Where we are
Last session set the rhythm of review. This one is about the conversation itself - the highest-leverage and most-mishandled part of measuring people. Three ideas carry it. First, frequency beats formality: a short honest check-in this week is worth more than a polished form once a year. Second, the three talks must stay separate - steering (are we on track?), development (how do you grow?), and evaluation (rating, pay) - because the moment someone knows a conversation feeds their pay, they stop admitting what they are struggling with, and the coaching dies. Third, how you have the conversation - dignity, an honest explanation, a chance to respond - shapes whether it is received as fair as much as the outcome does.
Continuous check-ins over the annual review 6 min live
The once-a-year review is a memory test that mostly records recent events and the rater's mood. Companies that reformed it did not add more forms - they added more conversations, smaller and more often.
LiveCFRs - Conversations, Feedback, Recognition3 min▶
Doerr pairs OKRs with CFRs - the continuous human layer that makes goals work:
- Conversations - regular, structured one-to-ones between manager and report.
- Feedback - specific, timely, two-way; peer as well as manager.
- Recognition - noticing good work publicly and often, not saving it for review season.
The key move: CFRs run on their own cadence, deliberately separate from formal evaluation. When Adobe scrapped annual reviews for lightweight "Check-ins," and Deloitte moved to frequent "performance snapshots," the reported gains came from talking more often, not from a better scoring form. Gallup's finding is blunt: engagement tracks with feedback frequency, not with the polish of the annual document.
Adobe's "Check-in" (2012). Adobe replaced the annual review with frequent, informal check-ins focused on ongoing feedback and goals. It publicly reported less time lost to review admin and lower voluntary attrition. The lesson to steal is not the brand name - it is that frequency and separation did the work, not a cleverer rating scale.
Self-studyRecognition is the cheapest lever you are underusing2 min read▶
The R in CFRs gets skipped, because recognition feels soft next to goals and ratings. It is not. Frequent, specific recognition is one of the strongest and lowest-cost drivers of engagement, and it belongs to its own rhythm - not saved up for review season.
- Specific, not generic. "Great job" fades; "the way you rewrote that returns flow cut support tickets" sticks, because it names the behaviour and the impact - the same shape as good feedback.
- Timely. Recognition close to the event carries far more weight than a line in an annual summary.
- Peer as well as manager. Recognition does not have to flow only downward; peer recognition scales and reaches work a manager never sees.
Because recognition is not evaluation, it can be frequent and public without corrupting anything - it is the one performance conversation you can safely have every week.
Keep the three conversations separate 5 min live
This is the linchpin of fair measurement: steering, development, and evaluation are three different conversations, and merging any of them into the pay talk corrupts the other two. People will not reveal weaknesses to the person deciding their compensation.
Self-studyProcedural and interactional justice - fairness is in the how2 min read▶
Greenberg's work on organizational justice shows people judge fairness on more than the outcome. Two dimensions decide whether a hard conversation lands as fair:
- Procedural justice - was the process fair? Consistent criteria, known in advance, the same rules for everyone, decisions based on accurate information.
- Interactional justice - was the person treated with dignity? An honest explanation of the reasoning, respect in delivery, and a genuine chance to respond before anything is finalised.
The practical upshot: a person can accept a disappointing rating if the process was clean and they were treated with respect - and will reject a generous one that felt arbitrary or was delivered dismissively. Fairness is built in the how, not only the what.
Turn vague feedback into specific + actionable ★ 10 min · everyone tries
A Cartwheel manager wants to tell an analyst to "be more proactive." As written it is useless - and, because "proactive" is a personality label, it is exactly the kind of vague feedback that lands unevenly across people. Rebuild it into behaviour, impact, and a next step.
Name the specific behaviour you actually observed, with an example - not a trait.
State the impact it had, so the "why it matters" is concrete, not a scolding.
Give a next step the person can act on this week, and offer support.
Check it is about the work, not the personality - "proactive," "abrasive," "not a leader" are labels; behaviour + impact are observations.
| Element | Vague (before) | Specific + actionable (after) |
|---|---|---|
| Behaviour | "Be more proactive." | "In the last two sprint reviews the checkout-bug data arrived after the decision was made." |
| Impact | (none stated) | "So the team chose without your numbers and had to revisit it." |
| Next step | (none stated) | "Circulate the key figures the morning before each review - want to set a shared reminder?" |
| About the... | person ("proactive") | work (a behaviour and its effect) |
Map a manager's conversations ★ 8 min · do your own
A Cartwheel team lead has three recurring conversations with each report. Sort which is steering, which is development, and which is evaluation - then see why the pay talk must never ride along in the weekly 1:1.
List the recurring conversations: the weekly 1:1, the quarterly growth check-in, and the annual rating + pay review.
Tag each with its purpose and cadence, and note what the report should feel safe saying in it.
Trace what happens if you fold the pay decision into the weekly 1:1 - and confirm the OKR score-review is reviewed for learning, not wired to the pay number.
| Conversation | Purpose | Cadence | Safe to say |
|---|---|---|---|
| Weekly 1:1 | Steering | Weekly | "I'm stuck on X, I need help." |
| Growth check-in | Development | Quarterly | "I'm weak at Y, I want to build it." |
| Rating + pay review | Evaluation | Annual / rare | The formal verdict, explained. |
Why never merge the pay talk into the 1:1. The weekly 1:1 only works if the report can say "I'm stuck" out loud. The moment they suspect that same meeting feeds their raise, they stop admitting where they struggle - they defend the number instead. You lose the early-warning signal the 1:1 exists to give you, and the development conversation dries up too. Keep the pay talk in its own rare, well-explained meeting; let the 1:1 stay a safe place to be honest.
Try it yourself - this week ◐ 30-45 min total
- Take one piece of feedback you gave recently and rewrite it into behaviour + impact + next step. Strip out every personality label.
- Map your own recurring conversations with a report onto steering / development / evaluation. Find any that are quietly doing two jobs.
- Check whether your OKR score-review and your pay conversation are separate meetings. If they are not, sketch how you would split them.
- Before your next hard conversation, plan the procedural and interactional justice: consistent criteria known in advance, an honest explanation, and a real chance to respond.
- Scan a few recent feedback notes for words like "abrasive," "aggressive," or "not a culture fit" - vague, often bias-loaded labels. Rewrite them as behaviour + impact before b7.
Sources covered (cited)
Taught from the performance-conversation canon. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · Why should the pay conversation never be merged into the weekly 1:1?
Keep steering, development, and evaluation in separate rooms. If the 1:1 feeds pay, the report defends the number instead of raising blockers, and you lose the honest signal the 1:1 exists to give.
2 · What makes feedback specific and actionable rather than vague?
Behaviour + impact + next step can be acted on and does not depend on whether the manager likes the person. Personality labels like "proactive" or "abrasive" are vague and land unevenly.
3 · Greenberg's justice research says a hard rating is received as fair mainly when...
Procedural justice (fair, consistent, known process) and interactional justice (dignity, honest explanation, a chance to respond) shape perceived fairness as much as the outcome itself.