Planning
Battle testedWhich goal system fits your size, how to run a quarter, and how to say no to the rest.
Which planning system fits?
Six questions. The rust line is your path. Each faint branch is where the other answer leads. Turn the dial to your stage; slide or type your headcount.
Can most people name the company’s top 3 priorities without looking?
- Yes: Keep what you have
- No: More questions
Is the company under about 10 people?
- Yes: Write a top 3, no OKRs
- No: More questions
Do 3 or more teams depend on each other’s output within the quarter?
- Yes: More questions
- No: More questions
Is most of the work building a product with uncertain scope?
- Yes: Use fixed-time bets
- No: Use 90-day priorities
Do you have reliable weekly metrics, and an owner who reviews them weekly?
- Yes: More questions
- No: Not OKRs yet
Will the scores touch pay, bonus or promotion?
- Yes: Separate scores from pay
- No: Run OKRs
Keep what you have
Do not add a framework to a system that works. The failure to fix is misalignment, and if there is none, more process is cost without benefit.
Write a top 3, no OKRs
A written top 3 for the quarter, and a written top 3 for each person each week. Specific goals beat “do your best” by d = .42 to .80 in the meta-analyses Locke and Latham cite.
Use fixed-time bets
A six-week cycle, a two-week cool-down and a betting meeting, with unfinished work cancelled by default. This is Basecamp’s Shape Up, and the evidence is one company’s account.
Use 90-day priorities
EOS Rocks: 3 to 7 per company (closer to 3), 3 to 5 per department head, 1 to 3 per person, one owner each. No independent controlled evidence.
Not OKRs yet
Use 90-day priorities first and build the metrics. Key results need a number you can grade. Without data they become task lists, which Google warns against.
Separate scores from pay
Keep them apart, or use committed goals only. Linked pay brings gaming, narrow focus and less intrinsic motivation, so add safeguards if you must.
Run OKRs
3 to 5 objectives, about 3 key results each, outcomes not activities, graded 0 to 1.0, with a mid-quarter check-in. Google’s 0.6 to 0.7 is for stretch goals, so do not demand it on committed ones.
Answer the last question to see your outcome.
Stage
Where the evidence disagrees
What share of goals should be hit
An average of 0.6 to 0.7 is the sweet spot, meant for stretch goals.
Google re:Work, OKR guide: primary source (opens in a new tab)Amazon
S-team goals are mostly controllable input metrics, and about three quarters are expected to be fully achieved.
Commoncog, Amazon weekly business review: secondary source (opens in a new tab)No evidence says which is better. The grading norm is a choice.
Do goals help or harm
For
Specific, difficult goals beat “do your best” by d = .42 to .80. The evidence is mostly simple, bounded tasks.
Locke and Latham, American Psychologist, 2002: primary source (opens in a new tab)Against
Narrow focus, distorted risk, unethical behaviour and less intrinsic motivation. Dose goals like a drug.
Goals Gone Wild, via HBS Working Knowledge: secondary source (opens in a new tab)The sides disagree about scope, not mechanism. Company planning is complex, interdependent work. OKRs themselves have almost no independent evidence beyond single-company case studies.
How it runs
Three flows. Each step has an owner, a time cost and a document.
Write the week
Three outcomes, each with a “done means” clause and when and where you will work on it. Implementation intentions have a mean effect of d = .65 across 94 tests.
Friday check
Done, not done, and one line on why.
Post the company top 3
The quarter’s top 3 where everyone sees it.
Spot-check alignment
Ask five people to name the company top 3.
Score and collect
Score the quarter: OKRs 0 to 1.0, or Rocks done or not done. Gather board feedback, metric trends, customer signals and capacity.
One-page proposals
Problem, appetite (the time they would spend, not an estimate), risks, and what they will not do.
Choose priorities
Silent read, then a leadership session. 3 to 7 company priorities, closer to 3, one owner each. Leave 20 to 30% of capacity unplanned for interrupts (estimate, not a sourced number).
Cascade
Each team writes its own top 3 within three working days.
Mid-quarter check
Re-score. Cancel by default any work that is not on track to ship in the cycle.
Keep the list visible
The top 3 and the not-doing list, always in view.
Triage in a day
Run each request through the answer-a-request questions within one working day.
Reply in three parts
The no or not now, the reason in one sentence, and the route. People overestimate how hard a refusal is to give and underestimate how often others comply.
Accept urgent work as a trade
Name what comes off the top 3. The owner of the displaced item agrees in writing.
Review the parking lot
Every logged request gets a review date, checked at each quarterly planning.
Who handles what
Set by the dial on the instrument. It follows your headcount until you turn it.
- Runs it
- Founder or CEO, with a co-founder as backup.
- You keep
- Choosing the top 3.
- Hand off
- Nothing.
My stance
Founder direction and growth in a few key areas. A whiteboard, posters on the wall, the calendar. No OKRs.
- Runs it
- CEO, with the first ops hire as backup.
- You keep
- The quarterly priorities and the not-doing list.
- Hand off
- Writing up, tracking weekly, the parking lot.
My stance
Ninety-day priorities with one owner each beat OKRs here. Use done or not done, because a score invites an argument about the number.
- Runs it
- CEO sets company priorities, function leads set team priorities. Chief of Staff or head of ops as backup.
- You keep
- The company top 3 and the final cut of what is not done.
- Hand off
- Planning calendar, scoring, cascade checks.
My stance
OKRs only once you have a history of meeting goals. Set them from what the team has shown it can do, not higher.
- Runs it
- CEO plus exec team. A Chief of Staff or head of BizOps runs the process.
- You keep
- The company priorities and the trade-off calls.
- Hand off
- Everything else in the cycle.
My stance
Keep pay out of the OKR conversation. The moment a score affects a bonus, expect sandbagging or gaming.
- Runs it
- Exec team with a planning office. Chief of Staff and finance partners as backup.
- You keep
- The narrative and the few company bets.
- Hand off
- Team-level goals.
My stance
Product teams get fixed-time bets with a cancel-by-default rule. Business teams get input metrics. Do not force one system on both.
How planning goes wrong
Failure modes
- OKR theatre
Written once, never reviewed, key results are tasks. Weekly owner updates, and key results must be outcomes.
- Too many goals
Five objectives per team. Keep to 3 to 5 objectives and 3 key results each.
- Pay-linked goals
Gaming and narrow focus. Wells Fargo paid a USD 185 million fine in 2016 and dropped its sales goals from 1 January 2017. Decouple, or add safeguards.
- The no never lands
The founder says yes to everything. Use a not-doing list and a trade-off note.
The biases behind them
- Planning fallacy
In one study of 37 students, about 30% finished in the time they predicted. Use your last four quarters as the base rate, and cut scope, not dates.
- Sunk cost
You keep a failing project because of time already spent. Ask: would I start this today?
- Parkinson’s law
Work fills the cycle length. Set the appetite first, then fit the scope.
- Narrow focus under goals
People neglect valued work outside the goal. Add a quality counter-metric to each goal.
Efficiency savers
What each step saves. Where nobody publishes the number, it says so.
Cap objectives at 3 to 5
Google says more leads to over-extended teams and a diffusion of effort. No hours figure.
Google re:Work, OKR guide: primary source (opens in a new tab)Cap company priorities at 3 to 7, individuals at 1 to 3
EOS reports completion rates fall when teams take on too much. No published rate.
EOS Worldwide, Rocks: primary source (opens in a new tab)Decouple OKR scores from pay
Avoids the side effects the critics list. No hours figure.
Goals Gone Wild, via HBS Working Knowledge: secondary source (opens in a new tab)Fixed-time cycle with cancel by default
Removes re-planning mid-cycle. No published data.
Basecamp, Shape Up glossary: primary source (opens in a new tab)Not-doing list and parking lot
No published data. Removes 1 to 3 ad hoc requests a week per leader from the live agenda.
Estimate, no published data: estimateImplementation intentions in the weekly top 3
d = .65 for goal attainment, not hours saved. No reliable benchmark exists for a company-level written top 3.
Gollwitzer and Sheeran, 2006: secondary source (opens in a new tab)
Rules by country
Tools that can monitor staff
BetrVG §87: primary source (opens in a new tab)Under §87(1) Nr. 6 BetrVG the works council has a co-determination right over technical devices meant to monitor behaviour or performance. An OKR tool that tracks individuals can fall under it.
How wide the right is
BAG, 8 March 2022, via Hensche: secondary source (opens in a new tab)In a 2022 ruling the BAG treated Microsoft Office 365 as covered. Commentary reads “meant to” as objectively capable of monitoring, whatever the employer intends.
General information, last checked 29 Oct 2026. Not legal or tax advice.
Disagree with a stance, or have data that should be here?
Email me