Revenue was up and net profit was down, the quiet killer of a volume business. The company could not see its cost-to-serve: the customers taking big discounts, demanding small express drops, returning goods and paying late cost more than they brought in, while reps kept giving blanket discounts to hit their targets.
The company could not attribute its indirect costs, fuel, warehouse hours, payment fees, the time customer service spent, to a specific account. So it had no idea which customers actually made money, and it was quietly nurturing the ones that lost it.
Pricing ran on gut and spreadsheets. Discounts lived in Excel, and a rep would give the maximum just to close a deal, with no view of what it did to the bottom line. Every "yes" chipped at a margin no one could see.
Promotions were just as blind. Trade budgets and retailer incentives were poured out with no way to tell whether a promotion drove real extra sales or simply pulled forward demand the company would have won anyway.
Make per-customer profit visible, quote to a floor, and let a human approve only the exceptions.
We pulled data from the warehouse, transport and finance systems and built a model that computes, per customer, what it actually costs to serve them, unloading time, service calls, credit days, and reveals their true profitability.
When a rep builds a quote, the engine weighs that customer's cost-to-serve, the stock cost and the market, and returns an optimal target price that holds a preset margin floor, in real time.
The system auto-approves only discounts that still meet the margin model. A request that breaks it is escalated straight to the trade manager with a simulation of the loss it would cause, so exceptions are a human call, not a reflex.
Every trade promotion is measured for the extra sales it genuinely drove versus demand simply pulled forward, so the promotion budget goes where it actually grows the business.
Flagging the loss-making customers let the company change how it serves them, from small daily drops to a weekly consolidated delivery, turning a loss into a profit without losing the relationship.
Reps and trade staff get data-backed, transparent pricing instead of an argument, so the negotiation is with the customer, not with head office.
The volume stayed; the profit came back.
Gross margin widened about 4.5% within two quarters without losing a single strategic customer, because the gains came from pricing to true cost, not from walking away from accounts.
Identifying the loss-making customers let the company reshape their service model and turn 80% of them profitable, while data-backed pricing cut internal discount arguments 50%.
Volume businesses bleed margin because the true cost of each customer is invisible and discounting is a reflex. The value here is allocating real cost to each account, quoting every deal to a floor, and sending only genuine exceptions to a person with the loss spelled out. It is the same margin-guardrail discipline behind our dynamic-pricing work, aimed at per-customer cost-to-serve in B2B distribution rather than the retail shelf.
A retailer's promotions quietly sold at a loss. A hard 12% margin guardrail blocked loss-making promotions before they ran and lifted gross margin about 4%, without a pricing analyst in the loop.
Read the case → Portfolio & cost · Consumer goodsAn over-extended catalogue of 1,500+ SKUs hid its true cost to serve. AI cost analysis cut the range 25% with no revenue lost, dropped line downtime 35%, and freed working capital tied up in surplus stock.
Read the case →
Reef TRH