“Week two of Retention Signal Studio forced us to admit our ‘activated’ flag was a marketing email click. I was embarrassed in the room. I was also grateful we did not take that flag to the board.”
I booked because our iOS retention looked healthier than Android and nobody could say why. Amira made us draw sessions on paper first. The ‘healthier’ curve was a background-fetch costume. The sitting did not give us a growth lever. It gave us a quieter chart and a store-ticket we should have filed months earlier.
★★★★☆
Noted on the alumni card: “Strong on cohort birth. I wanted more SQL. They refused, correctly.”