In Q2 2020, Brian Chesky walked into a room of Airbnb marketing leaders and told them to turn off paid marketing — not pause, turn off. Total marketing spend dropped from $1.62B in 2019 to $545M in 2020, and by Q4 91% of Airbnb's traffic was direct or unpaid. The performance team's worst fear never showed up. This episode unpacks what Chesky was actually testing — brand-search cannibalization, not "brand vs. performance" — and why the math on incremental paid lift gets brutal once your direct-traffic share crosses 70%.

What we cover:

• The real question behind the cut: how much paid-attributed traffic would have come anyway
• Brand-search cannibalization — why it can absorb 60–80% of paid-search budget without anyone noticing
• The three numbers from Airbnb's S-1 that made the call obvious: 90% direct/unpaid floor, $1.08B spend cut, 22% YoY Q4 2020 revenue growth
• Why "incremental ROAS" on last-touch attribution is a vanity number, not an incrementality number
• A three-question framework any marketer can run on their own paid budget — direct-traffic share, holdout-test discipline, six-week paid-off survivability
• Three moves you can make this week: brand-vs-non-brand split, one-week branded-search holdout, taking the result to the CFO before the CMO
• Why the Airbnb lesson isn't "cut performance" — it's that brand strength sets a ceiling on how much incremental work paid can ever do

Closing question we leave you with: if your CFO ran the same incrementality test on your paid budget tomorrow, how much of your spend would survive the room?

Go to marketingcasebootcamp.com for hands-on case simulation practice on this exact business problem — pull a brand-vs-non-brand split, design a holdout, and pressure-test your own paid-search budget the way Chesky pressure-tested Airbnb's.

If this one helped, share with other marketer folks.

Read the full written case on Marketing Case Bootcamp.