Johor Bahru · feature experiment readouts
Feature Vertex Hub reads the table you already have — primary metric, guardrails, who actually saw each variant — and writes a readout a product manager can take into a launch review. Then we sit with the team until ship, iterate, or stop is written down.
What you receive
Most feature experiments in Malaysian product squads already have a results table. What they lack is a reader who was not in the Slack thread, and a sentence about who was assigned, what moved, what was peeked, and whether the window covered a payday or a festive campaign.
Our flagship work is a feature experiment readout: written in English in five to eight working days, then briefed in seventy-five minutes. You keep your own experiment software. We do not install a thing.
Flagship engagement
From RM 4,800 · Johor Bahru or video
For product managers, designers, and engineers who have already run the experiment and now owe the rest of the company a decision. The pack you send is the hypothesis in its original wording, metric definitions, assignment method, sample sizes, and the results table.
Included: a memo, a one-page decision slip, and a live briefing. Not included: rerunning the experiment, standing access to your accounts, or a second test in the same fee.
Read the full engagement, including what we refuse to claim when assignment was broken.
Related work
Before traffic is assigned
A written review of hypothesis, metrics, sample size, assignment, and stop rules before the feature experiment goes live.
Several experiments, one room
A half-day session with one product squad to walk a backlog of finished experiments and leave with written decision notes.
When two weeks is not the story
An interpretation of holdout groups and longer windows when a short experiment period is likely to mislead.
When the overall number looks clean
A focused memo when the overall experiment looks fine but a region, device class, or new-user group moved the wrong way.
“They made us write the original hypothesis on the briefing slide, even though we had already drifted to a nicer metric in Slack.”
Hafiz Mahmud, product manager, digital banking squad, Johor Bahru — on a savings-pocket experiment we were asked to read after the flag had already been paused.
From a product team
Variant B had lifted the primary rate. The readout spent as much time on who never entered the experiment as on the lift. The decision slip said iterate. The next test was narrower, and the squad read that one themselves.
How a briefing day runs
We work from Bang Mara in Johor Bahru, and by video with squads in Kuala Lumpur, Penang, and further afield. The product manager who owns the flag should be in the room. So should someone who can say no to a ship.
See the briefing agenda — who should attend, what to send, and what we will not reopen once the slip is signed.
Journal
A readout has a job: help a product squad decide. Charts that do not change that decision can live in an appendix.
Crossing a threshold on the primary metric is not the same as being ready to give the feature to everyone.
Guardrail metrics exist to stop a reckless ship, not to give every side chart equal billing in the memo.
Next step
We reply within two working days. Incomplete packs are sent back before a date is held.