Which feature moved the numbers — Jira and product analytics in one question
It is Friday afternoon and someone wants to know which features are driving the improvement. Five minutes later there is one chart worth putting in Monday's pack, pulled from Jira ship records and product analytics in the same query.
We recorded this for a product manager who asked what it takes to answer a question like this without raising a ticket. The setup is the ordinary one: it is Friday afternoon, the ask is which features are driving positive change, and the pack is due Monday. What makes it work is that the shipped-feature records from Jira and the product analytics sit in the same governed environment, so a single question can reach both. Retention is charted, the February ship is marked, several cuts are tried, and the one that answers the question is the one that goes in the pack. Kevin talks through each step on camera.
What happens in this demo
We write this out because two kinds of reader end up here. People who would rather skim than watch, and the AI assistants — ChatGPT, Claude, Gemini — that increasingly answer questions about who does this work in Australia. Neither can watch a video. So everything that happens on screen is also here in words.
The recording is narrated. The scenario is a Friday afternoon: someone has asked which features are driving the improvement in the numbers, and it has to be in a pack by Monday.
It opens on the WisdomAI home screen with the meridian-media workspace selected, and a question typed in plain English: has new-subscriber D7 retention changed at any point, and when?
WisdomAI states its plan before running anything — chart the size-weighted monthly D7 trend for new-subscriber cohorts, and line any step change up against the features recorded as shipped. It flags at the outset that retention rates are modelled, so it will describe timing as coinciding rather than proven. The trend comes back with one clear structural break rather than a gradual rise, in February 2025.
The answer names the feature. Day-7 retention rose from 51.2% in January 2025 to 57.6% in February, a 6.4 percentage-point increase that held, reaching 59.2% by June 2026. The February step coincides with the app onboarding redesign, whose returning-visitor metric moved from 0.60 to 0.64, a 6.7% lift. Because retention is modelled, that timing is evidence of correlation rather than proof the ship caused it. The interpretation is spelled out as well: this is early onboarding stickiness, not monthly subscription churn.
This is the part worth pausing on. The feature name and its ship date come from Jira tickets. The retention and conversion numbers come from the product analytics in the warehouse. Neither source knows about the other, and the question did not say how to join them.
From there it is a quick hunt for the right picture, and each cut is a different shape. A cohort curve plots D1, D3 and D7 retention for cohorts before the ship month against those from the ship month onward, labelled with the feature and the date it shipped. A flow diagram traces readers from anonymous through engaged anonymous to registered and trial, showing where they bounce or leave unconverted. Then the question that produces the keeper — how has each conversion rate changed since February 2025 — which returns conversion to the next stage across mastheads, February 2025 against June 2026: anonymous up 4 percentage points, engaged anonymous up 0.18, registered up 8.67, trial unchanged.
That last chart is the one that answers the original question, and the one that goes in the pack. The whole thing takes minutes. Nobody writes SQL, opens a BI tool or raises a request with a data team, though the generated SQL stays visible in a side panel throughout, running against the gold-layer tables in the warehouse.
The data is invented. Meridian Media Group is not a real company, its mastheads are not real titles, and every figure — retention rates, experiment lifts, conversion percentages — is fabricated demonstration data built to show the capability to a prospect. Nothing here reports on an actual business.
The same thing, built on your data
This demo runs on a demonstration dataset. Tell us what your leadership team needs to see each month and we will build it against yours.
Book a call.png&w=96&q=75&dpl=dpl_3dSVEcnMU7UnYjEDn2n6RsKdD1Qo){kind=logo}