OKR Examples for Founders: The Intel Goal Sheet That Started It

An Intel goal sheet from the 8086 years fits on one page.

Establish the 8086 as the highest-performance 16-bit microprocessor family.

Four key results sit under it: develop and publish five benchmarks showing superior 8086 family performance, repackage the entire family, get the 8MHz part into production, and sample the arithmetic coprocessor by June 15.

That page is in John Doerr's own example library, sourced to Measure What Matters. It is also the shape of every OKR you've filled in since, minus the part that made it work.

The summer intern who wrote the benchmarks

Doerr arrived at Intel as a summer hire. He cold-called his way into Bill Davidow's office, and by the end of the summer he was writing benchmarks that showed Intel's chips were faster than Motorola's.

Andy Grove told him something that summer he has repeated ever since. It almost doesn't matter what you know at Intel. It's execution that's everything.

Management by objectives was already standard practice. Drucker had written about it, Hewlett-Packard ran on it. Grove's version, which Doerr calls Intel's MBO, changed the settings.

Quarterly instead of annual, public instead of private, bottom-up instead of top-down, and separate from your bonus.

Public is the setting small teams skip, and it's the one that does the work. Doerr cites Box CEO Aaron Levie's estimate that around 30 percent of people at any company are working on the wrong things, and that it's hard to know which ones. You can't see the misalignment when five people hold five private goal docs.

1999, in a garage

Doerr took the slide deck to Larry Page and Sergey Brin in 1999. They were 24, working out of Susan Wojcicki's garage.

Sergey's answer, as Doerr tells it: we don't have any better way to manage this company.

Google has run OKRs every quarter since, and the system has since been adopted by the Gates Foundation, Dropbox, Spotify, Samsung and BMW, among others.

Where they actually die

Doerr names the first failure mode himself. The system fails most often when the leader doesn't personally commit to it.

The second one shows up in almost every team between five and fifty people. The goal survives the quarter and dies in the week.

It gets set in January. February is a blur of shipping, hiring and fires. By March nobody can reconstruct why those four key results were chosen, so nobody argues with them, and the number at the top becomes decoration.

The Intel page worked because it recorded the reasoning: why the 8086, why those four measures, why June 15. A reader in October could see the decision, not just the target.

The test

Take the goal you're on this quarter. Write one sentence under it: what has to be true for this to matter?

If that sentence needs three tabs, two docs and a phone call to answer, the goal is already gone. The page that survives is the one that kept its own reasoning.