Budget negotiation: real-world application

Spring 2020: Airbnb's bookings fall off a cliff, and Brian Chesky's later account is that about 80% of the business disappeared in eight weeks. In April the company raises roughly $2 billion in debt and equity, at reported all-in costs in the region of 10%. Cash has a price now. The largest discretionary line on the P&L is marketing, about $1.6 billion of sales and marketing in 2019 against $4.8 billion of revenue. This lesson stays inside that single cut: what came off, what was defended, what the numbers looked like two years later, and which parts of it transfer.

The number that had to move

Sales and marketing was $1,621M in 2019 and about $1.18B in 2020, on revenue that fell from $4.8B to $3.4B. The annual comparison flatters the decision, because Q1 2020 spend was largely committed and spent before travel stopped. The in-year run rate through the middle of the year was far below the annual average, close to nothing in some channels.

That gap matters when you are the one negotiating. A full-year percentage is the wrong unit for judging a freeze. A finance team asking for "27% off the marketing line" in month four is asking for near zero across two quarters, and the cost lines the foundations lesson sets out do not all switch off at the same speed. The April financing set the other half of the arithmetic: every dollar kept in marketing was competing against rescue capital priced in the double digits, so the internal hurdle ratehurdle rateThe blended return a company must earn to satisfy everyone who funds it (lenders and shareholders). It sets the bar every investment has to clear.View full definition → for spend moved overnight, without anyone rewriting a marketing model.

Sub-concept 1: what came off, and in what order

Paid, auction-fed acquisition went first, because it is the only spend a company can stop on a Tuesday and see the invoice fall the same week. Then the adjacencies: Chesky's 5 May 2020 letter paused investment in Transportation and Airbnb Studios and scaled back Hotels and Lux, and cut 1,900 roles, about 25% of the company.

The sequencing logic is worth separating from the total. Two different tests were applied. Media spend was judged on whether output stops when payment stops. The bets were judged on whether payback landed inside the cash runway, which for a company burning through a collapsed booking base meant months, not the three to five years those programmes needed. Headcount was last of the three in speed terms: severance, notice periods and process mean the biggest number takes the longest to arrive. The fastest lever is rarely the largest one, and in a liquidity crunch you are buying weeks as much as dollars.

Sub-concept 2: what was defended

Hosts. Brand and communications. The core stays product. On a spreadsheet, spending against supply while demand is at zero reads as indefensible, and it is the reason there was inventory to sell when demand returned. A marketplace cut has an asymmetry a retailer's does not: starve the supply side and you cannot serve the rebound you are cutting to survive.

Brand was not so much protected as substituted. The Go Near campaign in mid-2020 pushed domestic, drive-to travel, cheap to run, heavily PR-led, and aimed at the only demand that existed. That is the useful distinction inside the case: the cut fell on the part of marketing that buys strangers in an auction, not on the part that keeps the name in people's heads.

Sub-concept 3: why the cut was survivable

Rebuilt from zero, in the way the frameworks lesson describes, most of Airbnb's demand does not have a line item. Roughly 91% of traffic in 2020 came through direct or unpaid channels, per the company's own IPOIPOThe first sale of a private company's shares to public investors on a stock exchange, converting private ownership into publicly traded stock.View full definition → filing. Chesky's public version is blunter: they turned marketing off and got back around 95% of the traffic.

Now the failure mode, because this episode is quoted more often than it is read. It is not a clean experiment. Global travel demand had collapsed, so there was far less worth buying in any auction, and click prices fell with it. Traffic is not bookings. The comparison baseline was already crushed. Nobody can produce the counterfactual where 2020 marketing ran at 2019 levels. What the case supports is narrow and still valuable: a company with a name people type directly into a browser can suspend auction spend for two quarters at bearable cost. It does not support the broader reading that the spend was worthless, and a CMO who repeats the 95% figure without the caveats has handed finance a weapon.

Sub-concept 4: what happened after, and the ratchet

2021 is the part that should worry anyone agreeing to a "temporary" reduction. Revenue passed 2019 levels, about $6.0B against $4.8B, while sales and marketing stayed roughly flat on 2020 at around $1.2B. Brand work came back, including the Made possible by Hosts campaign in 2021, but the paid acquisitionpaid acquisitionVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition → base did not return to its old size. By 2022 marketing was running at under a fifth of revenue, against roughly a third in 2019.

Read that as a second-order consequence of the negotiation, not just an efficiency win. Once a cut has been survived, the lower number becomes the baseline that finance owns, and restoring it needs a fresh business case rather than an appeal to what used to be normal. Airbnb happened to like the new baseline, and the leaner cost structure was part of the story it took into its December 2020 IPO. A CMO in a business with 30% direct traffic instead of 90% would inherit the same ratchet with none of the cushion.

The same shock, a different structure: Expedia

Expedia entered 2020 spending on a different order entirely, roughly $6 billion of selling and marketing on about $12 billion of revenue, the bulk of it flowing into paid channels. It cut hard too, down to something in the region of $2.5 billion, but revenue fell to around $5 billion alongside it. Management said publicly it would come back with a permanently lower, more brand-weighted mix and less dependence on paid search. Spend still climbed with volume as bookings recovered, because a business whose customers arrive through an auction pays again for every one of them.

Same quarter, same instinct, different outcome. The difference is not negotiating skill or discipline. It is where demand comes from, which is decided years before the budget meeting.

CMO action items

• Measure your direct and unpaid share of traffic and of revenue now, not during the crisis. That single ratio decides whether a marketing freeze costs you 5% of demand or half of it, and it takes years to move.
• Convert any annual cut you are offered into an in-year run rate before you accept it, month by month, so you know whether "20% off the year" means near zero for a quarter.
• Name the line you will defend even when its return is unattributable in the current quarter, the supply-side equivalent of Airbnb's hosts, and get the CEO to agree to it in writing before the cut, not during it.
• Write the re-entry case at the same moment you sign the cut: the trigger metric, the amount, the date. Nobody restores a budget out of fairness.

Common mistakes that kill results

Quoting the survivable cut as proof that marketing does not work. Airbnb's 2020 result came with a collapsed market, cheaper auctions, an unmeasurable counterfactual and a decade of brand equitybrand equityThe commercial value your brand adds beyond functional product attributes: the price premium, preference and loyalty it generates.View full definition → behind it. Presenting it to your own board without those conditions invites a cut your business has not earned the right to survive.

Cutting the part that has no near-term attributionattributionA framework for assigning credit to the touchpoints that contributed to a conversion, so you can measure which channels and interactions actually drive results.View full definition → because it is the easiest thing to defend removing. Supply, hosts, service, the community side of a marketplace: none of it produces a clean weekly return, and all of it determines whether you can serve the recovery. The cut that looks cleanest in the model is often the one that caps the rebound.

Treating a freeze as reversible. It rarely is. Assume the number you agree to in April is the number you argue up from for the next three years.