Revenue concentration in distribution means a large share of sales sits on a few accounts, usually wholesalers. Total volume can look fine while the network narrows. Didrah’s AI agent sees this on invoices: the share of the top three or five customers in one region, one period, one brand or the whole book. If that share crosses a threshold you set, the claim should be a warning, not a growth toast.
Why a wholesaler is both reach and risk in the distribution supply chain
A wholesaler lifts coverage fast and pulls stock out of your warehouse. If that account stalls, weighted coverage drops overnight, and you learn late because you do not see the retail secondary sales behind them. Customer churn in distribution is not always announced. Sometimes the gap between invoices just grows. The agent should read an active customer by last-invoice gap, not by an “active” tick in master data.
Which rule an AI agent uses to score concentration on invoices
The rule must stay fixed: the same top-N, the same period, the same item scope. Comparing “top three this year” with “top ten last year” is a contest of definitions. Agentic AI earns its keep by running the same definition every week and pinning the claim to rows. If customer identity on the invoice is inconsistent, it should say data quality, not invent a concentration percentage.
After a concentration warning, which distribution action is rational
The action is rarely to cut the wholesaler. It is to open a retail beat in the same region, cap credit on that account, or see that a promotion only reached those few points. The autonomy ladder is strict here: an agent must not execute a cut to a large wholesaler. The approval gate belongs to sales and finance. The agent should only say what share of weighted coverage disappears if that account stalls.
What concentration question to ask the supply-chain agent
- Ask for the top-three share every week on the same definition.
- Read concentration apart from volume growth; both can rise together.
- If a wholesaler is holding weighted coverage, check numeric coverage on its own.
- Do not climb past the propose rung on cutting a large account.
Where a concentration warning sits in weekly distribution operations
Do not just read concentration next to volume on the same page. Join it to numeric coverage and the last-invoice gap on those same three accounts. If share rose and the invoice gap also grew, the risk is an account stalling, not “loyalty.” The AI agent should put those three in one claim. That week’s action is usually a credit cap or two new retail points, not a breakup meeting. If the warning ran four weeks and no cap was set, the autonomy ladder is not the problem. The human decision was not taken.