Cohort Signal Lab
Six weeks of guided App Analytics practice for product managers, analysts, and growth partners who already ship — and want telemetry they can defend.
Learning outcomes
- Design an event taxonomy with clear ownership and versioning rules.
- Diagnose funnel leakage without inventing false precision.
- Build retention and cohort narratives that separate product habit from acquisition mix.
- Run a one-hour weekly signal review that produces logged decisions.
- Brief stakeholders with confidence ranges instead of theatrical certainty.
Who it suits
Teams with at least one live product surface and an analytics tool already in place. Absolute beginners in statistics will need extra reading; we do not teach SQL from zero.
Modules
Each week ends with a critique session and an artefact check.
Instrumentation archaeology
Inventory existing events, spot duplicates, and draft a retirement list with engineering allies.
Taxonomy that survives release pressure
Naming patterns, property contracts, and how to reject “just add one more event” tickets.
Funnels without vanity overlays
Define step honesty, handle optional paths, and present incomplete journeys without apology theatre.
Cohorts & retention storytelling
Read curves for habit formation, seasonality, and platform shifts — then write the narrative.
Experiment readouts in practice
Primary vs secondary metrics, peeking habits, and documenting inconclusive results as progress.
Operating cadence & handoff
Install the weekly ritual, assign owners, and package a 90-day improvement backlog.
Instructor
Amira Kell
Former product analytics lead for consumer apps across London and Edinburgh. Amira focuses on decision quality, not tool evangelism, and has coached 14 Portal Vectorcore cohorts.
Informational pricing
Listed for planning only — no checkout on this site.
£1,480 / learner · cohort seat
Includes live critiques, artefact templates, and 30 days of async office hours after the final module. Team packs appear on the pricing page.
FAQ
Do you teach a specific analytics vendor?
No. Examples reference common patterns across Amplitude, Mixpanel, and GA4, but the coursework is tool-agnostic. Bring the platform you already pay for.
What is a real limitation of this course?
We do not cover advanced causal inference or warehouse modelling. If your bottleneck is dbt pipelines or econometrics, this studio will feel too product-facing — and that is intentional.
How much time should I reserve weekly?
Plan for two live sessions (90 minutes each) plus three to five hours of applied work on your own product data.
Can engineering partners join?
Yes. Mixed product–engineering pairs usually finish taxonomy modules faster and with fewer rework cycles.
Learner notes
“Module 4 forced us to admit our D7 spike was a marketing calendar artefact. Awkward in the room — useful on the roadmap.”
“Clear templates. I still struggle with executive briefings, so I paired this with Stakeholder Signal Briefings afterward.”