Usable with a quick edit11 of 14 checks passed
The memo is well-structured, evidence-based, and covers all required elements; a PM could use it with minor edits, such as adding a condition that would change the recommendation.
Output
Memo: One North Star for Crewmate
To: Exec team | From: Staff PM | For: Offsite decision, ahead of the board report
Recommendation
North Star: Shifts worked as booked, per month. This counts shifts that were filled, worked, and not disputed. It is always reported split by chains and independents.
Paired business measure: Net revenue. This is our 18% take on worked shifts, minus worker incentives (sign-up and rescue bonuses).
A shift that gets filled and actually worked is the one event that keeps both sides of the marketplace. Venues at 90%+ fill keep posting (84% are still posting at six months, against 41% for venues under 70%). Workers who get three or more shifts in their first month stay (66% against 12%). Every team's work feeds this one number.
Why not the four metrics on the table
Q3 looks good on each team's number. The data underneath says otherwise.
- GMV (CEO): up 15%, from $8.5M to $9.8M a month. All of that growth is chains. Chain GMV rose from $2.9M to $4.5M, helped by an 11% holiday rate rise that will unwind. Independent GMV fell from $5.6M to $5.3M, down 6%. Four customers now produce 46% of GMV. GMV hid our biggest problem, and next quarter it will flatter or punish us on chain pricing we don't control.
- Shifts posted (Demand): up 19%. About 13,000 of the 15,000 extra shifts were chains. The metric also counts a shift that never fills as a win.
- Active workers (Supply): registrations up 30% after the $25 bonus. Only 2,370 of the 7,900 Q3 registrants (30%) have worked a shift.
- Fill rate (COO): this is the closest to right, but it fell from 83% to 77%. Chains held at 93%, while independents dropped from 79% to 68%. It has three flaws:
- It assumes the 36% of unconfirmed shifts were worked.
- It is propped up by rescue fills, which have a 13% no-show rate.
- It improves if venues simply stop posting hard shifts.
The real story: the average independent venue is now below the 70% fill line where retention collapses. Independent churn nearly doubled, from 5% to 9% a quarter. Chains, at 93%, lost no venues.
Estimated baseline. This applies the 6% no-show/dispute rate seen on confirmed shifts to all filled shifts.
| Q2 | Q3 | Change | |
|---|---|---|---|
| Chains | ~18,300 | ~29,700 | +62% |
| Independents | ~41,500 | ~37,000 | −11% |
| Total | ~59,900 | ~66,700 | +11% |
This is the honest version of Q3: real growth from chains, with the independent base eroding.
Metric tree
Shifts worked as booked = Shifts posted × Fill rate × Worked-as-booked rate
1. Shifts posted by active venues. Owner: Head of Demand - 1a. Independent venues still posting (quarterly churn, now 9%) - 1b. Share of independent posts made 24+ hours ahead (now 69%). Same-day shifts fill only 58% of the time and are 31% of independent posts, against 8% for chains. - 1c. Chain shift volume (account management) 2. Fill rate, by segment - 2a. Fill rate on shifts posted 24+ hours ahead. Owner: Head of Supply. The driver is new workers reaching three shifts within 30 days, because that is the pool that becomes reliable supply. - 2b. Same-day fill rate (58%). Owner: COO 3. Worked-as-booked rate. Owner: COO - 3a. No-show rate (4% standard, 13% on rescue fills) - 3b. Venue confirmation rate (64%). Product builds the confirmation flow; the COO owns the rate.
Fill rate by segment, and the share of independents at 90%+ fill, remain diagnostics the COO reports weekly.
The business pairing
Our gross take is about $1.76M a month. Net revenue subtracts what we pay to manufacture activity, which has become material.
Rescue bonuses. An independent shift averages about $134 GMV, so we earn about $24 on it. A rescue bonus costs $15. After the 13% no-show rate, that is about $17 per rescue shift actually worked, or roughly 70% of our take. Last month that was $93,000.
Sign-up bonuses. At $25 per sign-up, 7,900 registrations cost up to about $200,000, if the bonus was paid on sign-up. Fewer than a third of those registrants have worked.
Net revenue makes these trade-offs visible. GMV hides them.
We should still show the board GMV as context, with the chain price effect disclosed. It should not be what we steer by.
How this could be gamed, and the guardrails
| Gaming risk | Guardrail |
|---|---|
| Counting unverified shifts as worked | A shift counts only if confirmed, or unconfirmed after a 7-day dispute window. Confirmation rate is a tracked guardrail, raised from 64% toward a target we set at the offsite. |
| Buying fills with bonuses | Pairing with net revenue. Incentive cost per worked shift is capped. Rescue and sign-up bonuses are paid only on completion. |
| Growing chains while independents erode | The North Star is always shown split by segment. Guardrails: independent churn back to ≤5% a quarter, and the share of independents at 90%+ fill rising. |
| Venues posting fewer hard shifts to lift fill | The North Star counts volume, so fewer posts lowers it. Independent shifts posted is tracked. |
| Loosening vetting to add workers fast | No-show and dispute rate guardrail; a rise triggers review. |
| Price changes inflating the headline | The North Star is a count of shifts, so it is immune to rate changes. |
What each team's metric becomes
| Team | Today | Becomes | Practical change |
|---|---|---|---|
| CEO | GMV | Board context, reported alongside net revenue | Disclose chain concentration and the holiday price effect |
| Demand | Shifts posted | Independent venues retained; share of independent posts made 24+ hours ahead; chain volume | Push independents to post earlier, where we can actually fill |
| Supply | Active workers | New workers reaching 3 shifts in their first 30 days | Move the $25 bonus from sign-up to third shift; prioritise first-shift matching for new workers |
| COO | Fill rate | Same-day fill rate and worked-as-booked rate | Pay the rescue bonus on completion only; fill rate becomes a diagnostic |
Nobody loses their number. Each team's number becomes an input, and no team's metric can go up while the North Star goes down.
Decisions for the offsite
- Adopt the North Star and its definition, including the 7-day dispute window.
- Adopt net revenue as the paired measure, with GMV as board context.
- Confirm the owners for each input above.
- Approve the incentive changes: both bonuses paid on completion.
- Agree the board narrative: real chain growth, an independent problem we have identified, and a plan to fix it.
Before the board meeting
Data will backfill the North Star by segment for Q2 and Q3, using actual confirmed and disputed records instead of my estimate. We do not yet know the dispute rate on unconfirmed shifts. That is the main uncertainty in the baseline.
Data will also separate how much of the Q3 fill rate came from rescue fills, so we report the underlying independent rate. I expect it to be below 68%.
Grades and run details
Decision model 89 · LLM judge 87
Decision model checks
- passUses the supplied evidence correctly10%
- passAddresses the actual decision89%
- passRespects explicit constraints35%
- passIdentifies material uncertainty86%
- partialAvoids unsupported claims52%
- passProduces the required deliverable52%
- passSees the growth is narrow100%
- passChecks the North Star can be measured100%
- passSees the fill rate being bought100%
- passOne North Star for both sides100%
- passMeasures value created, not captured95%
- failGets the base of every number right30%
- passBreaks it into inputs teams can move100%
- passSays how it could be gamed100%
Run
- Run
- #1
- Time to output
- 82 s
- Submitted
- 5 Oct 2026