OKR and SMART goals for meeting-driven work

They are routinely presented as competitors. They are not — they answer different questions, and using the wrong one at the wrong altitude is what makes both feel like paperwork.

The 70% rule is the part most often misapplied. In Doerr's account and in Google's re:Work guidance, an OKR scoring around 0.7 is a well-set objective: consistently hitting 1.0 means the targets were sandbagged. That logic holds for aspirational quarterly outcomes and breaks completely at the level of a single meeting. "We reached 70% of a decision" is not a partial success, it is a meeting that has to happen again. Individual meeting goals are SMART goals, binary, and the ambiguity between the two frameworks is not a nuance to gloss over.

Specific, Measurable, Achievable, Relevant, Time-bound. Applied to an agenda, the criteria stop being an acronym exercise and turn into a filter that kills a surprising number of meetings before they are scheduled.

• Specific — "discuss the roadmap" is not a goal. "Choose between the two Q4 sequencing options" is. If the purpose cannot be written as a sentence with a verb in it, that is a finding.
• Measurable — the test is whether an observer could tell afterwards whether it happened. A decision recorded with an owner is measurable; "alignment" is not.
• Achievable — in the time booked, with the people invited. Most overrunning meetings were unachievable before they began: four decisions in a 30-minute slot, or the person who has to approve it not in the room.
• Relevant — it ladders up to something. If a recurring meeting cannot be traced to a quarterly objective, that is the audit finding it looks like.
• Time-bound — not the meeting's end time, which is given. The date the outcome is due, which is what stops a decision being deferred indefinitely.

One line at the top of the invite carries all five: "By the end of this meeting we will have chosen a launch date and named the owner for the go/no-go, or we will have written down what is blocking that choice and who resolves it by Friday." The second clause matters as much as the first — it converts a failed meeting into a recorded state rather than a vague feeling that it went nowhere.

The more interesting use is one level up: applying OKRs not to what your meetings discuss but to how the meeting system performs. This is where the framework earns its keep, because it converts "we have too many meetings" into something with a number attached that can be true or false at the end of a quarter.

A worked example:

• Objective: Our meetings produce decisions instead of scheduling more meetings.
• KR1: 90% of recurring meetings end with at least one recorded decision or action item with a named owner.
• KR2: Cut recurring meeting hours per person from 14 to 9 per week.
• KR3: No decision appears on three consecutive agendas of the same meeting.
• KR4: Every recurring meeting has a written purpose that names the objective it serves.

Note what these have in common: they are countable from records you either have or could have, and none of them measures satisfaction. Sentiment surveys about meetings mostly measure how the last bad week went. KR3 is the sharpest of the four, because a decision that has survived three agendas is the clearest available signal of a meeting that discusses rather than decides.

The standard move is to multiply attendees by salary by duration and present a cost. It is a useful shock and a poor metric: it counts the cost of every meeting equally and the value of none, so it makes a decisive 30-minute meeting with eight people look worse than an aimless 60-minute one with three.

Two cheaper measures tell you more. Decisions per meeting hour, counted from the record over a month, ranks your recurring meetings honestly and takes an hour to compute once you have a decision log. Decision latency — the elapsed days between a decision first appearing on an agenda and being made — catches the failure mode that the first metric misses, which is the meeting that produces decisions on schedule but only after four attempts.

Both depend on having a record. The full audit procedure, including how to gather a quarter's worth of calendar data, is in the meeting audit.

Every metric on this page assumes a record of what meetings produced, and most teams do not have one. Notes are taken inconsistently, by different people, in different places, and the meetings least likely to be documented are precisely the ones that produced nothing — which biases the data in the direction that flatters the calendar.

Earkeep makes the record a byproduct rather than a discipline. It records mic and system audio continuously while it runs and transcribes locally on your device, so every meeting has a transcript whether or not anyone volunteered to take notes — no bot to invite, no start/stop to remember. Meetings are defined afterwards by selecting a span on a timeline, or automatically from a connected calendar. Audio is processed only in memory and never written to disk, and transcripts are plain files, so counting decisions over a quarter is a search across a folder rather than a reconstruction from memory.

Goals decide what your meetings are for; the surrounding practices — batching, buffers, defended focus — decide whether you survive them. Start from the pillar guide on time management for meeting-heavy roles, and use the meeting audit to get the baseline your first key results will be measured against.

Download Earkeep free for macOS and start the quarter with a searchable record of every meeting.