Take rate is a product decision

On a $100 booking, a 9% take rate keeps $9. Move from 6.8% to 8.9% on the same bookings and you keep about 31% more revenue. On $1 billion of bookings, each percentage point of take rate is worth $10 million.

The reflex is to ask partnerships for another commission point.

But contracts only set the terms available to you. Your product decides which supply gets booked, which discount gets paid, what appears first and what happens when a booking goes wrong. Take rate is a product decision. The team building those screens should own the number, with conversion beside it.

All figures for the hotel example and its experiments are invented.

First, agree on what you are measuring

For this essay, net take rate = revenue retained after platform-funded discounts, refunds and fee reversals, divided by the room value of confirmed bookings. The booking value excludes taxes and stays in the denominator even if the guest cancels later. That lets us see what cancellations cost the original booking cohort.

Public marketplaces use different definitions. Some report gross booking value after cancellations. Others divide revenue by the value of completed orders. Even two sensible percentages can describe different things.¹

Document one definition and stick to it. Take rate measures revenue retained, not profit. Payment fees, support and marketing still have to be paid.

The ownership problem

In many marketplaces, partnerships negotiates the commission. Finance reports the net take rate at month-end. Product reviews searches, clicks and bookings.

Each team can hit its target while the marketplace keeps less money.

A supplier offers a good commission, but that rate barely appears in results. A coupon lifts conversion, but funds bookings that would have happened anyway. A refund wipes out a fee nobody tracks.

Those are product choices. Four surfaces matter most.

1. Supply mix: which offer do you actually sell?

A hotel can reach your marketplace through more than one supplier. The room may be identical while the commission, available rate and cancellation terms differ.

When your catalog treats these as separate rooms, guests see clutter. Ranking cannot compare equivalent offers. The better-margin option may disappear.

Clean hotel and room mapping makes supply mix a choice. Put genuinely equivalent offers together, compare what the guest pays and select the supplier that leaves more value without degrading the stay.

Higher commission can still lose money if it brings worse prices, unreliable availability or harsh cancellation terms. Fix the data first. Then test the mix.

2. Pricing: whose money is paying for that deal?

A $10 discount can be funded by the hotel, by the marketplace or by both. The guest sees the same price. The amount you keep can be very different.

Did that discount change the decision? A coupon offered at checkout might only thin your margin.

Test supplier-funded offers against your own coupons. Keep a comparable group undiscounted. Measure extra bookings, not coupon redemptions.

An October 2026 filing from a large delivery marketplace cited lower discounts and promotions as one factor supporting its revenue margin.²

3. Ranking: the shelf is part of the deal

Search results decide which commissions you earn. A beautifully negotiated rate on page six may as well live in a drawer.

That makes ranking tempting. Sort by the highest margin and the spreadsheet looks better immediately.

Guests also care about price, location, reviews and availability. Show the wrong options and you win a bigger share of fewer bookings.

A better test gives margin some weight among offers that already meet the guest's needs. Keep relevance and price competitiveness in the model. Track booking conversion as closely as the take-rate lift.

4. After booking: what you thought you kept

A booking confirmation is an optimistic moment. The guest can cancel. The hotel can refuse the room. A payment dispute can arrive weeks later.

When the marketplace reverses its fee or pays for a service failure, some of that expected revenue disappears. The original booking still sits in our denominator, so the loss shows up plainly.

Better availability checks, clearer terms and faster resolutions can cut avoidable refunds. Give each booking cohort enough time to cancel or complete its stay before assessing results.

What the four levers can add up to

The hotel marketplace starts at a 6.8% net take rate. Better supply matching adds 0.6 percentage points, smarter promotion funding 0.8, balanced ranking 0.5 and fewer fee reversals 0.2.

That reaches 8.9%, a gain of 2.1 percentage points. On unchanged booking value, retained revenue rises 30.9%. The extra revenue would be $21 million on $1 billion of bookings. None of that is a forecast. These effects need to survive a combined test before anyone puts them in a budget.

The first four tests explain the proposed lift. The others reveal the trade-offs.

Conversion changes are relative to each test's control group. “pp” means percentage points. These are separate tests; their effects may overlap.

Put a floor under conversion

Suppose the team agrees before testing that booking conversion must not fall by more than 1% relative to control. If conversion starts at 5%, the floor is 4.95%. The balanced-ranking test passes at a relative decline of 0.4%. The commission-only sort fails at a decline of 3.8%, even with a larger take-rate lift.

That floor is a business choice, not a universal benchmark. It protects selection, supplier relationships and the repeat customers who make a marketplace useful.

Keep one commercial measure above both: net revenue retained per eligible search. It catches cases where a higher take rate comes from fewer bookings or a change in average booking value. Check contribution profit too. Retained revenue can rise while costs eat the gain.

Put it on the weekly product dashboard

Show net take rate beside conversion and retained revenue per search. Break it down by supplier, property group, traffic source and promotion type. Include refunds from older booking cohorts. Review experiments with partnerships, finance and product in the same room.

On Monday, ask four questions:

1. Which supply did we show when two suppliers had the same room?
2. Which discounts changed a booking decision, and who funded them?
3. Did our ranking tests keep more revenue without crossing the conversion floor?
4. How much of last month's expected revenue came back out in refunds or reversals?

If take rate only appears in your finance review, move it to the product dashboard, next to conversion.

---

Try the tool: Marketplace vs Direct Calculator