Split by Segment
Find which customer segment explains a change in performance. Reconcile weighted totals, test mix effects and avoid counting the same sale twice.
- By customer type
- By region
- By channel
- By product
Watch out: Use the cuts the client already reports; ask which ones they track.
On a case
A bank's credit card spending fell 10%.
- Salaried or self-employed
- Metro or smaller city
- Online or in store
Illustrative case. Figures and conditions below are assumptions for practice, not market data.
Assume last month's card spending was Rs 60 crore from salaried customers and Rs 40 crore from self-employed customers. Salaried spending remains Rs 60 crore; self-employed spending falls to Rs 30 crore. Total spending falls from Rs 100 crore to Rs 90 crore, matching the stated 10 percent decline.
Now split the self-employed segment by geography or channel to locate the change. Do not add a metro subtotal to an online subtotal: the same transaction may appear in both. Use one mutually exclusive segmentation at a time.
For mix effects, suppose two groups spend Rs 1,000 and Rs 2,000 per active card. Equal shares produce a Rs 1,500 average; a shift to 80 percent in the first group lowers it to Rs 1,200 with no change within either group. Ask whether customer mix, active cards or spend per card moved.
Sources and further reading
- US Small Business Administration: market research
Questions for testing customer demand, market size, location, saturation and competing prices.