Frank & Erin · May 2026 · Pre-read for the retro meeting
Pre-read assumption. This memo assumes you’ve read Crescendo Pilot — What We Ran, which covers what the program is, what we taught, who participated, and how it went operationally. This memo is about the decisions we need to make for Crescendo 1 in September.
We need to leave this meeting with answers to seven decisions. The biggest one is whether to reposition Crescendo’s spine around Lyric-specific modeling and composability — at the cost of some storytelling time and a fifth week added to the program. We think yes, but we want to debate it before locking in.
The other six decisions are about scale, price, resourcing, pipeline, instrumentation, and Empower. They’re listed at the end.
The energy and the Finale are non-negotiable. Every survey respondent rated session energy 4/5 or 5/5. Chicago Finale logistics rated 5/5 from everyone. Andy named the in-person day as the single best thing. The Finale is the asset to protect, and the energy is the design choice to repeat.
The named frameworks worked. Students applied Lenny’s 5 in Build 2 and Mike Adams’ Seven Stories in Build 3 — we could see specific frameworks showing up in the Chicago presentations. Sara’s “Lead the horse to water” got cited by name. Mike & Ganesh’s “Humanize it” and “Man in the hole” got used. Naming frameworks made them sticky. We should keep doing this and convert more guest-speaker tips into named, reusable handles.
Peer connection became a real outcome. Teresa wrote that she has “so much to learn about other industries.” Ross built rapport across recurring breakout partners. Three of six want to come back as moderators. This is a marketable outcome — and a moderator pipeline for Crescendo 1.
Storytelling craft is genuinely differentiating. Tommy named the “upper-right engineer” — technical depth plus the ability to communicate. Ross discovered he was a better storyteller than he expected. No other supply chain credential teaches this. It stays.
But the curriculum was lighter on Lyric-specific modeling than students wanted. This was the loudest signal in the survey and in the secondhand feedback. CSM quote: “a few comments of it not being overly technical and didn’t really increase their model building skills.” Ross: “I was a better storyteller than I anticipated. And I was a worse data engineer than I thought I would be.” Composability was cited in four of six survey responses. The word composability doesn’t appear anywhere in the pilot’s session materials.
This is a positioning issue, not just a curriculum issue. The pilot was explicitly framed as a prototyping program — wrap existing science notes in apps and learn to tell the story. Students arrived expecting modeling depth that framing didn’t promise. We need to either change the framing, change the curriculum, or both.
Hands-on AI was underdelivered. AI showed up as an option in assignments — Claude Artifacts as a build path, “Feed AI your work” as a pre-pitch step — but never as a teaching focus of its own. Second most-cited gap.
Deadlines were punishing. Multiple students reported working into the early morning. Tommy’s specific ask: move draft deadline from Wednesday midnight to Thursday noon, peer review window 4–8 hours after that, keep final at Monday noon. Cheap fix.
Frank and Erin did everything. Content, sessions, emails, logistics, follow-up. At 18 students this was hard. At 40 it’s impossible without help. This is the most important operational risk for Crescendo 1.
How do we rebalance the curriculum between storytelling and Lyric-specific technical depth, without losing what made the pilot work?
Storytelling worked. The energy worked. The Finale worked. Some students explicitly loved the storytelling-heavy parts of the program. But the gap students named — and Teresa articulated most sharply — is that storytelling and presentation training is commodity content available everywhere, and Lyric-specific modeling skill is not.
“Less focus on the storytelling and pitch aspect. I love that stuff and it’s incredibly important, but data storytelling and presenting training opportunities are common whereas focused Lyric-specific support is not.” — Teresa Viola
Our proposal: Storytelling stays as a strong supporting element. The spine of Crescendo 1 becomes Lyric-specific modeling and composability — the full sequence of data, pipeline, and optimization, taught with named reusable frameworks, threaded with hands-on AI interaction. The program lengthens from four weeks to five to make room for it.
We think this is the right move both for student outcomes and for Crescendo’s strategic role in the Lyric customer lifecycle. But it’s a repositioning of the program, not just a curriculum tweak — and we want to debate that openly before committing.
Scale. 40 students, capped. Venue capacity is higher but we don’t want to bite off everything at once.
Length. Five weeks, up from four in the pilot. The new week is the modeling and composability week.
Pricing. $800 per student, unchanged from the implicit pilot price-point. Travel and lodging on the customer.
Curriculum changes:
Moderator program (new). Pilot graduates engage as moderators in exchange for free Crescendo 1 access plus the Chicago Finale. Time commitment 2–4 hours per week. Responsibilities: warm up the chat at session start, engage in breakout rooms as the icebreaker, give feedback on student projects, be a friendly face at Finale. Confirmed interested from the survey: Teresa, Tommy, Ross.
Operations at 40 students.
Swag. Real Crescendo gear by September. Coordinate with marketing now.
Guest speakers. Re-invite Sara, Mike, Ganesh, Dinesh. Add a survey question: “Were guest speakers useful?” so we can quantify.
| Pilot | Cohort 1 | Cohort 2 | Cohort 3 | |
|---|---|---|---|---|
| Finale event costs | $6,295 | $8,040 | $10,680 | $10,680 |
| Manhours (Frank & Erin) | $25,000 | $25,000 | $25,000 | $25,000 |
| Total cost | $31,295 | $33,040 | $35,680 | $35,680 |
| Student count | 18 | 40 | 60 | 60 |
| Cost per student | $1,739 | $826 | $595 | $595 |
| Revenue per student | $0 | $800 | $800 | $800 |
| Margin per student | –$1,739 | –$26 | $205 | $205 |
Crescendo 1 at 40 students is effectively breakeven (–$26/student) with a $25K notional load for Frank and Erin’s time. The program crosses into modest margin only when we fill to 60 students at Cohort 2 — and that margin is $12,300 across the whole cohort. The Frank-and-Erin manhours line is the most sensitive: if we hire help to take work off the founders, that number moves and breakeven shifts with it.
Demand for Cohort 1. We’re assuming 40 paid seats fill. We don’t have hard pipeline data yet. If fill is soft, we need to know early and decide whether to discount, delay, or shrink. Need a pipeline checkpoint and a decision rule.
Frank and Erin capacity. Even with moderators, Crescendo 1 will demand more time than two people can sustainably give while doing their other roles. The $25K manhours line is notional — in practice this is founder time we can’t keep paying ourselves with forever. If we don’t resource this differently, quality degrades or one of us burns out. This is the operational risk we most want help solving.
The “unprofitable programs get cut first” risk. Crescendo Pilot operated at a real loss. The cost model shows modest margin only at Cohort 2’s scale. $12,300 per cohort is small relative to what one bad quarter could pressure. We want leadership to see Crescendo as a customer success investment with measurable retention and expansion impact, not as a P&L line reviewed in isolation. Action: instrument account-level tracking now so that by Crescendo 2 we can show real downstream impact data, not just program economics.
These are the decisions we want to leave the meeting having made:
Worth raising in the retro but probably not deciding in the meeting.