On April 24, 2025, Intel’s new CEO, Lip-Bu Tan, sent a memo to the company titled “Our Path Forward.” Most of it was about layoffs, fewer meetings and a return to the office. One sentence was about goals: “In addition, I have decided to make our formal Insights and OKR requirements optional.” The next sentence kept the point: “While it’s crucial for us to stay accountable for our results and receive feedback on our performance, I believe we can achieve this in a simpler and more flexible way.” [1]

The memo did not say OKRs had failed. It categorized them as things that “do not move our business forward.” [1] That is a strange place for them to end up, because Intel is where OKRs were born.

A very brief history

Peter Drucker introduced Management by Objectives in 1954. He defined objectives as a tool of self-control for the manager, “not a tool of control from above.” [2]

Andy Grove brought the idea to Intel in the 1970s and added a key result: a measurable test of whether the objective was met. As John Doerr tells it in Measure What Matters, Grove wanted a key result to be binary: “Did I do that or did I not do it? Yes? No? Simple. No judgments in it.” [3]

Doerr carried the system to Google in 1999. [3] Google still uses it.

What OKRs are supposed to produce

When a company adopts OKRs, it is usually looking to level up a few different things:

  1. Focus on what matters
  2. Alignment across the organization
  3. Accountability that can be measured
  4. Turning information into insights and insights into organizational learning

OKRs produce another thing, and very few people put it on the slide. A goal system drives culture and behaviors. OKRs are a very efficient way of telling a few thousand people what gets noticed. They will learn to adjust their behaviors to the system.

Where have OKRs worked?

The honest answer is shorter than the OKR industry would like.

The strongest evidence of effective outcomes has nothing to do with OKRs. In 1991 Robert Rodgers and John Hunter pooled 70 studies of Management by Objectives, the system OKRs descend from. Sixty-eight showed productivity gains. When top management was strongly committed, the average gain was 56%. When commitment was low, the average gain was 6%. [4]

A randomized field experiment points the same way. Researchers randomly assigned 138 branches of a distribution company to conditions; branches given goals plus benchmarking improved sales relative to controls. [5] Again, this is goal setting, not OKRs specifically.

Then there are the famous stories. In 1980 Intel ran Operation Crush with the objective “Establish the 8086 as the highest performance 16-bit microprocessor family,” and a target of 2,000 design wins. By 1986, the 8086 family held about 85% of the 16-bit market, and the 8088 went into the first IBM PC. Bill Davidow’s account notes that “not one Intel product was modified for Crush.” [6] It is a great story. It comes from an insider, in a book by OKRs’ chief advocate, and he credits OKRs.

Google adopted OKRs with roughly 40 employees and still used them past 60,000. [7] That proves Google kept them. It does not prove they made Google. Cagan says it bluntly: successful companies “aren’t successful because they use OKR’s.” [8]

So the case for goal systems is strong, and the case for the OKR brand in particular is thin.

Why OKRs fail in so many companies

Start with the size of the problem. Donald Sull and colleagues found that two-thirds to three-quarters of large organizations struggle to execute their strategies. [9] In one MIT Sloan study, only 28% of executives and middle managers could name three of their company’s five strategic priorities. [10]

The failure modes are familiar to anyone who has lived through an OKR rollout.

  1. Cascading numbers game.
  2. Key results that are really a to-do list.
  3. Sandbagging.
  4. Too many objectives.
  5. Set and forget.
  6. No link to strategy.

Some might interpret that list as a list of people problems. It is not. Each item is a predictable output of the systems as they were built.

Zach Saw was a junior engineer at Intel in 2003. Twenty years later, he wrote about how he gamed the OKRs there. [11] They were tied to performance reviews, which meant pay and promotion. So he picked visible, easy tasks, quantified trivial work, piled up key results, and avoided hard, long-term problems. Teams competed with each other and with other departments. “We weren’t lazy,” he wrote. “We were responding rationally to the incentives.”

That is one person’s memory, two decades on, so weigh it accordingly. But it is exactly what the research predicts. And pay is only one way the system shapes behaviors.

In Out of the Crisis, Deming estimated that 94% of problems are system-related, management’s responsibility, and 6% are special causes. [13] That figure is his judgment from experience, not a measurement. At a seminar in 1993, he reportedly put it more sharply: [12]

“A bad system will beat a good person every time.” - W. Edwards Deming

When teams are handed objectives and a fixed roadmap at the same time, the key results end up being decoration. And when OKRs are set on a semi-annual or quarterly basis, anytime the roadmap flexes to adapt to new insights or a changing ecosystem, the actual work and the documented objectives become disconnected.

A company that runs OKRs as a separate, graded reporting ritual will eventually notice it is paying for a ritual and getting very little return on the investment. Making it optional, as Intel did in 2025 [1], is a reasonable response. It’s not the only one.

OKRs fail because they either start disconnected from what actually matters for the business or quickly become disconnected. The faster things change, the faster they become disconnected. But if the disconnected system is still tied to real-world incentives, it will drive behaviors that diverge from what is best for the organization.

What to Do About That

If the system produces an undesirable behavior, fix the system. Training people to write better key results won’t fix a system that rewards gaming. The fix comes in three parts.

1. Goal orientation needs to be woven into the daily operating model, not treated as a separate process.

First, fix where goals live. In most companies, they live in a tool you open at the start of the quarter and again at the end. The real business runs somewhere else: the weekly operating review, the pipeline meeting, the hiring plan, the budget.

In practice, this means that relevant goals are revisited at the top of the agenda of the meetings that already decide things. If a key result never changes a staffing call, a budget line, or a priority, it is a report, and people will treat it as one. This answers set-and-forget, the to-do-list key result, and the leadership gap in one move.

2. Goals should be set and tuned at a frequency that is aligned with what makes sense for each type of goal and the rate of inflow of new information and dynamic changes in the company’s ecosystem.

The second thing to fix is the clock. Most OKR programs run every goal on the same quarterly cycle. That is convenient for the calendar and wrong for the work.

Some goals should move weekly, like a support backlog or a conversion rate. Others, like building a new capability, need a longer horizon and should be judged by what you learned. Gerard Seijts and Gary Latham found that in new or complex work, specific performance targets can actually hurt, and learning goals work better. [14] One cadence for every goal guarantees that some goals are tuned too often and others too rarely.

A target set in January and defended in November is a sandbagging machine. Tuning goals when new information arrives, rather than when the calendar says so, removes most reasons to sandbag and keeps goals tied to the market the company is actually in.

3. Goals should be identified, shaped, and elevated based on the business’s highest points of leverage and most constraining bottlenecks. “Theory of Constraints”

The third thing to fix is aim, and it matters most.

Eli Goldratt’s The Goal makes a simple claim: every system has a constraint that limits its output, and improving anything other than the constraint mostly wastes effort. His five focusing steps are to identify the constraint, exploit it, subordinate everything else to it, elevate it, and then start again without letting inertia become the new constraint. [15]

The results can be large. Victoria Mabin and Steven Balderstone reviewed more than 80 published Theory of Constraints applications and found average lead-time reductions of 69%, inventory down 50%, and revenue or throughput up 68%. [16] Treat those numbers with care; they come from published successes, and the authors found no reported failures to compare them with. The direction is still clear.

Now look at a typical OKR set through that lens. Every team has three objectives. Each team’s objectives focus on its own function. Nobody has asked where the company is actually stuck. That is how a company ends up with too many OKRs, goals disconnected from strategy, and teams competing to look good. Each team optimizes its own number, and the constraint sits between them, owned by no one.

Aiming goals at the constraint changes what the system teaches. A team whose key result is to relieve another team’s bottleneck learns that helping across the line is the job. Fewer goals become possible because most of the company’s goals can be subordinate to the one that matters this quarter. And when the constraint moves, which it will once it is relieved, the goals move with it, which brings us back to cadence.

None of this requires dropping OKRs. Intel’s memo kept accountability and feedback and dropped the formal requirement. [1] A company can keep the form and change the system around it: goals inside the operating rhythm, on a clock that fits each goal, aimed at the place the business is stuck. That is where the 56% lives. [4]

References

  1. Intel, “Lip-Bu Tan: Our Path Forward,” April 24, 2025. https://www.intc.com/news-events/press-releases/detail/1738/lip-bu-tan-our-path-forward
  2. Peter Drucker, The Practice of Management (1954). https://archive.org/details/practiceofmanage00druc
  3. John Doerr, Measure What Matters (Portfolio, 2018). https://www.penguinrandomhouse.com/books/546304/measure-what-matters-by-john-doerr-foreword-by-larry-page/
  4. Rodgers and Hunter, “Impact of Management by Objectives on Organizational Productivity,” Journal of Applied Psychology 76(2), 1991. https://doi.org/10.1037/0021-9010.76.2.322
  5. Goal-setting and benchmarking field experiment, Journal of Management 24(1), 1998. https://doi.org/10.1177/014920639802400106
  6. What Matters, “Operation Crush.” https://www.whatmatters.com/okrs-explained/john-doerr-operation-crush
  7. Felipe Castro, The Beginner’s Guide to OKR. https://studylib.net/doc/27271728/-felipe-castro--the-beginners-guide-to-okr
  8. Marty Cagan, “Team Objectives – Overview,” SVPG. https://www.svpg.com/team-objectives-overview/
  9. Sull, Homkes and Sull, “Why Strategy Execution Unravels—and What to Do About It,” HBR, March 2015. https://hbr.org/2015/03/why-strategy-execution-unravelsand-what-to-do-about-it
  10. “No One Knows Your Strategy — Not Even Your Top Leaders,” MIT Sloan Management Review. https://sloanreview.mit.edu/article/no-one-knows-your-strategy-not-even-your-top-leaders/
  11. Zach Saw, “I’ve Waited 20 Years to Tell This Story: How I Gamed Intel’s OKRs and Emailed the CEO About It,” Medium. https://zach-saw.medium.com/ive-waited-20-years-to-tell-this-story-how-i-gamed-intel-s-okrs-and-emailed-the-ceo-about-it-e00521ea9c1c
  12. The Deming Institute, “A Bad System Will Beat a Good Person Every Time” (from seminar notes, February 1993). https://deming.org/a-bad-system-will-beat-a-good-person-every-time/
  13. W. Edwards Deming, Out of the Crisis (MIT, 1986).
  14. Seijts and Latham, “Learning versus Performance Goals: When Should Each Be Used?” Academy of Management Executive, 2005. https://doi.org/10.5465/ame.2005.15841958
  15. Theory of Constraints Institute, “The Five Focusing Steps.” https://www.tocinstitute.org/five-focusing-steps.html
  16. Mabin and Balderstone, “The performance of the theory of constraints methodology,” International Journal of Operations & Production Management 23(6), 2003. https://www.tocinstitute.org/uploads/1/1/0/9/110976539/_toc_impact_study_3.pdf