Gross margin by product is known in every distribution business. Cost to serve by customer is known in very few, and it is the more decision-useful number.
Every distributor can rank customers by revenue. Most can rank them by gross margin. Almost none can rank them by what it actually costs to serve them, which is the ranking that should drive commercial decisions.
The gap is not analytical sophistication. It is that the cost data sits in five places and nobody has been asked to join it up.
Everything between the order and the cash that is not the cost of the goods.
The technical build is not the difficult part. Getting the commercial team to accept the allocation basis is.
Our advice is to be conservative and transparent. Use activity drivers people recognise — number of order lines, number of deliveries, number of returns — rather than a sophisticated model nobody can follow. A defensible approximation that sales trusts beats a precise model they dispute.
Publish the drivers alongside the result so anybody can see why an account scored as it did.
Two things, reliably. A group of accounts that look healthy on gross margin and consume disproportionate resource. And a group of small accounts that are quietly excellent because they order predictably and never call.
Neither is actionable in isolation. The point is that the conversation shifts from 'grow revenue' to 'grow the right revenue', which is a different and considerably more valuable discussion.
Not by firing customers. In our experience the productive responses are ordering minimums, delivery consolidation, a change to payment terms, a self-service channel for routine orders, or a straightforward price adjustment with the reasoning explained.
Most customers who are expensive to serve do not know they are, and a proportion will change behaviour when asked.
This is a good first workload for a governed semantic model, because the definitions are contested and the value of settling them is obvious. Build it once, with row-level security so account managers see their own book, and the argument about whose number is right stops.
We will build a cost-to-serve model against one customer segment using data you already have, and show it to your commercial team before you commit to anything.
If this describes your situation, tell us where it hurts most. We will tell you what it would realistically take to fix in your environment, what we would measure, and whether we think it is worth doing at all.
How do I manage IT across multiple retail locations? Centralize your network, POS, and vendor relationships around one standardized store blueprint — so store #50 looks like store #1, just newer. Here's how multi-site retail IT actually works in 2026.
What are the biggest retail IT trends in 2026? Pragmatic, ROI-focused AI, unified data platforms, and ambient in-store technology are reshaping retail — but Gartner also warns 30% of generative AI retail projects will be abandoned. Here's what's actually working.
Available-to-promise calculated from a nightly snapshot is not availability. It is a guess that the sales desk is contractually committing to.
Describe the situation in your own words.
