These notes come from people who finished a Datapacketgrid desk. We edit for length and to remove employer secrets. We do not invent five-star copy.
★★★★☆
Journey Atlas forced us to kill eight campaign-tagged events that were polluting path analysis. The live critique of our Android identity join was the week I stopped arguing with engineering in Jira comments. I wanted more time on subscription pause states; we only skimmed them.
Nattapong V. · analytics manager, Bangkok
Took Funnel Cartography because checkout looked “fine” in the default funnel and terrible in support tickets. Drawing retries as loops — not drop-off — changed the conversation with payments. Still a bit too theoretical if you have never written SQL.
Lina, Chiang Mai
8.4 / 10 · internal L&D form
We sent two PMs to Atlas Desk. One shipped a two-page journey narrative that finance actually annotated. The other found the reading load heavy during launch week and skipped two seminars — recordings helped, but office hours were harder to catch up.
Platform-style review from a fintech L&D lead
Event Taxonomy Atelier was four evenings and a spreadsheet. Unromantic. We finally have an owner for each event family. I would not call the sessions warm; Pim is precise and occasionally blunt about sloppy names.
Client in travel · anonymous by request
A grocery app’s growth review kept celebrating a rising “checkout start” rate while completed orders were flat. The team joined Atlas Desk with a hypothesis that the payment SDK was slow. Journey mapping showed something else: a large share of users opened checkout, backgrounded the app to copy a PromptPay reference from Line, and returned after the session window had closed. The old funnel counted them as drop-off.
Work during weeks five to eight rebuilt session stitching across that backgrounding pattern and split “left forever” from “left to fetch a reference.” The product change was small — persist the draft cart across a longer idle window. The reporting change was larger: the weekly pack now has a loop count next to the conversion rate. Completed orders moved; the original SDK theory was retired.
Extended note, product director · name withheld · Atlas Desk, 2025
Retention Signal Lab, not the flagship. The team tracked “churn” as a single cancel event. Interviews suggested people paused for travel and came back. Instrumentation had no pause state, so journey reports treated every cancel as terminal. After the lab, they added pause / resume to the contract and stopped mailing win-back offers to people who were explicitly resting. Reactivation mail volume fell; complaint volume fell with it. Revenue did not magically rise. The report became honest, which was the brief.
First-name attribution · K. from Phuket, growth ops