Food DistributionTen margin levers, and what can be budgeted
Ten workstreams quantified on the company’s own data, never on market ratios.
What was done : Ten workstreams identified and costed on the company’s own data, never on market ratios.
What can be budgeted : About €0.36m net a year, from a measured potential of €0.85m to €1.6m gross. Payback in 9 to 26 months.
The situation
A food distribution group: a warehouse that sells fresh and ambient products to trade customers and supplies the group’s own network of shops. Warehouse turnover of more than €20m, a gross margin of around 20%, an ageing business management system and several till systems. Price-setting and knowledge of the systems rest with very few people.
The management team asked a simple question: what can artificial intelligence do for us, in what order, and for what gain? The framework was set on the first day: value first, and the data stays with the client.
The approach, step by step
| Step | What was done |
|---|---|
| Economic Baseline, the reference performance | A data foundation built during the engagement from a controlled copy of the management database and the shops’ till systems. First yardstick: one point of margin on warehouse sales is worth around €0.2m a year. |
| Value Discovery, identifying the sources of value | Interviews with the teams. More than 160 use cases identified, grouped into ten workstreams that share the same data, the same owner and the same expected result. |
| Value Quantification, quantifying the gains | Each gain quantified from a measurement taken in the company’s own data. No market ratios. |
| Capital Prioritization, prioritising the investments | Workstreams ranked from A to D by return on investment: annual gain relative to the cost of implementation. |
| Value Realization, proposed to the management team | No workstream moves to the next stage without a measured result. The tool advises; the person decides. |
What the data showed
| Measured finding | Order of magnitude | What it opens up |
|---|---|---|
| Sales below cost | Several thousand sales over twelve months, around €0.6m of losses | The first source of value: price recommendations |
| Price dispersion | The same item sold at prices more than 20% apart in 60% of cases | Fragmented price decisions, and therefore room for improvement |
| Stock with no recorded sale | Around €0.45m at purchase cost | The upper limit of the unsold stock opportunity |
| Inactive suppliers and customers | Several hundred dormant suppliers, dozens of customers with no order for three months | Tool-supported reactivation, with a gain that can be quantified |
| Payments | Nearly 40% of payments late | Three to five days gained release €0.18m to €0.3m of cash, as a one-off |
| Consolidated margin | No measurement of margin from the price paid to the supplier to the till receipt | The first building block of management reporting |
The ten workstreams
| Workstream | Estimated annual gain | Class |
|---|---|---|
| Data foundation and product master data | Indirect: a precondition for the other workstreams | Foundation |
| Management reporting: dashboards and plain-English queries | €45k to €60k of avoided cost | Foundation |
| Price recommendations and margin management | €250k to €450k | A |
| Reactivation of inactive suppliers and customers | €90k to €180k | A |
| Clearing slow-moving and unsold stock | €120k to €250k | B |
| Warehouse reliability and cost of poor quality | €100k to €200k | B |
| Accounting, receivables and cash | €80k to €160k, plus €180k to €300k of one-off cash | B |
| Management of the shop network | €100k to €180k | C |
| Automated processing of supplier offers | €60k to €150k | C |
| Critical risks and continuity of know-how | Not quantifiable | D |
| Total | €845k to €1,630k gross per year |
Each workstream was costed with its implementation budget and its running cost: that ratio determines its class. The detailed budgets can be discussed in a first conversation.
What can be budgeted
Around €0.36m net per year can be budgeted. A measured potential is not entered in the budget as it stands. We start from the lower end of each gain and apply a reduction that depends on how robust the measurement is:
| Robustness of the measurement | Share retained |
|---|---|
| Value observed line by line, or a known avoided cost | 70% |
| Gain that depends on behaviour: a recommendation that is followed, a supplier that responds | 50% |
| Figure extrapolated from a sample | 40% |
After these reductions, the gains amount to around €0.48m a year. After the running cost of the tools, taken at its upper estimate, around €0.36m net a year remains. On this prudent basis, the warehouse gross margin moves from around 20% to 21%. Part of the gain relates to cash, avoided costs and the shop network: only the share linked to warehouse sales shows in that margin. The potential of €0.85m to €1.6m would take it to 22% to 24%: that is the target trajectory, not the budget line.
| Period | What is in service | Net annual gain achieved, prudent basis | Annual potential |
|---|---|---|---|
| Months 0 to 3 | Data foundation, management reporting, first price scope | No euros yet, a first measurement | None |
| Months 3 to 6 | Prices, reactivation of inactive accounts, management reporting | €135k | €665k |
| Months 6 to 12 | Slow-moving stock and shop network | €240k | €1,090k |
| Year 2 | Warehouse, accounting and cash | €345k | €1,445k |
| Year 3 | Supplier offers | €360k | €1,585k |
The reservations, stated before deciding
- Margin precision has a limit. Margin is measured on entry into stock, not in the accounting ledger: a gap of around 0.8 points remains and limits every margin figure.
- End-to-end margin is not complete everywhere. For the share of volume transferred to the shop network, margin stops at the transfer price. It is complete for only some of the shops, and one shop’s result cannot be extrapolated to the others.
- The gain assumes adoption. Sound advice that is not followed produces not a single euro. No technology removes that limit.
- Team time is not costed, although the gain depends on it.
What the engagement teaches
- Separate the potential from the prudent basis from the outset. The figure put forward before any access to the data, €0.45m to €1m a year, was confirmed by the measurements. But what can be entered in the budget is less than half of the top of that range.
- Check the scope before quantifying. An initial reading underestimated the share of volume transferred to the shop network by a factor of ten, because each legal entity had not been assigned to the correct scope.
- The measurement most lacking was not a workstream. Consolidated margin, from the price paid to the supplier to the till receipt, did not exist anywhere. It is obtained by combining two sources that were already available, and it is published at fixed intervals.
The next steps chosen by the management team
A practical start: complete the data foundation, launch management reporting as a project in its own right, then test price recommendations on a first product family. Three conditions were set: a dated launch with dedicated resources, a business owner for each workstream with time genuinely set aside, and a target gain and cost fixed before any provider is consulted.
Has your business measured what one point of margin is worth?
We reply in writing within 48 working hours.
