Cross-sell rate = customers who bought an additional product ÷ eligible customers, in a period
The word doing the work is "additional". A different product, not more seats of the one they have.
Cross-sell is not upsell
| Means | Driven by | |
|---|---|---|
| Cross-sell | A different product | Product fit, awareness, and whether the second product solves a problem the first surfaced |
| Upsell | More of the same product: seats, tier, volume | Usage growth and value realised |
Blending them produces a number nobody can act on, because the two respond to completely different work. A cross-sell problem is usually that customers do not know the second product exists. An upsell problem is usually that the first one is not being used enough to justify more.
Get the denominator right
"Eligible customers", not all customers. A customer who already owns every product cannot cross-sell, and leaving them in the denominator drives the rate down as you succeed. That is the trap: the better you do, the worse the metric looks, and nobody can explain why.
Define eligibility per product and recompute it each period.
Bundling breaks it
If the second product ships inside a bundle, a customer who "has" it never bought it. Counting bundle inclusion as cross-sell inflates the rate to meaninglessness; excluding it entirely understates real adoption.
Decide which, write it down, and annotate the series on the day the bundle launched. A metric that changes definition silently at a pricing change is worse than no metric.
A worked example
400 customers. 90 already own both products, so 310 are eligible. In the quarter, 24 of those 310 bought the second product, and 15 more received it in a new bundle.
Cross-sell rate = 24 ÷ 310 = 7.7%
The 15 bundle customers are excluded and reported separately as bundle adoption. Including them gives 39 ÷ 310 = 12.6%, which would credit a pricing decision to the customer success team.
The traps
- Blending with upsell.
- All customers in the denominator instead of eligible ones.
- Counting bundle inclusion as a purchase.
- Counting the same customer twice across periods for the same product.
- Reading it without expansion rate,
which captures the revenue effect this metric does not.
Where this sits
Cross-sell is Measurement in the Tenbound Pipeline Architecture Standard, and it is downstream of Market: customers acquired outside the ICP rarely have a use for the second product, so a persistent cross-sell problem is often a targeting finding.
Read beside net dollar retention and ARPA.