Where we are
In b1-b4 you built the goals and the metric tree: fair, outcome-based OKRs with an owner on every branch. This session is about watching them move - the rhythm of review that steers the work early enough to matter, without turning into surveillance. Most teams get this wrong in one of two ways: they check too rarely (problems surface at quarter-end when it is too late) or they check obsessively on everything (the team spends more time reporting than working). The fix is a layered cadence - different horizons answer different questions - and a discipline that never gets old: the meeting where you steer the work is not the meeting where you judge the person.
Three horizons, three different questions 7 min live
A good monitoring rhythm is not one meeting done more often - it is three meetings with three jobs. Confuse them and every review becomes a status readout that steers nothing.
LiveThe 1:1 is the report's meeting, not yours3 min▶
Grove's point in High Output Management is easy to miss: the weekly 1:1 belongs to the person being managed, not the manager. It is where they raise what is stuck, and where you catch a problem while it is still cheap to fix - a slipping key result in week two, not a failed objective in week twelve.
- Weekly 1:1 - is the work moving? Blockers, next steps, early signals. Short and frequent beats long and rare.
- Monthly review - is the unit healthy? Trends across the team, resourcing, cross-team dependencies. Zoom out from any single person.
- Quarterly OKR review + retro - did the bets pay off, and what did we learn? Score the OKRs (for learning, not pay), keep what worked, reset the next set.
Andy Grove ran Intel on this. He argued a 1:1 that surfaces one problem early "buys back" hours of a manager's own time later. The cadence is not overhead - it is the cheapest place to catch a drifting result before it becomes a missed quarter.
Self-studyGrove's task-relevant maturity - cadence follows the task, not the title2 min read▶
How often you check in should track how new the work is to the person, not how senior they are. Grove called this task-relevant maturity (TRM): maturity is per-task, not per-person.
- Low maturity on the task (new hire, or a veteran doing something unfamiliar) - check in weekly, structured, hands-on. Not distrust; scaffolding.
- High maturity on the task - lighter touch, monitor by exception, let them run. Over-checking here reads as micromanagement and erodes trust.
The trap is anchoring cadence to rank. A senior leader learning a brand-new domain needs more structure, not less; a junior who has mastered a routine task needs less. Set frequency to the task in front of the person, and revisit it as they grow.
No naked numbers 5 min live
A number on its own cannot be read. "Revenue: $1.2M" - is that good? Up or down? Against what? Every tile on a dashboard needs three companions before it means anything: a target, a trend, and context.
Self-studyThe Balanced Scorecard reflex - and vanity metrics to drop2 min read▶
Kaplan and Norton's Balanced Scorecard (HBR, 1992) made one durable point: a board built only of financial numbers steers you off a cliff, because money is a lagging result. Balance it across four lenses - financial, customer, internal process, and learning/growth - so you see cause before effect.
- One metric that matters per horizon. Pick a single headline number for each review level so the board has a spine. Everything else is supporting context, not competing headlines.
- Leading and lagging on the same board. Pair the slow result (revenue, retention) with the fast signal that predicts it (add-to-cart rate, activation). The leading one is where you can still act.
- Drop the vanity metrics. Totals that only ever go up (cumulative signups, page views, "total revenue since launch") always look good and steer nothing. Prefer rates and ratios you can compare period to period.
A dashboard is a decision aid, not a trophy case. If a tile has never once changed a decision, it is decoration - retire it.
Self-studyThe reporting tax - when the cadence eats the work2 min read▶
Every review meeting and every dashboard tile has a hidden cost: the time your team spends preparing for it. A cadence that looks thorough on paper can quietly become a tax that crowds out the work it was meant to steer.
- Reuse one source. The weekly, monthly, and quarterly views should draw from the same numbers at different zoom levels, not three separate reports built by hand.
- Kill the deck. If people build slides for a review, the review is about the slides. Look at the live dashboard together instead.
- Budget the check-in. If a review takes longer to prepare than the decisions it produces are worth, thin it out. Frequency should lower stakes, not raise the reporting burden.
The test is simple: does this meeting change what someone does next week? If not, it is overhead wearing the costume of rigour.
Set a monitoring cadence for a Cartwheel team ★ 12 min · everyone tries
Cartwheel's Growth team just picked up a brand-new paid-social channel alongside its steady email programme. Build the cadence: what gets reviewed, how often, and with whom.
List the work and rate its maturity. The email programme is high maturity (runs itself); the new paid-social channel is low maturity (nobody has run it before).
Set frequency by maturity, not by who owns it: the new channel gets a tight weekly structured check; the mature email programme gets a light monthly look.
Assign each horizon a question and an audience - the 1:1 steers the owner, the monthly checks unit health, the quarterly retro scores and resets.
Write it into a table so it is a shared expectation, not a mood. Revisit maturity ratings each quarter as the new channel stabilises.
| Horizon | What is reviewed | Question | Who |
|---|---|---|---|
| Weekly 1:1 | New paid-social KR + blockers | Is the work moving? | Manager + channel owner |
| Monthly review | Whole Growth funnel + email trend | Is the unit healthy? | Growth team + peers |
| Quarterly retro | Score the OKRs, keep/kill the channel | Did the bets pay off? | Team + leadership |
Fix a bad Cartwheel dashboard tile ★ 8 min · do your own
Cartwheel's weekly board opens with a single tile: "Revenue: $1.2M". It is a naked number - nobody in the room can tell whether that is a good week. Rebuild it.
Add a target: the week's plan was $1.15M, so $1.2M is +4% ahead. Now the number has a verdict.
Add a trend: show the last several weeks as a sparkline, and the change vs last month (+9%). One point is noise; a line is a story.
Add context: what drove it? Revenue rose on higher AOV, not more sessions - a very different situation to explain.
Pair a leading indicator: add-to-cart rate sits beside revenue so you see next week coming, not just last week's result.
The naked number vs the tile that steers. "$1.2M" ends a conversation - everyone nods and moves on. "$1.2M, +4% vs target, +9% MoM, driven by AOV not traffic, add-to-cart trending up" starts one: is the AOV lift durable or a one-off promo? RAG colour helps only if it is a prompt to ask why, never a verdict that closes the discussion. A green dot means "looks fine, confirm it"; a red dot means "let us talk", not "someone failed".
Try it yourself - this week ◐ 30-45 min total
- Write your team's three-horizon cadence table: what is reviewed weekly / monthly / quarterly, and who is in each meeting.
- Rate each stream of work as low or high task-maturity, and set its check-in frequency from that - not from seniority. Flag anything you are checking on out of habit.
- Pick the worst naked-number tile on your current dashboard and rebuild it with target + trend + context + one paired leading indicator.
- Find one vanity metric (a total that only goes up) and either convert it to a rate or retire it.
- Audit your last three 1:1s: were they steering the work, or quietly grading the person? Note where the two got mixed.
Sources covered (cited)
Taught from the monitoring and dashboard canon. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What should set how often you check in on a piece of work?
Grove's rule: cadence follows maturity on the task, not rank. A senior in an unfamiliar domain needs more structure; a junior who has mastered a routine needs less. Revisit as they grow.
2 · Why is "Revenue: $1.2M" a bad dashboard tile?
Every tile needs target + trend + context, ideally with a paired leading indicator. Without them a number cannot be read: good or bad? up or down? against what?
3 · A red RAG status on the board should be treated as...
RAG is a prompt to ask why, not a judgment. Tracking progress and evaluating people are different meetings - steering is decoupled from grading and pay.