Marketplace Take-Rate Model (Assign vs. Assist)
Whether a marketplace can charge Uber-level take rates or is capped at Airbnb-level take rates depends on who the customer thinks they're buying from.
- Difficulty
- Easy
- Time to result
- ~ongoing to results
- Steps
- 4
- Confidence
- 85%
A decision framework for predicting the sustainable take rate of a marketplace business by classifying it as 'marketplace assign' (homogenous supply, platform picks the specific provider, customer feels like they bought from the platform) versus 'marketplace assist' (differentiated supply, customer actively chooses the specific provider, customer feels like they bought from that provider). Assign-type marketplaces like ride-sharing can command 25-30%+ take rates because the platform is the perceived merchant of record; assist-type marketplaces like Airbnb are structurally capped at lower take rates because the host is the perceived merchant, and the platform is 'just helping' facilitate the transaction.
Origin
Articulated by Ben Gilbert during the Airbnb episode's Power discussion as a way to explain why Airbnb, unlike Uber, will likely never reach 30%+ take rates.
How to run it
- 1
Assess supply homogeneity
Determine whether units of supply are functionally interchangeable (a ride, a commodity item) or meaningfully differentiated (a specific home, a unique product).
- 2
Identify who is the perceived merchant
Ask whether the end customer mentally attributes the purchase to the platform itself or to the individual supplier being showcased on the platform.
Pro tip Run a simple test: would a customer say 'I got an Uber' or 'I stayed at John's house'? The phrasing reveals the perceived merchant.
- 3
Classify as assign or assist
If the platform can auto-assign any interchangeable unit of supply to fulfill demand, it's 'marketplace assign.' If the customer browses and selects a specific, differentiated listing, it's 'marketplace assist.'
- 4
Set realistic take-rate expectations
Benchmark expected long-term take rates against comparable assign or assist marketplaces rather than assuming your category can converge on the highest known take rates in the industry.
Pro tip Use this to sanity-check founder or analyst projections that assume take-rate expansion beyond what the category structurally supports.
Watch out Don't assume that adding more fees on both sides of the transaction will overcome the structural ceiling set by perceived merchant identity.
In the wild
Because any nearby driver can fulfill a ride request and the experience feels identical to the rider, Uber is perceived as the actual seller of the ride, letting it command a much higher take rate.
→ Ride-sharing companies can sustain take rates above 25-30%.
Guests actively browse and choose a specific home, so in the customer's mind the real seller is the host, not Airbnb, even though Airbnb facilitates payment and trust.
→ Airbnb's blended take rate stays around 13-15%, well below ride-sharing-style marketplaces.
Common mistakes
Assuming all marketplaces converge on similar take rates
Projecting that an assist-style marketplace will eventually reach assign-style take rates just by adding more fees or features.
Ignoring customer perception of merchant identity
Focusing only on unit economics and ignoring the psychological ownership of the transaction, which structurally caps pricing power.
Is it for you?
Not ideal for
Single-sided businesses (pure SaaS, direct retail) where there's no marketplace intermediation to classify.
From the transcript
“something like Uber is marketplace assign versus something like Airbnb is marketplace assist.”
“they're never gonna get to that 30 plus percent um that you see in like ride sharing where there's you know people feel like they're…”
From the episode
Airbnb