The Monday report that took until Wednesday
Plant reporting assembled by hand each week arrived too late to act on. Replacing it returned two days of decision time and a day and a half of analyst effort.
An industrial manufacturer, three plants · First outcome in 4 weeks
Composite. This describes a pattern we see repeatedly rather than a single client, and the figures are illustrative. No client is named without written consent.
Where it started.
Every engagement begins from a written problem statement rather than a role description. This was theirs.
- 01Three plants submitted spreadsheets in three different shapes; one analyst merged them by hand.
- 02The weekly pack landed on Wednesday, describing a week that ended on Sunday.
- 03By the time a yield problem was visible in the report, it had been running for over a week.
In order, and why that order.
Map the manual path first
Before automating anything, the existing process was documented end to end — every copy, paste and judgement call. Two of the manual steps turned out to be unnecessary rather than automatable.
Standardise the input, not just the output
The three plant submissions were reduced to one schema. Most of the effort in the engagement went here, because it was the actual constraint.
Automate the assembly, and monitor it
The pack builds itself from source systems on a schedule, and a failure watcher alerts when a feed stops — the failure mode that quietly kills most automations.
We thought we needed a dashboard. We needed the three plants to agree what a shift was.
Operations director, manufacturing
- →The weekly pack is waiting before the Monday operations meeting starts.
- →Yield variance is caught within days rather than after a fortnight.
- →The analyst who assembled the pack now investigates what it shows.
Automation Pod
Automation fails when it starts from tooling. The Automation Pod starts from the work: which recurring task consumes senior time, what it costs, and whether it deserves a human at all. Then it removes it — and keeps ownership of what it built.
Different industry, same shape.
Daily reconciliation, from a three-day scramble to a morning report
A grocery chain losing margin to unreconciled outlet settlements moved from a three-day manual close to a nightly automated match, with exceptions surfaced before anyone opened a spreadsheet.
Signups were fine. Activation was the business problem.
A SaaS team optimising the top of the funnel discovered the drop-off was three steps in. Instrumenting activation changed what the roadmap was for.
Start with one problem. Not a hiring plan.
A short sprint tells you more about the work than three rounds of interviews.