Somewhere right now, a leadership team is having the framework fight. One founder read Measure What Matters and wants OKRs. The COO ran EOS at her last company and wants rocks. Someone from a big-tech background thinks both are overkill. The meeting ends with a committee to evaluate goal frameworks, which is itself a bad sign.
Here's the thing the framework fight misses: you're not choosing one thing. You're choosing two, and they've been bundled together by the frameworks' marketing.
Untangle them and the decision gets dramatically easier.
First, the plain-English versions
OKRs (Objectives and Key Results): each goal is a qualitative objective ("Make onboarding effortless") plus 2–5 quantitative key results that define success ("completion rate 35% → 60%," "time-to-first-value under 3 days"). Objectives can nest — a team's objective links to a company objective — forming an alignment tree. Orthodox OKRs add stretch scoring (0.7 is success), confidence ratings, and separate committed vs. aspirational tiers.
Quarterly priorities (rocks, big rocks, must-dos — the EOS tradition and its cousins): each is a single concrete thing that must get done this quarter ("Launch the partner portal," "Hire two AEs who ramp"). One owner, binary done/not-done at quarter's end, reviewed weekly with a simple on-track/off-track call. No key results, no scoring, no tree — alignment happens in the planning conversation, not the data structure.
Both agree on more than their advocates admit: a 90-day cadence, a short list (3–7), named owners, and a weekly check. The genuine differences are just two:
- Goal format — is a goal a measured outcome (KRs) or a completed thing (done/not-done)?
- Alignment formality — is the company↔team connection a structural link (the tree) or a conversation (planning review)?
That's it. That's the whole fight.
The two-decision insight
Notice that those two decisions are independent. You can have measured outcomes without an alignment tree. You can have a formal cascade of simple done/not-done priorities. The frameworks bundle them — OKRs bundles measurement and formal alignment; priorities bundles simplicity and informal alignment — and the bundling is why teams end up with a framework that's half right for them.
The most common mismatch: a 40-person company adopts full OKRs because they want measurement discipline (a real need), and gets the alignment apparatus free with purchase (a need they won't have for two years). Now every team goal requires picking a parent objective, grading confidence, and debating whether 0.7 counts — and by quarter two, people are reverse-engineering parent links to justify work they were doing anyway. That's not an OKR failure; it's a bundling failure. They bought a suit for the body they hoped to have.
The reverse mismatch exists too: a 200-person company runs simple priorities long past the point where "alignment by conversation" scales, and teams quietly drift into local optimizations nobody connected to strategy.
An honest scorecard
Where OKRs genuinely win: measurement forces clarity — "launch the portal" can ship broken, but "portal live with 40% partner adoption" can't hide. The tree makes strategy legible in a larger org. Outcome framing keeps teams from confusing shipping with succeeding.
Where OKRs genuinely lose: the overhead is real — writing good KRs is a skill (we wrote a whole guide on it), and stretch scoring confuses everyone the first four quarters ("we hit 0.68 — did we... win?"). Below ~50 people, the tree formalizes conversations you could just have. And OKR orthodoxy's "OKRs aren't tied to run-the-business work" leaves 60% of what companies actually do unrepresented.
Where priorities genuinely win: anyone understands them in one meeting. Done/not-done kills grade-inflation debates. The weekly on-track/off-track call takes ten seconds per rock. For a company that's never written goals down, this is the on-ramp that actually gets used.
Where priorities genuinely lose: "done" is gameable — a thing can be completed while the outcome it existed for doesn't happen (the binary ship-it problem). No structural alignment means that at scale, the connection between team work and company direction lives in people's heads, and heads turn over.
The hybrid, without the dogma
Most companies that make goals stick converge on the same synthesis, whatever they call it:
- A short list of company goals — 3–5, one owner each, with a plain-English success measure (that's the OKR insight, minus the apparatus: one measurement per goal, no scoring, no 0.7 debates).
- Team goals that are mostly concrete deliverables — priorities-style, done/not-done, because most team-level work genuinely is "ship the thing" — each with its milestones or checkpoints.
- Alignment by visibility, not by mandate — teams plan with the company goals literally in view, most team goals obviously support one, and a team is allowed a flagged "local" goal (the billing migration, the compliance work) that defends itself in planning review rather than faking a parent.
- One weekly rhythm for all of it — owner calls the status out loud, on track / at risk / off track, every week (the meeting is its own playbook).
Graduate to the formal tree when — and only when — the org actually needs it: multiple layers of teams, strategy that no longer transmits by conversation, new leaders who can't absorb context from hallways. Formal alignment is a capability you add to a working goal habit, not a starting requirement. Companies that start with the tree before the habit get neither.
The decision: five questions
Answer honestly — about the company you are, not the one you're becoming.
1.Do written, owned goals exist today?
2.If you asked five people "what are the company's top goals," how many answers would match?
3.Does anyone own a number they report weekly?
4.How many layers of teams plan somewhat independently?
5.When a goal misses, what happens?
Start with simple goals.
One sentence, one owner, one success measure per goal, company level only, reviewed weekly. Your bottleneck isn't framework sophistication — it's the habit. Adding KRs or trees now is buying weights before learning to walk. Revisit in two quarters.
Goals and milestones, two layers.
Company goals with success measures; team goals as concrete deliverables with milestone checkpoints; teams plan with company goals visible; local goals allowed but flagged. This is the hybrid above, and it's where most 20–200 person companies should live — many, permanently and happily.
You're ready for full OKRs, selectively.
Add key results where measurement genuinely sharpens the work (usually company goals and metric-driven teams first), and the alignment tree where conversation no longer scales. Skip stretch scoring until the honesty layer is solid — sandbagged 0.7s are worse than honest done/not-dones.
A tie or a mix? Round down. In goal systems, the under-engineered version that gets used every week beats the correct version that gets abandoned in week five — and upgrading a working habit is easy, while resuscitating an abandoned framework is nearly impossible, because the framework takes the blame.
The answer nobody markets
"OKRs vs. priorities" is framed as a war because frameworks are sold as identities. In practice it's a dial, not a door: start simple, measure what matters as the habit hardens, formalize alignment when the org outgrows conversation. The companies that win the quarter aren't the ones with the most sophisticated goal architecture — they're the ones where every goal has one owner who says a true status out loud, every single week.
Pick the dial position that matches the company you are. Then go run the week.