We publish what we actually believe about data and decisions.
No gated PDFs, no vendor talking points, no trend pieces. These are the arguments we make inside engagements — that most dashboards go unread, that most forecasts are never tied to a decision, that the third pipeline rebuild is never a technology problem — written up so you can disagree with them before you hire us.
Six pieces a year, not sixty
Everything here came out of work we actually did. Each piece carries a claim we are prepared to defend in a boardroom.
The forecast is not the decision
Accuracy is a means, not a result. We show how to trace every model back to the order, the price or the roster it is meant to change, and how to retire the ones that change nothing.
Put the privacy boundary before the cloud
Real-time dashboards usually ship raw events to a cloud store and control access afterwards. Stripping the personal data in the collector costs two days, closes the compliance question, and turns out to be the cheaper design on four separate axes.
Two short things we noticed this month
Smaller than an essay, too specific to sit in a newsletter. Both come from work in flight across the practice.
Nothing to subscribe to yet
There is no mailing list. When there is something worth sending, it will be announced here first. Until then the published projects are the newsletter — they change when the analysis does.
See the published workPrefer to argue about it in person?
Bring the piece you disagreed with. In 20 minutes we will take one of your recurring decisions apart and tell you plainly whether a better forecast would change it — and if it would not, we will say that too.