Abstract network of connected points on a dark field

Observatory

Product-Led Growth Analytics as four weather layers, not one funnel poster.

The Observatory is Sensor Cloudhub’s working method. It is what faculty mark against in studio, and it is what we mean when we say we teach Product-Led Growth Analytics rather than a tour of a BI tool.

Most PLG diagrams stack acquisition, activation, retention, revenue, and referral as if they were floors in the same building. In practice they are different climates. A campaign can heat acquisition while expansion cools. A PQL can look “hot” while contraction has already started in the same account family. The Observatory keeps those layers separate so one lucky week cannot impersonate product-market fit.

We developed the language in Bangkok because the imported diagrams kept failing local commercial weather: quarterly invoices, LINE-adjacent onboarding, bilingual copy, procurement calendars that ignore your in-app checklist. The method is portable. The examples in class often are not — on purpose.

Signal layers

What we actually look at

I

Activation weather

A time-bounded sequence of jobs, not a single “aha” event. First value, first collaboration, and first invoice are recorded as distinct moments. If you cannot name the window, you do not have an activation metric yet.

II

Qualification weather

Product-qualified leads with explicit vetoes. A score that only rises is a mood board. Sales must be able to send a PQL back without starting a theology debate.

III

Expansion weather

Seat, usage, and package expansion modelled apart from new-logo motion. Weekly review rituals are written down so the Monday pack does not quietly revert to acquisition charts.

IV

Contraction weather

Downgrades, idle admin seats, and silent identity splits. Contraction is not “churn’s little sibling”; it often arrives months earlier and is visible only if you stop averaging it into retention.

Person reviewing product analytics on a phone and laptop

How the atelier uses it

Definitions you can version

Every Observatory artefact has a date, an owner, and a known limitation. When the warehouse changes, the definition is versioned rather than silently edited in a dashboard. That sounds clerical. It is the difference between analytics as folklore and analytics as a craft.

Signal to Expansion walks all four layers. The PQL lab lives mostly in layer II. Activation Desk Intensive stays in layer I until the job sequence can survive an exec question. Fellowship teams often arrive already drowning in layer III charts that were never separated from acquisition.

See which room matches the layer you cannot read

Or write the desk if you are unsure