In my Marketing Budget Experiments project development, I segment customers in two different ways.
- New customers stay in a cohort through the rest of the year. They stay in the cohort in the subsequent calendar year as well.
- Existing customers are modeled via regression models, and are split into twenty-five (25) segments of varying levels of customer quality.
Now, each marketing channel appeals to a different audience. Paid Social might appeal primarily to prospects, but not entirely to prospects. If you increase Paid Social spend by 50%, you will mostly generate new customers, but some existing customers will purchase as well.
This is where our Experiments become interesting.
With new customers, we balance the cost of acquisition with CLV ... customer lifetime value. Most clients look to be paid back within about 7-12 months. Most of the work I've performed suggests the payback window should be more than 7-12 months ... your mileage will vary.
With existing customers? Spicy nacho dip. Here you are looking at ICLV ... or "Incremental Customer Lifetime Value". When you take a customer who had Recency = 7 month and Frequency = 4 purchases and you convert the customer to Recency = 1 Month and Frequency = 5 purchases, you change ICLV (incremental customer lifetime value). The customer might have been worth $30 of profit in the next five years and the customer is now worth $45 of profit in the next five years. The difference ... $45 - $30, is ICLV. The marketing effort added $15 of CLV to this existing customer.
It turns out that ICLV plays just as big a role as CLV in determining if the long-term impacts of marketing outweigh the short-term cost of marketing.