Real-world application of integrated media planning

In the second quarter of 2020 Airbnb's revenue fell by roughly two thirds year over year. The company let go about a quarter of its staff, shelved its adjacent bets, and did the thing most demand teams treat as unthinkable: it switched off essentially all marketing, including the paid search and paid social that had been feeding bookings for years. Airbnb has put the saving at around $800 million.

Then the bookings came back. Not to 2019 levels, and not to the same places (city breaks collapsed while drive-to and rural stays surged), but they came back with almost no paid mediapaid mediaVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition → pushing them. That accident became the most consequential planning input the company had ever received, and in February 2021 Airbnb told investors it would not rebuild the plan it had before. This lesson stays with that one decision: what got turned off, what the numbers actually said, how the shift was defended to a CFO, and where the read breaks if you copy it. The coordination mechanics are assumed here, in the sense the foundations lesson sets out; the work is in the evidence.

What airbnb turned off, and what came back anyway

The pre-2020 plan was performance-dominated. Sales and marketing ran at roughly $1.6 billion in 2019, close to a third of revenue, and the last large brand campaign before the pause had run back in 2016. Most of the money was doing pursuit work: bidding on category and brand terms, retargetingretargetingShowing ads to users who have previously visited your site or interacted with your brand, to bring them back and drive conversion.View full definition →, feeding an auction that rewarded whoever measured fastest.

When it stopped, Airbnb saw how much of that spend had been buying demand that already existed. Brian Chesky and CFO Dave Stephenson have said publicly that around 90 percent of traffic arrives direct or organic, and that they came to treat marketing as education about what the product is rather than as a mechanism for purchasing customers. The rebuilt plan led with brand: "Made possible by Hosts" launched in February 2021, the first major brand push in about five years, running television and video in a handful of core markets, with performance channels demoted to a defensive and closing role.

The commercial outcome is the part worth holding onto. Airbnb's 2021 revenue reached roughly $6 billion, above 2019, while sales and marketing stayed below its 2019 dollar level, and by 2022 marketing had fallen to under a fifth of revenue. Growth and a lighter media bill at the same time is what made the shift durable internally.

The measurement that defended the shift

Three measurement moves let Airbnb argue the case rather than assert it.

The pause worked as an unplanned global holdout. Spend went to near zero everywhere at once, which produced something no test budget buys: a baseline for how much demand exists with no paid media at all. That baseline reframed every performance channel's reported ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → as a question about incrementalityincrementalityThe share of results (sales, conversions, revenue) that only happened because of a marketing action, not what would have occurred anyway.View full definition → rather than volume.

Second, the defence ran on longer measurement windows. Click-window 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 → reports the last few days of a decision that takes weeks; mix modelling and brand trackingbrand trackingRegular measurement of brand health metrics (awareness, image, preference, and purchase intent) over time, so shifts can be detected and linked to marketing activity.View full definition → (unaided awareness, consideration, search volume on the brand name) put a number on the part that accrues later. Nielsen, which sells media measurement and so has an interest in the argument, has made this point for years: short attribution windows systematically undercredit upper-funnel media because the effect lands after the window closes. Airbnb's pause demonstrated it in reverse, with performance channels losing far less than their dashboards implied they would.

Third, channel roles were rewritten rather than just resized. Branded paid search stopped being an acquisition line and became insurance against competitors buying the Airbnb name. Once a channel's job is defence, you stop judging it on cost per booking and start judging it on how much traffic you lose without it, which is a smaller and far cheaper mandate.

Where the airbnb read breaks

This case gets quoted badly, usually as "brand beats performance, cut your paid budget". Four conditions decide whether the read transfers.

The experiment was confounded. Spend went to zero in the same months that travel was illegal in much of the world and then, when it reopened, in the same months when everyone wanted a house two hours from a city. Some of the recovery was pent-up demand meeting a supply base that happened to suit the moment. Crediting brand equity for all of it overstates the effect, and Airbnb's own longer-window modelling is the honest version of the claim.

Brand strengthBrand strengthThe commercial value your brand adds beyond functional product attributes: the price premium, preference and loyalty it generates.View full definition → was the precondition, not the result. Airbnb is used as a verb and has near-universal unaided awareness in its core markets, built over more than a decade. A brand with single-digit awareness that switches off performance media gets silence, not organic demand. The correct question is not "should we do what Airbnb did" but "what is our direct and organic share today, and what would it be in nine months of no paid media".

Turning off branded search is an invitation. The cheapest clicks you cancel are the ones aggregators and rivals then buy against your name. Airbnb could hold that line because a user typing the brand name mostly persists to the app; a brand whose customers are indifferent between three options will hand over margin the day it stops bidding.

Two audiences, two different plans. Airbnb's media has to recruit hosts as well as guests, and host acquisition in a new market has no organic base to lean on. The parts of the plan facing a cold audience kept looking a lot more like performance buying than the guest-side brand work did.

The second-order consequence is governance. A performance-weighted plan gives a daily read; a brand-weighted plan gives a quarterly one. If your reporting cadence and your incentives do not change with the plan, the brand budget becomes the first line cut in the next downturn, and you rebuild the thing you just dismantled. LinkedIn's B2B Institute, sitting inside a company that sells ad inventory, has pushed a version of the same argument for B2B with its 95-5 framing: most buyers are out of market on any given day, so media has to reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → people who buy later, which only survives if the measurement system can wait.

CMO action items

• Get your direct-and-organic share of sessions and of bookings or signups, by market, before you argue about mix. Airbnb's whole case rests on that one number being high; if yours is low, the argument runs the other way.
• Run a deliberate holdout instead of waiting for a crisis to run one for you. Two comparable geographies, paid social or branded search dark for six to eight weeks, and a pre-agreed read with finance on what counts as a loss.
• Rewrite the mandate of every channel that is currently judged on cost per lead but is really doing defensive work. Branded search, retargeting of existing customers and app reinstall campaigns usually fall in this group, and their targets should be leakage, not acquisition.
• Set the measurement window before the plan, not after. If the plan carries brand spend, the reporting cycle has to tolerate a lag, and someone in finance has to have signed the sentence saying so.

Common mistakes that kill results

Mistake 1: treating a rebalance as a one-way door

Airbnb did not abolish performance media; it lowered its share and changed its job. Teams that read the case as abolition lose the closing infrastructure and then discover that brand campaigns raise intent they can no longer capture.

Mistake 2: cutting spend without building the defence

The pause was survivable because Airbnb could show, quarter after quarter, what unpaid demand was doing. Cutting first and measuring later gives you no evidence when bookings dip for unrelated reasons, and the reflex fix is to switch performance back on at a worse price than you left it.

Mistake 3: copying the output instead of the diagnosis

The transferable part is the method: use an incrementality test to find out what your paid channels are genuinely adding, extend the measurement window until it covers the real decision period, then reassign channel roles. The 2021 split that came out of that process was specific to a category-defining consumer brand recovering from a demand shock, and there is no reason your split should match it.