Login Create Account

Basic Customer Lifetime Value

Customer Analytics Full Version →

How much is a customer worth over the months ahead? Answer it with two small models: one for whether they stay, one for how much they spend. Same engine as the full CLV tool, with one marketing lever and no extra dials.

1 ✓
Get Data
2 ✓
Check Columns
3 ✓
Run
4 ✓
Read Results

👨‍🏫 Professor Mode: Guided Learning Experience

Brand new to lifetime value? Turn on Professor Mode for a step-by-step walkthrough.

THE IDEA IN ONE PAGE

A customer is not a sale. They are a relationship that may keep producing sales for months or years. Customer Lifetime Value is what that relationship is worth in today's money.

To get there you only need two questions, asked for every future period:

🔄
Will they be active?

A logistic regression predicting the 0/1 is_active flag. It gives a probability, not a yes or no.

💰
If active, how much?

An ordinary regression fitted only on the periods where the customer actually showed up. It answers how much, never whether.

Multiply them, discount, and add up:

$$\text{CLV} = \sum_{t=1}^{T} \frac{P(\text{active}_t)\;\times\;E[\text{spend}_t \mid \text{active}]}{(1+r)^t}$$
  • \(T\) — how many periods ahead you look (the horizon)
  • \(r\) — the discount rate per period; money later is worth less than money now
💡 Why two models instead of one?

Because "loyal" and "big spender" are different things. A customer can turn up every single month and buy almost nothing. Another can vanish for a quarter and then place a huge order. One combined model blurs those two people together; two models let you see — and act on — each half separately.

🔄 Does a quiet month mean they are gone?

Usually not, and this is the single most common mistake in CLV. In a subscription business cancelling is permanent, so a customer who leaves is gone and the model should write them off. That is called contractual.

In most retail it is the opposite. Someone who skips March is very often back in April. That is non-contractual, and writing off everyone who has a quiet month would badly understate what your customers are worth.

So the tool estimates two probabilities rather than one, and tells you which world your data lives in before it shows you any money:

  • Retention — the chance of being active next period given they were active this period.
  • Reactivation — the chance of coming back next period after having gone quiet.

When reactivation is near zero the maths collapses to the classic survival curve you may have seen in a textbook. When it is not, the tool keeps those returning customers.

🎓 What the full version adds

This basic tool deliberately stops early. The full CLV tool adds:

  • Several marketing levers at once, with saturation (diminishing-returns) terms
  • Separate models per customer segment
  • Customer attributes as extra predictors
  • A what-if simulator with costs, ROI and targeting
  • Churn-risk triage, deciles, and a per-customer spotlight

STEP 1 — GET SOME DATA

📚

Use a Case Study

📤

Upload Your Data

One row per customer per period. Required columns: customer_id, period, is_active (0 or 1), spend. Plus at least one marketing column.

Drag & Drop raw data file (.csv, .tsv, .txt)

Period-level data — one row per customer per period