Microsoft Handed Yahoo 88% of Search Revenue in 2009 — and Wrote the Step-Ups Into the Contract

In July 2009, Microsoft did something that looked, on the surface, like a company throwing money into a hole. It agreed to run search and search advertising on Yahoo's websites, and to pay Yahoo 88 percent of the revenue those searches produced.¹ Eighty-eight cents of every dollar Bing earned on Yahoo's turf went straight back to Yahoo. Microsoft kept twelve cents, ran the infrastructure, and ate the cost. The framing was David-versus-Goliath: this was the deal that would finally give someone a fighting chance against a Google that already owned more than 70 percent of the market.¹ The headline everyone kept was simpler and crueler — Microsoft overpaid, badly.

The tidy version that survives is that Microsoft handed over 88 percent in a moment of weakness, watched it hurt for six years, and then in 2015 finally had the leverage to renegotiate it down. Almost every beat of that is wrong. Microsoft did not claw the 88 percent back — because the number was never meant to stay at 88. The step-ups were already in the contract.

Here is the thesis a smart friend can repeat at dinner: the 88 percent was not a panicked overpay Microsoft later corrected. It was a self-amortizing loss-leader with the rate schedule written in from day one, and the famous 2015 'renegotiation' changed how many queries Yahoo had to send and whether Microsoft was exclusive — not the share Microsoft paid.

The price paid for scale that Bing couldn't buy any other way: a search engine is worthless without query volume, and Yahoo had a block of it to sell

To understand why any sane company gives away 88 cents on the dollar, you have to understand what a search engine actually sells. It doesn't sell answers. It sells the loop between advertisers and queries: more searches produce more data, better ad matching, higher revenue per search, which funds a better engine, which attracts more searches. Below a certain volume, that loop doesn't spin at all. In 2009, Bing was below that line and Google was miles above it. Microsoft could not buy queries at retail — it had to acquire a whole population of them at once. Yahoo, still one of the largest destinations on the web, was exactly that population, sitting there for sale. So Microsoft paid what the queries were worth to a company trying to reach ignition: nearly all of the near-term revenue, in exchange for the scale that makes the long-term revenue possible. That is the definition of a loss-leader — you lose on the transaction to win the customer, or in this case, the volume.

The economics on Yahoo's side make the trade legible. Yahoo told analysts the arrangement would add about $500 million a year to operating profit and save roughly $275 million a year in capital expenditure by letting it stop building its own search technology altogether.⁵ Yahoo was, in effect, selling its search business for a rich slice of the revenue and firing the engineers who used to run it.⁵ Microsoft was buying a search business's worth of query flow and paying for it out of the very revenue that flow produced. Neither side needed to be foolish for 88 percent to make sense; it was the clearing price for turning two subscale search operations into one that might reach escape velocity.

The rate was always going to climb — Yahoo's own filing says so: the step-up to 90% or 93% was baked into the 2009 contract, not won in a later fight

This is the part the popular retelling gets exactly backwards. The 88 percent was contractually fixed for only the first five years. After that, the original December 2009 agreement already scheduled the rate to change — and mostly upward. Per Yahoo's own FY2010 10-K, once year five passed the Revenue Share Rate was set to become 93 percent if Microsoft took over exclusivity for selling to premium advertisers, 90 percent if it did not, or 83 percent only in the single case where Yahoo chose to keep its own ad-sales exclusivity.² Read that again: the only path to a lower number required Yahoo to give something up. The default trajectory ran from 88 to 90 to 93. There was no clawback to negotiate, because the escalator was already in the elevator shaft.

Meanwhile the deal was doing precisely what a working loss-leader does: growing into the base it was subsidizing. Revenue tied to the Microsoft search agreement rose from roughly 20 percent of Yahoo's total revenue in 2011 to about 25 percent in 2012 and around 31 percent in 2013.⁴ For Yahoo that was a lifeline becoming a dependency. For Microsoft it was the query volume compounding on exactly the schedule the contract had priced in. The 88 percent wasn't bleeding out; it was buying a growing share of the search market's mind, one that let Bing climb above 20 percent of U.S. search by early 2015 even as Yahoo's own brand share slid to 12.7 percent.⁷

What actually changed in 2015 was volume, not price: the nominal rate went up to 93%, and Yahoo's guaranteed queries got cut roughly in half

On April 15, 2015, the two companies signed the Eleventh Amendment, and the press coverage reached for the obvious frame: Yahoo finally getting free of a bad deal. Look at the mechanics and the frame collapses. The nominal Revenue Share Rate did not fall — it rose to 93 percent.³ What made that increase less than it appears is a change in the base it was calculated on: the earlier 88 and 90 percent rates were computed after deducting affiliate sites' share of revenue and Microsoft's costs, while the new 93 percent was calculated before deduction of the affiliate's share.³⁶ A higher rate on a differently defined base can wash out to roughly the same money. The one thing the 2015 amendment did not do was make the deal cheaper for Microsoft on paper.

The substantive change was not economic at all — it was structural, and it moved in Yahoo's favor on the one axis that mattered to a CEO who wanted independence. Microsoft went from Yahoo's exclusive PC search provider to a non-exclusive one, and Yahoo's mandatory routing of desktop queries to Bing dropped from effectively 100 percent to a 51 percent 'Volume Commitment' starting May 1, 2015.³ Yahoo was suddenly free to send up to 49 percent of its desktop search ads through its own Gemini platform or other partners.⁸ The amendment also let either side terminate the ten-year deal on or after October 1, 2015 with four months' notice.⁶ Marissa Mayer got autonomy from the arrangement her predecessor Carol Bartz had signed.⁷ What she did not get — and what nobody was actually fighting over — was a discount on the 88-going-on-93 percent.

Wasn't Microsoft just cornered into a lousy deal it later escaped?: the honest counter is that 88% is a brutal number — but brutal is what buying scale costs

The fair objection is that this reads too kindly to Microsoft. Eighty-eight percent is a punishing giveaway, and a company confident in its position doesn't hand a rival nine of every ten dollars. True enough — and worth sitting with. But two facts push against reading it as a blunder. First, if the rate were the mistake, the correction would have been to cut it; instead the scheduled path ran upward, and Microsoft accepted that in 2009 with eyes open.² Second, the 2015 change targeted volume, not price — precisely what you'd renegotiate if the problem was never the rate but the exclusivity lock that kept Yahoo from wandering.³ The honest concession is that a loss-leader can still be a bad bet if it never reaches ignition, and Bing never toppled Google. But 'we bought scale at a steep price and it stalled short of the leader' is a very different story from 'we panicked, overpaid, and clawed it back.' Only the first one is what the contract and the filings actually describe.

Microsoft's 88 percent was a bet that you could rent your way into a market that punishes anyone who arrives small. It paid away nearly all the revenue to buy the one thing money alone can't manufacture on a two-sided network — volume — and it wrote the price to rise as the volume matured. The deal that everyone remembers as a humiliating overpay quietly corrected the record in Yahoo's own filings: the rate went up, not down; the fight in 2015 was over freedom, not money. The lesson isn't that Microsoft got a great deal. It's that a giveaway with a schedule attached is a strategy, and a giveaway remembered without its schedule is just a story people prefer to tell.