CPO Day-90 Readout Skeleton

One memo, SCQA-shaped, five working sections. Written to be read, not presented. This is the actual deliverable of your first 90 days, and people will quote it back to you for a year, so every sentence has to be one you can defend with a ledger row.

Length budget: 4 to 6 pages. If it runs longer, cut the situation section first. Your audience lived the situation. They need your read of it, not a retelling.

---

The memo skeleton

# [Product name]: Where We Are and What I'm Betting

**Author:** [You]
**Date:** [Day ~90]
**Audience:** Exec team, then board
**The answer in one sentence:** [Your whole readout compressed. "We have a quality problem masquerading as a growth problem, and I am betting the next two quarters on evals and margin before any new surface area." If you cannot write this sentence, the memo is not done.]

---

## 1. Situation: what the audits found

[1 to 1.5 pages. The output of your four audits, stated as findings with evidence, not impressions. Each finding is one bolded sentence plus 2 to 3 supporting sentences citing where the evidence lives.]

[Reality audit finding: what using the product and sitting in raw calls showed. Quote your own day-zero customer experience, it is the one read nobody can accuse of being curated.]

[Money audit finding: cost per outcome by workflow. Name the workflow that is underwater if you found one, with the number.]

[Quality audit finding: how many production AI features have an eval score with a trend, how many have a number somewhere, how many returned silence. The pile sizes ARE the finding.]

[Decision audit finding: how the last ten significant decisions actually got made. One sentence on the pattern: where decisions stall, who really decides, what evidence is typically in the room.]

[Rule: every claim here traces to a claim-ledger row marked Confirmed. If a finding rests on an Untested row, either test it this week or cut it.]

## 2. Complication: the two or three constraints that matter

[Half a page. Not a list of everything wrong. The two or three things that make the obvious plan unavailable. "We cannot grow into new segments while eval scores on the core workflow are trending down." "Margin math caps how much agent compute we can spend per customer at current pricing."]

[Each constraint gets: the constraint in one sentence, the evidence in one sentence, what it rules out in one sentence.]

[If you have more than three, you have not finished thinking. Merge or cut.]

## 3. Question

[Two or three sentences. The question the complication forces. Frame it so your answer is a choice among real options, not the only door in the room. "Given X and Y, do we spend the next two quarters fixing the core or expanding the surface? And in either case, what do we stop funding?"]

## 4. Answer: the bets, the anti-bets, the kill list

[1.5 to 2 pages. The heart of the memo.]

### The bets (two or three, no more)

[For each bet:]
- **Bet:** [One sentence. What we are doing.]
- **Why this, why now:** [Two sentences, citing audit findings from section 1.]
- **What we'll see if it works:** [The measurable signal and the date we check it.]
- **Reversibility:** [Can we walk this back? At what cost? Weight irreversible bets accordingly and say so.]
- **What would change my mind:** [The evidence that would make you kill this bet. Writing it now is what makes the bet honest.]

### The anti-bets (what we are explicitly not doing)

[Three to five items. One sentence each, with the one-line reason. This paragraph replaces a hundred future hallway debates. "We are not building a second vertical this year, because the core workflow eval trend says we have not earned it."]

### The kill list

[The initiatives, ceremonies, and features the audits showed are running on inertia. For each: name, what it costs (money, people, attention), what killing it frees up, and the date. Two of these should already be dead by the time the memo lands, killed publicly in month two or three, so the list has precedent behind it.]

## 5. Scoreboard: judge me on this

[Half a page. The three to five numbers you want to be measured on for the next year. For each: the number, today's baseline, the direction you are committing to, and when it gets reviewed. Margin-aware, eval-backed, outcome-shaped. This section should match the scoreboard you renegotiated with your CEO around day 50. If it does not match, you skipped a conversation, go have it before this memo ships.]

## Appendix

[Audit detail, methodology, the full kill-list cost math, claim-ledger summary stats. Rule: nothing load-bearing lives in the appendix. If a reader skips it entirely, the memo still stands.]

---

The one-page exec summary variant

For the board pre-read, or for the CEO who asks for the short version. Same skeleton, compressed:

# [Product]: Day-90 Readout, One Page

**The answer:** [Same single sentence as the memo.]

**What I found (4 bullets):** [One bullet per audit. Finding plus the single strongest number.]

**The constraints (2-3 bullets):** [One sentence each.]

**The bets (2-3 bullets):** [Bet plus the signal-and-date we check.]

**Not doing:** [Anti-bets, one line.]

**Killing:** [Kill list, one line, with what it frees.]

**Judge me on:** [The scoreboard numbers, baselines, direction.]

[Rule: the one-pager is written AFTER the memo, never instead of it. It compresses conclusions you can defend, it does not replace the defending.]

---

The read-through meeting agenda (60 minutes)

Run it Amazon-style. The memo is read in the room, not presented. Slides hide weak arguments. Silence with a document exposes them.

0:00-0:20  Silent read. Everyone, including you. No phones, no laptops except the doc.
0:20-0:40  Questions on the findings (sections 1-3 only). You answer from evidence,
           citing audits and ledger rows. If a finding gets successfully challenged,
           say so in the room and mark the doc. Defended-when-wrong costs more
           than corrected-in-public.
0:40-0:55  Discussion of the bets and the kill list (section 4). The question for
           the room: "what evidence would change this?" not "do you like it?"
0:55-1:00  Confirm the scoreboard and the next review date. Name who owns what
           by when. Same-day note follows to all attendees.

[Pre-wiring rule: nobody influential reads this memo for the first time in this meeting. Run the pre-wiring checklist from the coalition map worksheet starting 10 days out. The two people whose objection could kill it get the draft first, one on one.]

[Sequence: exec team first, board second. The board version leads with the scoreboard and the asks, and the product detail compresses. Boards fund trajectories, not features.]

---

QA before it ships

• The one-sentence answer at the top survives being read alone, with no context.
• Every finding in section 1 traces to a Confirmed claim-ledger row.
• Every bet has a "what would change my mind."
• Two kill-list items are already dead, publicly, with written explanations.
• The scoreboard matches what you agreed with the CEO.
• The retelling test: give the memo to someone outside the exec team, take it away, ask them to retell the argument. If the retelling does not survive, the structure is broken, not the reader.