Every gap between sales and procurement ended on the CEO’s desk.
An industrial distributor asked for an AI strategy. We measured three options against the company’s own numbers, and advised against the cheapest one.
- Client
- Industrial distributor
- Size
- About 25 people, four warehouses
- Asked by
- The CEO
- What we did
- Find
- To the map
- One week, two meetings
- Status
- Map delivered. The decision is the client’s.
- Published
- 16 May 2026
What the map found
The client’s own figures. Not our results: the starting point every option was measured against.
789
offers sent in April. 420 of them were won.
The client’s own figures, April 2026~75
offers a month lost to missing follow-up: at least 20% of those not won, by the sales lead’s estimate.
The client’s own estimate, April 20262–6
hours a day each sales rep spent entering data into tables.
The sales team, May 202680%
of the CEO’s time on numbers and papers, 20% with people.
The CEO, May 202601 · The question
“80% of my time goes on numbers and papers, 20% on people. I want to flip that.”The CEO, translated from Serbian
The company distributes industrial products from four warehouses, with about 25 people. Customers order in metres, suppliers deliver in kilograms, and someone converts between the two by hand.
Five sales reps wrote offers in the ERP, then logged their time, contacts and follow-ups in spreadsheets. What procurement needed reached it through one shared tracking sheet. When a column broke or a line went missing, procurement couldn’t order, the two teams passed it back and forth, and it ended with the CEO.
She asked for an AI strategy. The obvious move on the table was a Microsoft Copilot licence for everyone: about €415–525 a month for 25 users.
1–2
hours a day the CEO spent sorting out gaps between sales and procurement.
The sales lead’s estimate, May 202602 · What we found
The business ran on one shared spreadsheet and on what people remembered.
2
places the same stock lived: the ERP and a spreadsheet the warehouse kept, matched by hand.
Process walk-through, May 202620–30
minutes lost each time a formula column broke, which happened most days.
The sales team, May 20260
follow-up reminders. Each rep kept their own 10–20 regular customers in mind; everyone else was left to chance.
The sales team, May 2026Copilot answers when someone asks. The CEO had no time to ask, and the knowledge that mattered wasn’t in a document it could read: it lived in one spreadsheet and in people’s heads. AI on top of the same spreadsheet would have sent her alerts about the same gaps. Either way, she stayed the place where everything met.
03 · What we delivered
An AI investment map, in a week.
Written for the CEO and the sales lead after two meetings, before any data work.
04 · The map
Three options, measured against their own numbers.
Not against features. Four questions from their April figures decided the recommendation.
The map · three options, four questions
Their own numbersThe readout: what each option would change.
If the demo didn’t convince the team before the start, the work wouldn’t start. If the pilot didn’t cut entry time in its first four to six weeks, it wouldn’t go to the whole team.
05 · What happened next
The decision is the client’s. The map is theirs either way.
They haven’t started a build with us, and we don’t know yet what they chose. What they have is the map:
Their own numbers in one place, which they had never put side by side.
Three options priced as kinds of investment, with what each would cost to run.
A recommendation, and the reasons the two cheaper options would leave their numbers where they were.
A phased path with stop points they control.
06 · What it shows
Chat doesn’t fix silos.
When the knowledge lives in one spreadsheet and in people’s heads, AI on top of that spreadsheet leaves the CEO as the place where everything meets.
If this sounds like your company