Client stories

These are from people who paid for a review or a readout. Titles are the ones they used on the day. We do not publish star rows or invented sample sizes.

Notes from the work

“They made us split Raya weekend sends from ordinary Tuesdays. The combined open rate we had been quoting in board papers was not a number anyone should have been defending. I still wish they had been faster with the draft — we lost a week waiting on our own legal team to release logs, and the five-week clock felt tight.”

Wei Jun refused to add iOS and Android opens until we admitted the Android figure was shade views. Finance were annoyed. The next quarter’s pack was shorter and, for once, nobody asked why the line jumped.

The workshop was uncomfortable in a useful way. Our copywriter had never met the person who named events. We left with ten sentences on a page, which is less than I wanted and more than we had ever agreed.

Office hours stopped me from presenting a ‘recovery’ that was only a change in quiet hours. Priya also chased us when we sent a screenshot instead of the export. Slightly fussy. Correct.

A longer account: service messages after a bank outage

A KL payments app asked for a single-campaign readout after an unplanned outage notice went to 1.4 million devices at 1:12 a.m. MYT. The internal slide claimed a “strong engagement rate.” What the export actually showed was a spike of accidental taps from people silencing the lock screen, plus a second send at 7 a.m. that many devices treated as a duplicate.

We spent the 90 minutes reconstructing the two sends as separate rows. The useful number for their incident review was not the open rate. It was how many people received a third, calmer message at 9 a.m. with a working status page — and how many of those still hit support.

They did not hire us for a full programme review afterwards. They did send the two-page note to operations. That is a reasonable outcome.

If you want work of this shape, request a review.