What is OKR program health?
Health is about the process, not the score
Most OKR dashboards answer one question: are we hitting the numbers? That is attainment. It matters, and it is the wrong first question. A team can be green on every key result and still run a broken program: goals that are vague, unaligned, and last touched in week one. Attainment measures the goals. It says nothing about whether the organization is running the system that produced them.
OKR program health is the second question, made measurable. It asks whether people set OKRs, connect them to strategy, keep them current, and write them well. When health is high, an attainment number means something. When health is low, the green dashboard is theater, and the theater lasts right up until the quarterly review, when it is too late to change anything.
Here is the distinction in one line. Attainment is the result. Health is whether the machine that produces the result is switched on. You can hit 70 percent of your key results by accident, in a program that is quietly dying. Health is how you tell the difference.
Why this metric was missing
Walk the OKR software market and you find fifty tools that all do the same three things: author goals, cascade them, track progress. WorkBoard, Quantive, Lattice, Betterworks, Perdoo, Weekdone, Profit.co, Viva Goals, and the rest differ on price, polish, and who they sell to. Not one of them leads with a program-health score. Health of the process is, at best, a buried feature or a blog-post concept. Never the product.
The open-source world tells the same story, more bluntly. Read the schemas of the serious open OKR projects and every one models the content of OKRs: objectives, key results, progress, cycles. None models whether the org follows the process. No adoption metric. No check-in compliance score. No quality checks. The best of them stores a next-check-in date on each goal, and still never rolls it up into a number a leader can act on.
So the gap is real and specific. The word compliance in this market usually means security and privacy: SOC 2, GDPR, audit logs for data. Process compliance, whether teams actually run OKRs, has no owner. That is the metric this guide defines, and it is the reason Keel exists.
The four pillars
Program health is not a vibe. It is four things you can count. Score each from zero to one hundred, weight them, and you get one number plus a map of where to look. Here is each pillar in full: what it is, why it decays, how to measure it, and the failure it catches.
Pillar 1: Adoption
Adoption is the share of teams and owners actually running OKRs this cycle. It sounds trivial. It is the first crack in every failed program.
Most rollouts treat adoption as a launch event: an all-hands, a template, a deadline. Six weeks later the energy is gone and nobody tracks which teams stopped. Adoption is not a launch. It is a decay curve. It falls over time unless someone watches it.
Measure it as two numbers. One, the share of teams with at least one objective set this cycle. Two, how fast a gap gets noticed: does a team with no OKRs get flagged this week, or at the quarterly review? A program where two of fifteen teams skipped the quarter and nobody said anything is a program with an adoption problem, no matter how green the other thirteen look.
The failure adoption catches is the quiet opt-out. A team decides OKRs are not for them, stops setting them, and the org never notices because the dashboard only shows the teams that participate. Adoption is the pillar that makes the missing teams visible.
Pillar 2: Alignment
Alignment is the share of team and individual OKRs that connect to a company objective. An objective with no parent is a goal with no strategy behind it.
The old model was strict cascade: the CEO's goals split into VP goals, split into team goals, straight down the tree. Tidy, and it kills ownership, because teams do not fight for goals they did not write. The better model is alignment by negotiation. Leadership sets three to five company objectives. Teams read them, write their own OKRs, and connect each one to a parent. Direction from the top. Goals from the teams.
Measure it by counting. Take every team and individual objective and check whether it links to a parent. The share that do is your alignment score. When the number is low, one of two things is true: teams are working on things that connect to nothing, or the strategy above them was never made clear. Both are worth knowing before the quarter ends.
The failure alignment catches is drift. A marketing team sets a goal to run more campaigns. Nobody asks which company objective it serves. Three months later the campaigns shipped and moved nothing that mattered. Alignment is the pillar that forces the line from every team's work to the company's direction.
Pillar 3: Cadence
Cadence is the check-in rhythm. It is the single most reliable predictor of whether a program survives, and it is the first thing to decay.
This is the core lesson of Christina Wodtke's Radical Focus: the cadence, not the goal-writing, is what produces results. A quarterly goal with no weekly rhythm is a New Year's resolution. A good check-in is five minutes per person: the current value of each key result, a confidence score from one to ten, and one line on what changed. The confidence number is the tell. When it drops, you talk about why this week, not at the review.
Measure cadence as the share of key results checked in within your window, say seven days, across every team. Then rank the teams that have gone quiet. Studies of OKR programs report that teams checking in weekly complete more of their OKRs and stay active far longer than teams checking in monthly or not at all. The discipline is the whole game.
The failure cadence catches is the slow freeze. One team skips a week, then two, and the dashboard still shows last month's green. By the time anyone notices, the quarter is gone. Cadence is the pillar that surfaces the freeze while there is still time to thaw it.
Pillar 4: Quality
Quality is whether your objectives are well-formed: measurable key results, the right count, outcomes rather than task lists, and kept off compensation.
The most common quality failure is the key result that is really a task. "Launch the newsletter" is a task. "Grow active subscribers from 4,000 to 6,000" is an outcome. Felipe Castro's test is blunt: if it can go in your backlog, it is not a key result. The second failure is count: two to four key results per objective. Fewer and the objective is thin. More and focus disappears. The third is comp-linkage: tie OKRs to bonuses and people set safe, gameable targets. A healthy program keeps pay on a separate track.
Measure quality as the share of objectives that pass these checks: measurable outcomes, two to four key results, no compensation link. Low quality means the program looks busy and measures nothing.
The failure quality catches is the wish-list. Objectives that inspire and commit to nothing, graded green at quarter end because there was never a number to miss. Quality is the pillar that keeps OKRs from becoming a slide.
Turning four pillars into one number
Four scores are a diagnosis. Leaders need a headline. Weight the pillars and average them. A common split is adoption thirty, alignment thirty, cadence twenty-five, quality fifteen, and the exact weights are yours to set against what your org keeps getting wrong. The output is a single, defensible number a Chief of Staff can take to the board, plus a per-team ranking that says where to start.
Read the number against bands. Eighty and up is healthy. Sixty to seventy-nine needs attention. Below sixty, the program is at risk. Watch the trend across cycles more than the absolute number: a program moving from 62 to 71 is healthier than one sitting flat at 78 while three teams quietly slip.
Health versus attainment: two axes, not one
Do not collapse these into one score. Read them together. High attainment and high health: the program works, keep going. High attainment and low health: the green is theater, dig in before it collapses. Low attainment and high health: the process is sound, the goals were hard, adjust the targets. Low on both: restart the program, do not just reset the goals. Most orgs track only attainment, which is exactly why the decay stays invisible until the quarter is lost.
How to run a health check
Run it mid-cycle, around week five or six, while there is still runway to act. An end-of-quarter read tells you why the program died. A mid-cycle read lets you save it. Score each pillar across the org, roll them into the number, and rank teams worst-first. That worst-first list is the program owner's work queue: the two teams with no OKRs, the team whose check-ins froze, the objectives with no parent. Then help those teams. The point is support, not a scorecard used as a stick. Adoption climbs when the program owner shows up to fix, not to punish.
Who owns program health
The buyer and the user are the same person: the Chief of Staff, the Head of Strategy, the PMO lead. The one who owns the OKR rollout and gets the uncomfortable question when it stalls. Program health is their instrument. It turns "I think OKRs are slipping" into "adoption is 71, cadence is 54, and here are the four teams to fix this week." That is the difference between a hunch and a plan.
The one-line version
OKR program health measures whether your organization runs the process, not whether it hits the goals. Four pillars, one score, one work queue. Measure it mid-cycle, help the teams that slip, and the green on your dashboard starts meaning something again.