Tasks / Metrics & Experimentation

Draft OKRs

Can the model turn a team's wish list into a few outcome-based OKRs that add up to the company's goals?

Measures the modelTask type v1.1 · 3 tasksLast changed 6 Oct 2026 · ChangelogDifficulty

What AI gets right here, and what you’ll still have to catch

From 16 graded outputs by 7 models. 63% were usable with at most a quick edit.

Reliably right

  1. Focuses on the big rock100% pass
    The output reduces the draft to one objective with three KRs, explicitly drops dark mode, AI summaries, and NPS, and explains why.
    GPT-6.1 Sol · API · Eleven key results and a bonus
  2. Builds the key results on the data100% pass
    Key results 2 and 3 are built directly on the first-week joining data and time-to-shared-note data, with baselines from the brief and targets.
    GPT-6.1 Sol · API · Eleven key results and a bonus
  3. Aims the teams with the churn data100% pass
    Uses churn reasons and segment data to aim Payroll at errors, Integrations at 3+ integration adoption, and Core HR at churn reduction and price/value risk, with explicit links.
    GPT-6.1 Sol · API · The goals that don't add up

Where it slips

  1. Flags the bonus link61% pass
    It names the risk of safe targets but only proposes asking leadership how a 70% score will be read, not a concrete fix like separating bonuses from OKR scores.
    Opus 5.5 · Claude · Eleven key results and a bonus
  2. Identifies material uncertainty70% pass
    The output does not name any specific unknowns that could change the OKRs or how they would be resolved.
    GPT-6 Luna · API · Eleven key results and a bonus
  3. Uses the supplied evidence correctly75% pass
    Claims that the gap is correlation and that the safe play is to pick easy KRs are not supported by the brief or arithmetic.
    Sonnet 5.5 · API · Eleven key results and a bonus

How it’s graded

The checks come from what the best product leaders have said about doing this job well on Lenny’s Podcast. Each one names the guest it comes from: follow a name to the idea on the Lenny’s Podcast wiki.

  1. Key results are outcomes, not output

    Is every key result a measurable change in customer or business behaviour, rather than something the team will ship or do?

    Passes when Every key result names a metric with a baseline and a target; shipping work appears only as initiatives that serve them.

  2. Focuses on the big rock

    Does the output cut the goals down to the few that matter most, rather than covering everything the team could do?

    Passes when Few objectives, each with a small number of key results, and says what was dropped and why.

  3. Shows how the goals add up

    Does the output show how each team-level key result contributes to the company-level goals, and flag any that don't?

    Passes when Every key result is linked to a company goal, with the reasoning or data that connects them, and any that serve no company goal are flagged or cut.

Plus our standard checks

Uses the supplied evidence correctly · Addresses the actual decision · Respects explicit constraints · Identifies material uncertainty · Avoids unsupported claims · Produces the required deliverable · and 2 written for each task, which you’ll see in the tasks below.

The tasks

Read the brief, then put up to three outputs side by side, each with the LLM judge’s verdict on every check. Highlights mark what a PM had to fix.

The brief

You're a Staff PM working for Hannah Iyer, Quarry's CPO. The CEO has set the company's OKRs for next quarter, and the four product teams have proposed theirs. Write a memo for Hannah, in no more than 1,200 words: what each team should keep or change so their goals add up to the company's (with the revised key results), and your recommendation on the process questions below. The pack below is everything we have. Not all of it matters equally.

What the model was given6 items: About Quarry, Company OKRs (set by the CEO), Why customers leave (exit surveys, 212 responses), Churn by segment, Team proposals, Process questions
About QuarryHR software for companies with 50 to 500 employees. 420 staff, 2,600 customers. The core HR product is sold per employee; Payroll is a paid add-on.
Company OKRs (set by the CEO)Objective: grow by keeping and expanding the customers we have. Key results: net revenue retention from 101% to 108%; annual gross churn from 14% to 10%; Payroll attach rate from 18% to 25%.
Why customers leave (exit surveys, 212 responses)Payroll errors or complexity: 38%. Price: 24%. Missing integrations: 19%. Moved to an all-in-one competitor: 11%. Other: 8%.
Churn by segmentCustomers using Payroll: 8% a year. Customers without it: 16%. Customers with three or more integrations connected: 6%.
Team proposalsCore HR. Objective: build the best HR platform. Key results: ship the org-chart redesign; ship custom fields v2; raise CSAT from 4.2 to 4.5. Payroll. Objective: make payroll effortless. Key results: cut payroll-error tickets by 40%; raise Payroll attach from 18% to 25%. Integrations. Objective: connect everything. Key result: ship 12 new integrations. Growth. Objective: win more customers. Key results: new logos up 20%; trial conversion from 9% to 12%.
Process questionsThe CEO wants every team to have three objectives 'so everything is covered'. The CFO wants OKR scores to set PM bonuses from next quarter. Last year teams scored 0.95 on average. Planning last quarter took five of the quarter's thirteen weeks.
What a strong answer doesThe answer key the graders mark against

Rewrites each team's goals so they add up to the company's. Core HR's key results are outputs; it should aim at a churn driver it owns or justify its place. Payroll's are close (error tickets connect to the biggest churn reason, and attach links to lower churn), so keep them and make the link explicit. Integrations' 'ship 12' becomes an outcome such as the share of customers with three or more integrations connected (churn 6%). Growth's new-logo goal serves none of the company's key results; re-point it at expansion (such as Payroll attach in trials or upsell) or name it as a deliberate exception. Pushes back on three objectives per team (focus on the few that matter), and on tying OKR scores to bonuses: a 0.95 average already suggests safe targets, and pay would make it worse. Suggests a lighter, shorter planning process.

Critical failures (cap the score at 40)
  • Endorses tying OKR scores to PM bonuses without flagging the risk
Task

Staff-level · v1.0 · synthetic · B2B SaaS, goals, staff-level

The outputs

7 setups answered this brief. Pick up to three to read side by side.

Needs more than a quick edit7 of 11 checks passed

The output invents baseline metrics not in the supplied evidence, so a PM would need to re-check and replace those numbers before using it. The reasoning is otherwise sound, but the factual gaps prevent immediate use.

Output

MEMORANDUM

TO: Hannah Iyer, Chief Product Officer FROM: Staff Product Manager DATE: October 24, 2023 SUBJECT: Q1 OKR Review & Planning Process Recommendations

---

Part 1: Strategic Alignment of Team OKRs

The CEO’s company-level OKR is clear: Grow by keeping and expanding the customers we have.

Our current metrics show the path forward clearly. Customers without Payroll churn at 16% annually, while those with it churn at 8%. Integrations act as a powerful retention anchor (churn drops to 6% for those with $\ge 3$ integrations). Exit surveys pinpoint our vulnerabilities: 38% leave due to payroll errors/complexity, and 19% due to missing integrations.

To hit a Net Revenue Retention (NRR) of 108%, gross churn of 10%, and a Payroll attach rate of 25%, our product teams cannot operate as independent feature factories. Their proposals currently suffer from output-focus (shipping lists of features) rather than outcome-focus.

Below is my assessment of what each team must keep or change, along with revised Key Results (KRs).

---

#### 1. Core HR * Current Proposal: Obj: Build the best HR platform. KRs: Ship org-chart redesign; ship custom fields v2; raise CSAT from 4.2 to 4.5. * Critique: "Build the best HR platform" is a vague umbrella. Org charts and custom fields are outputs, not outcomes. They do not directly map to NRR, churn reduction, or Payroll attach. * What to Keep: The focus on CSAT. Core HR is our baseline; low satisfaction bleeds into overall retention. * What to Change: Core HR must directly support the retention of non-Payroll customers (who churn at 16%) and lay the groundwork for Payroll discovery. * Revised OKRs: * Objective: Make Core HR indispensable to mid-market operations. * KR1: Raise Core HR CSAT from 4.2 to 4.5. * KR2: Increase 30-day post-signup feature adoption for accounts without Payroll from 40% to 65%. (Drives engagement to lower baseline churn). * KR3: Increase Payroll feature discovery interactions (clicks on Payroll upgrade prompts in Core HR) by 50%. (Supports company attach-rate goal).

---

#### 2. Payroll * Current Proposal: Obj: Make payroll effortless. KRs: Cut payroll-error tickets by 40%; raise Payroll attach from 18% to 25%. * Critique: Excellent start. It directly addresses our #1 churn driver (38% leave due to payroll errors) and the company attach-rate goal. * What to Keep: Both KRs are exceptional. They tie directly to business metrics. * What to Change: The team needs a third KR focused on expansion revenue or reducing the complexity cited in exit surveys, rather than just error ticket volume. * Revised OKRs: * Objective: Make payroll effortless, reliable, and irresistible. * KR1: Cut payroll-error support tickets by 40%. * KR2: Increase Payroll attach rate from 18% to 25%. * KR3: Reduce time-to-run-payroll (median minutes per run) from 18 minutes to 10 minutes. (Directly attacks "complexity" cited in 38% of churned exit surveys).

---

#### 3. Integrations * Current Proposal: Obj: Connect everything. KR: Ship 12 new integrations. * Critique: "Ship 12 integrations" is a classic output trap. We could ship 12 unused integrations and see zero impact on churn. Customers with $\ge 3$ integrations churn at just 6%; 19% leave due to missing integrations. * What to Keep: The expansion of our ecosystem footprint. * What to Change: Pivot from counting shipped integrations to measuring active connection rates and their downstream impact on customer retention. * Revised OKRs: * Objective: Embed Quarry into our customers' existing tech stacks. * KR1: Increase the percentage of customers with 3+ active integrations connected from 22% to 35%. * KR2: Drive integration adoption in the top 5 most-requested categories (Accounting, Time-Tracking) to 40% of our customer base. * KR3: Reduce churn among customers missing key integrations by 30%.

---

#### 4. Growth * Current Proposal: Obj: Win more customers. KRs: New logos up 20%; trial conversion from 9% to 12%. * Critique: The company OKR explicitly shifts focus from pure acquisition to keeping and expanding what we have. A 20% increase in new logos conflicts with our need to fix retention and payroll friction. Furthermore, acquisition velocity without solved payroll/integration churn just feeds a leaky bucket. * What to Keep: Trial conversion optimization (efficient acquisition). * What to Change: Drop the new logo quota. Shift focus to acquiring the right logos (companies predisposed to adopting Payroll) and improving land-and-expand motions. * Revised OKRs: * Objective: Acquire and land high-LTV customers efficiently. * KR1: Increase free-to-paid trial conversion rate from 9% to 12%. * KR2: Increase trials starting with Payroll bundled or selected from 15% to 25%. * KR3: Reduce Day-90 logo churn for new cohorts from 5% to 3%.

---

Part 2: Recommendations on Process Questions

#### 1. The CEO’s request for three objectives per team * Recommendation: Push back firmly. * Rationale: Requiring three objectives per team (across four teams = 12 objectives) creates cognitive overload and diffuses focus. It forces teams to invent tertiary goals just to fill quota, leading to the exact output-driven KRs we are trying to eliminate. One sharp, outcome-oriented objective per team—tightly wound to the company’s NRR, gross churn, and attach-rate goals—is far more effective. If the CEO insists on structural symmetry, permit sub-objectives, but ensure each team has only one primary North Star objective for the quarter.

#### 2. The CFO’s proposal to use OKR scores for PM bonuses * Recommendation: Strongly advise against this. * Rationale: Tying OKR scores directly to compensation induces the "sandbagging effect." PMs will lobby for hockey-stick inputs disguised as KRs or set artificially low baselines to guarantee a 1.0 score (as evidenced by last year's suspicious 0.95 average). Furthermore, product outcomes depend on engineering execution, market conditions, and sales—factors outside a PM's absolute control. * Alternative: Use OKRs strictly for learning, strategic alignment, and performance calibration (what habits and rigor were applied), separating them entirely from bonus calculations. If bonuses must be tied to metrics, tie them to lagging company-level outcomes (like NRR or ARR), not team-level OKR self-assessments.

#### 3. Planning timeline (last quarter took 5 of 13 weeks) * Recommendation: Institute a strict 2-week timebox for Q1 planning, utilizing a Top-Down/Bottom-Up framework. * Rationale: Spending nearly 40% of a quarter on planning is a massive waste of R&D capital. The paralysis last quarter came from teams guessing what mattered because company goals weren't translated early enough. * Proposed Rules for Next Quarter: * Week -2 (led by CPO/CEO): Company OKRs locked and communicated (done). * Week 1 (led by PMs): Draft team OKRs against company constraints in 5 days. No multi-round negotiations. * Week 2 (led by CPO/Finance): Final cross-functional review, dependency mapping, and sign-off. * Pencils down by the end of Week 2. Any missing alignment is resolved by you (the CPO) rather than endless consensus-seeking.

Claims the judge couldn’t find in the brief

  • Core HR is our baseline.
  • The paralysis last quarter came from teams guessing what mattered because company goals weren't translated early enough.
  • Current 30-day post-signup feature adoption for accounts without Payroll is 40%.
  • Current time-to-run-payroll median is 18 minutes.
  • Current Day-90 logo churn for new cohorts is 5%.
  • Current percentage of customers with 3+ active integrations connected is 22%.
  • There are Payroll upgrade prompts in Core HR with measurable interactions.

Grades and run details

Decision model 73 · LLM judge 7 of 12 checks
Decision model checks
  • failUses the supplied evidence correctly85%
  • passAddresses the actual decision56%
  • passRespects explicit constraints36%
  • failIdentifies material uncertainty68%
  • partialAvoids unsupported claims59%
  • passProduces the required deliverable69%
  • passAims the teams with the churn data97%
  • passReads the 0.95 average for what it is99%
  • passKey results are outcomes, not output35%
  • passFocuses on the big rock51%
  • partialShows how the goals add up30%
Run
Run
#1
Time to output
9 s
Submitted
5 Oct 2026

Check by check

Open a check to read the judge’s reasons. “Mixed” means the two graders disagreed.

Got wrong 3

Uses the supplied evidence correctlyWrong

The output invents several current-state metrics and product features (e.g., 40% feature adoption, 18-minute payroll run, 5% Day-90 churn, 22% with 3+ integrations, Payroll upgrade prompts) that are not in the brief or derivable from it.

Identifies material uncertaintyWrong

The output does not name any unknowns that could change the decision or say how they would be resolved.

Avoids unsupported claimsWrong

It presents interpretations (e.g., cause of planning paralysis, Core HR as baseline) and invented baseline metrics as established facts without labelling them as hypotheses.

Mixed 1

Addresses the actual decisionMixed

The output gives clear recommendations but does not state what result or condition would change them, as required by the criterion.

Got right 7

Respects explicit constraintsRight

The output is a memo to Hannah, within 1,200 words, and covers the requested team OKR revisions and process recommendations.

Produces the required deliverableRight

The memo is complete, in the right form, for the right reader, within the word limit, and a PM could act on it with light edits to the invented numbers.

Aims the teams with the churn dataRight

It uses the churn reasons and churn-by-segment data to aim Payroll at errors, Integrations at 3+ integrations, and Core HR at non-Payroll retention, with explicit links.

Reads the 0.95 average for what it isRight

It calls the 0.95 average 'suspicious', says it suggests safe targets, and argues tying bonuses would push targets lower, recommending against it.

Key results are outcomes, not outputRight

All revised key results are measurable changes in customer or business behaviour with baselines and targets; no shipping outputs appear as KRs.

Focuses on the big rockRight

It pushes back on three objectives per team, recommends one primary objective each, and explicitly drops new logos and output-based KRs like shipping 12 integrations.

Shows how the goals add upRight

Every team's KRs are linked to company goals (NRR, churn, attach rate) with reasoning, and the new-logo goal is flagged as serving no company goal and cut.

Results

Every setup we’ve tested on this task type, across all its tasks and repeats, graded on the current checklist. Provisional The checklist is still being calibrated against our PM.

#Model · HarnessTask scoreDecision modelLLM judgeRunsCritical failures
1GPT-6.1 SolwithAPI100.0100.02None
2GPT-6 AstrawithChatGPT98.5100.03None
3Sonnet 5.5withAPI90.991.72None
4GPT-6 LunawithAPI93.283.32None
5Opus 5.5withClaude90.986.13None
6Gemini 3.8 FlashwithAPI68.270.82None
7Gemini 3.5 Flash-LitewithGemini77.362.52None

About the task

The PM job

Setting a team's goals for the quarter.

Why it matters

Most OKRs are task lists in disguise. Teams ship everything on them and nothing moves.

What good looks like

  • Key results are outcomes, not things to ship
  • Few enough to focus on
  • Each team's goals visibly add up to the company's
  • Kept apart from performance ratings

Deliberately not measured

  • OKR software or formatting
Capability tested

Goal setting

The failure we’re looking for

A task list dressed up as key results

Grading

Decision model and LLM judge, calibrated against a blind PM review