AI Hints
Home›Data Science Tutorial›Deciding Under Uncertainty
Data Science · Lesson 53 of 101

Payback Time

When does a customer pay off?.

⏱ 2 min read▶ Video at 1:30:18🎓 Free
Watch this lesson in the Data Science courseWatch on YouTube ↗Full Data Science course

It costs a hundred and thirty dollars in marketing to win a new customer. When do you get that back? If it takes three years, and most customers leave within one, then you've got a business that loses money with each new customer. Many fast growing startups have died from exactly this.

Payback Time
From the video: Payback Time
  • $130 to win a customer
  • When do you get it back?
  • Startups die from this

Working out payback

Each month, a remaining customer brings in twenty two dollars of margin. We add that up, month by month, while some of them leave. We try three churn rates. Run it. At three percent churn, it pays back in seven months. At ten percent, it takes nine months. A rule of thumb from investors is that twelve months or less is healthy. So all three are fine here. But what's the point? It's that this is a simple while loop, and it can tell the founder how much it's safe to spend on marketing.

payback.pypython
margin_per_month, cost_to_win = 22.0, 130.0

for churn in [.03, .06, .10]:
    alive, profit, month = 1.0, 0.0, 0
    while profit < cost_to_win and month < 60:
        month += 1
        profit += alive * margin_per_month
        alive *= 1 - churn
    print(f"Churn {churn:.0%}: pays back in {month} months")
Output
Churn 3%: pays back in 7 months
Churn 6%: pays back in 8 months
Churn 10%: pays back in 9 months

Unit economics

These numbers together are called unit economics, and they're the first things that an investor looks at. Customer acquisition cost is what you pay to win someone. Lifetime value is what they're worth. A common goal is a lifetime value of at least three times the cost of winning them. Payback time tells you how much cash you'll need to grow.

keyvalue
CACcost to win a customer
LTVwhat they're worth
LTV to CACaim for 3 or more
Paybackunder 12 months

Summary

Know what a customer costs, and when they pay you back. Next, back to data quality, starting with surveys.

Key takeaways
  • Add margin month by month
  • Churn slows payback
  • LTV should be 3× CAC
Next up: Fixing a Biased Survey — who actually answered?