Ask a leadership team to write a strategy and they will give you something coherent by Friday. Ask their organisation, six months later, what that strategy actually is, and you will get a different answer from almost every desk. The strategy isn't wrong. It just never made it down the building. Cascading goals, and the OKR system in particular, exist to fix exactly that failure: not to invent the plan, but to carry it intact from the boardroom to the person doing the work.

The quick version

  • The real failure is execution, not strategy. Research repeatedly finds that most managers can't even name their company's top priorities, so the plan never turns into aligned action.
  • OKRs are a cascading-goal system. An Objective says where you're going; three to five Key Results say how you'll know you arrived. Teams set their own to support the level above.
  • Stretch is built in. Google grades aspirational OKRs on a 0–1 scale and puts the sweet spot around 0.6–0.7, hitting everything means you aimed too low.
  • The move is connection, not control. A goal is only "cascaded" when each person can trace their work up to the strategy and back down again. Most of the value is in the conversation, not the spreadsheet.

How it actually works

Strategy doesn't die in formulation, it dies in translation

The uncomfortable evidence here is hard to wave away. In a study published in MIT Sloan Management Review in 2018, Donald Sull, Charles Sull and James Yoder surveyed leaders at a large technology firm they anonymised as "Generex." Ninety-seven per cent of senior leaders said they had a clear understanding of the company's priorities, yet only about a quarter could actually list three of the firm's five stated strategic priorities, and a third couldn't name a single one. Widen the lens to 124 organisations and the picture holds: only 28% of the executives and middle managers responsible for executing strategy could list three of their company's priorities (Sull, Sull & Yoder, 2018).

Worse, the misalignment compounds downward. The same researchers found the top team agreed on the strategic objectives only just over half the time (51%); among their direct reports, agreement fell to 22%. The strategy degrades at every handoff, like a photocopy of a photocopy.

So the move is: stop treating "we communicated the strategy" as a one-time broadcast. Each leader's first execution job is their own direct reports, making sure those few people can say back, in their own words, what the strategy is and how their patch fits it. If your reports can't pass that test, nobody three levels down has a chance.

One caveat before you run with this. It's survey and self-report data from a particular set of firms, not a controlled experiment proving that tighter alignment causes better results. The numbers tell you the gap is real and common; they don't promise that closing it guarantees performance. Read them as a warning, not a law of physics.

OKRs: the system that carries a goal downhill without losing it

Objectives and Key Results are not a 2010s startup invention. The method was built at Intel in the 1970s by Andy Grove, who adapted Peter Drucker's older "Management by Objectives" and added the crucial second half, measurable key results, documenting it in his 1983 book High Output Management. A young Intel salesman named John Doerr learned it there, carried it to Google as a venture investor in 1999, and three decades later wrote the book that made it mainstream, Measure What Matters (2018).

The structure is deliberately plain. Doerr renders it as a single sentence: "I will [Objective] as measured by [Key Results]." The Objective is qualitative and a little ambitious, where you want to be. The three-to-five Key Results are numeric and unforgiving, the evidence that you got there. "Improve onboarding" is not a key result; "cut time-to-first-value from 9 days to 3" is.

An objective is where you're going. A key result is how you'll know you arrived. If a key result can't be argued about with a number, it's just a wish.

Cascading is the part that does the strategy-execution work. Leadership sets a small number of company Objectives. Each team writes its own OKRs that visibly support one of them, and, importantly, the team writes them, rather than receiving a quota from above. Doerr's rule of thumb is that roughly half of any team's OKRs should bubble up from the bottom rather than be handed down. That tension, direction from the top, ownership from the team, is the whole point.

The move, in practice: next planning cycle, pick no more than three company Objectives, give each three to five Key Results, and ask every team to draft theirs in a sentence that names which company Objective it serves. If a team can't connect its work to one, you've found either a misaligned team or a missing piece of strategy. Both are worth knowing.

Be honest about the evidence here, though. There's no strong peer-reviewed proof that OKRs beat other goal systems; most of what exists is practitioner case studies from firms that were already winning. A bad strategy executed with crisp OKRs is still a bad strategy, now executed efficiently.

Build stretch in, and don't pay people for hitting it

The most misunderstood part of OKRs is grading. At Google, aspirational OKRs are scored on a 0.0–1.0 scale, and the "sweet spot" is an average of about 0.6 to 0.7. Consistently scoring 1.0 isn't excellence, it's evidence you set the bar too low (Google re:Work). Doerr draws a sharp line between committed OKRs (the ones you must deliver, payroll runs, the contract ships) and aspirational or "moonshot" OKRs, where 70% of a huge goal beats 100% of a small one.

So here's the move: label each OKR committed or aspirational before the cycle starts, and keep OKR scores out of compensation and performance reviews. The moment a stretch goal feeds a bonus, people stop stretching and start sandbagging, they set goals they know they'll hit. You'll have protected the metric and lost the ambition.

flowchart TD
  A(["Strategy
where we're going & why"]) --> B(["Company Objectives
3 max, qualitative"])
  B --> C(["Key Results
3-5 numeric per objective"])
  C --> D(["Team OKRs
each names the company objective it serves"])
  D --> E(["My work this quarter
traceable up & back down"])
  E -.->|"check-in & re-grade"| C
How a goal cascades, and loops back. The dotted line is the part most teams skip. Leaders Loop

A worked example

Picture a mid-sized software company, call it Northwind, and treat every number here as illustrative. The board's strategy is to move upmarket: fewer, larger customers, less churn. On a whiteboard that's a sentence. As a cascading goal it looks like this.

The leadership team sets one company Objective: "Become the obvious choice for mid-market buyers." Its Key Results: lift net revenue retention from 98% to 110%; win 20 customers above £100k annual value; cut enterprise onboarding from 9 weeks to 4.

Now it cascades. The product team owns the onboarding key result and writes its own Objective, "Make the first month effortless for big customers", measured by a guided setup flow for the top three use cases and a first-value time under two weeks. The sales team owns the £100k key result; customer-success owns retention. Crucially, none were handed down as quotas, each team drafted its own and named the company Objective it serves, so the connection is visible to everyone.

Twelve weeks later they grade. Product hits 0.7, the setup flow shipped for two of three use cases. Under an old MBO regime that's a missed target and an awkward review. Under OKRs it's a healthy aspirational score, and the check-in, why the third use case slipped, what it would take, is where the real management happens. The number started the conversation; it didn't end it.

flowchart LR
  O(["Company OBJ
Obvious choice for mid-market"])
  O --> P(["Product
Effortless first month"])
  O --> S(["Sales
Win 20 x £100k accounts"])
  O --> C(["Customer Success
NRR 98% to 110%"])
  P --> P1(["KR: setup flow live
score 0.7"])
  S --> S1(["KR: pipeline x3 built
score 0.6"])
  C --> C1(["KR: churn down 4pts
score 0.8"])
Northwind's cascade, with illustrative end-of-quarter scores in the 0.6–0.8 sweet spot. Leaders Loop

Frequently asked questions

Aren't OKRs just KPIs with a new name?

No, though they're cousins. A KPI is a health metric you monitor continuously, like a heart rate. An OKR is a focused, time-boxed change you're trying to drive this cycle, like a training goal. Most of your KPIs should stay flat and fine; your OKRs are the few things you're deliberately trying to move. Confusing the two turns an OKR list into a dashboard of everything, which is the same as having no priorities at all.

How many OKRs should a team have?

Few. Doerr suggests roughly three to five Objectives with three to five Key Results each, and most teams should aim lower than that ceiling. There's supporting logic from strategy research: in an analysis of nearly every S&P 500 company, Sull found well-designed strategies typically contained just three to five objectives over a three-to-five-year horizon. If everything is a priority, nothing is, the constraint is the feature.

Should OKRs cascade strictly top-down?

No, and pure top-down cascading is the most common way the system fails. If every goal is dictated downward, teams comply without owning, and the lower-level OKRs become a tax rather than a tool. The healthier pattern is alignment, not dictation: leadership owns the company Objectives, teams write their own OKRs to support them, and the two negotiate where they don't fit. Direction comes down; ownership comes up.

Do we tie OKR scores to bonuses and performance reviews?

Deliberately not, if you want the stretch to survive. The whole design depends on people setting ambitious goals they might only 70% reach. Attach money or ratings to the score and rational people start setting safe goals, you'll hit your numbers and quietly lose the ambition the system existed to create. Keep grading a learning conversation, and judge performance through a wider lens.

We're small. Is this overkill?

The ceremony can be, but the core idea scales down well. A ten-person company doesn't need quarterly OKR software; it needs three shared Objectives written on a wall and a fortnightly five-minute check on whether the work still points at them. The discipline that matters, fewer goals, measurable, visibly connected, costs nothing. The bureaucracy is optional and usually a sign you've over-engineered it.

Related in the Toolkit

Where to go next