The mix table

One table. Rows are your segments, however you genuinely differ: product line, customer size, channel, region. Columns are revenue this period, revenue last period, margin this period, margin last period, and the share of total each segment represents.

That is the whole instrument, and the reason it is worth a lesson is that most operations do not have it. They have revenue by segment, which is available, and margin in total, which is available, and no one place where the two meet.

What it shows that neither shows alone is shift. A business can hold total revenue and total margin flat while its mix moves substantially toward the segment that will be unprofitable next year, and nothing in either report will say so.

It also settles arguments that otherwise run on anecdote. Everybody has an opinion about which customers are worth having. The table replaces the opinion with a column.

Build it once by hand before automating it. Building it manually forces the definitional arguments to happen, and the arguments are where the value is: which segment does a customer who buys two products belong to, and what happens to shared cost. Automating before settling those produces a table everybody distrusts.