“The taxonomy week on Retention Kernel Studio forced us to delete twenty-one events. Painful. Also the first time our iOS and Android funnels agreed.”
Studio notes · Dacre Banks
The kernel is the event you can still explain six months later.
Featurekernelbase teaches App Analytics as a written discipline. We work with product, data, and engineering leads who are tired of dashboards that cannot survive a naming dispute.
Open the course ledgerHeard in the last two cohorts
“I came for charts. I left with a ledger. The studio is slower than a vendor workshop, and that is the point.”
“Session Cartography did not magically fix our sampling. It did stop us from presenting sampled DAU as if it were a census.”
Programmes
Three rooms, one ledger
Flagship · 16 weeks
Retention Kernel Studio
Design the few events that actually describe why people stay, then hold the rest of the catalogue to that standard.
Read the syllabus
Atelier · 6 weeks
Event Schema Atelier
A naming and property system that still makes sense after a reorganisation, a new platform, or a new contractor.
See listing
Intensive · 4 weeks
Funnel Geometry
Drop-off is rarely a straight line. This room treats each step as a shape with assumptions you have to write down.
See listing
How the studio teaches
We grade the sentence before we grade the chart.
Most App Analytics training starts in a vendor interface. We start on paper. If a teammate cannot say, in one sentence, what an event claims about a human action, we do not instrument it — however pretty the resulting tile.
Cohorts are small. Work is reviewed in writing. Instructors have spent years inside product organisations, not on a conference circuit. The United Kingdom consumer and privacy context is treated as a constraint, not a footnote.
You will leave with a telemetry ledger: a living catalogue of kernels, properties, and known lies. You will not leave with a folder of screenshots.
The studio’s method
11
studio cohorts since 2019, run from Dacre Banks with satellite days in Leeds.
73
product organisations that have sent at least one analyst or PM through a room.
7 yrs
median time our instructors spent inside analytics orgs before teaching here.
16 wks
length of the flagship Retention Kernel Studio cycle, including critique weeks.
What we actually change
Capabilities, not slogans
-
Kernel selection
Choose the handful of events that describe retention, activation, and paid value — then refuse to let vanity events share their names.
-
Property hygiene
Enums, nulls, and platform drift are treated as editorial problems. You will write a property bible that engineering can implement without a second meeting.
-
Sampling honesty
Mobile SDKs sample. We teach you to mark every chart that is not a census, and to stop mixing sampled and unsampled series in the same slide.
-
Board-ready narrative
A retention curve is an argument. We practise the version you can take to a UK leadership team without hiding the caveats in appendix six.
Next intake
Bring a messy catalogue. Leave a ledger.
Places are limited because critique does not scale. Fees are published without checkout — write to the studio if a room looks like the right pressure for your team.
Write to the studio