Notes from rooms we have sat in

A grocery controller, Selangor

We had three outdoor contractors and a digital agency all claiming the same Chinese New Year lift in basket size. App Meadow Core asked for invoices first, which embarrassed one contractor who had billed a flight we could not find on the road. The table that came back gave outdoor a smaller share than the panels’ own study, and a larger share than last-click. Finance could live with that. I still wish the first draft had been plainer about the seven-day versus fourteen-day lag; we spent half the readout on that one choice. Once it was written as a sentence, the argument ended.

A beauty media lead, Kuala Lumpur

The 11.11 voucher codes were eating every other channel in the last-click extract. The full-channel study put a first-touch window on the week before the sale and suddenly television and a modest radio burst reappeared. Nobody in the room loved the even-split alternative they also ran — it made search look too kind — but having it on a second page stopped the agency from saying we had “picked a method to hurt them.” We used the first-touch table in the next plan.

An app campaign owner, Penang-based brand sold nationally

I booked them because the install report looked triumphant and first purchases did not. Wei Jun kept installs and first purchases in separate columns, which was the whole point and also slightly bruising. Paid social got credit for cheap installs that never paid. A small search burst looked expensive on installs and reasonable on first purchase. We cut the install-chasing placement. The study did not tell us what creative to run next; we had to take that back to the agency ourselves.

Longer note: a fashion retailer’s Hari Raya burst

The brand ran television, KL-area out-of-home, and a marketplace push in the two weeks before Hari Raya. Store managers swore the panels drove the fitting-room traffic. Digital reported a clean week of tagged orders. App Meadow Core asked for store traffic counts (already collected at the door) and marketplace orders without customer names. They refused to blend Malaysia and a small Singapore leftover budget into one window, which delayed us four days while we split the files.

The post-mortem found a five-to-nine-day lag from panel upweight to store traffic, and almost no lag on marketplace orders. Last-click had given the marketplace the store traffic that arrived later, because some shoppers still searched the name after seeing a board. The reservation from their side: the door counts were messy on two rainy days, and the study had to leave those days in an “unresolved traffic” line instead of guessing. That honesty made the table slower to present, and more usable in the July planning meeting.

If your own argument looks like one of these rooms, brief a study.