Subscription product advisory
Subscription PM. Prior work shipped ~$1.7M at Match, ~$1M at HelloFresh, and lifted LTV/CAC 2–3× at Found. Ships production tools with AI coding agents when the wedge calls for it.
Performance Health · Pre-Series B · 3–4 engagements per year
Next availability: Q3 2026
Field note 01 · the pattern behind the numbers below
2020 · HelloFresh
$15M+ protected
2021 · Match
$1.7M incremental
2022 · Thirty Madison
+11% DTC conversion
2023 · Found
2–3x LTV/CAC
2024 · TaxAct
+18% add-on conversion
2025 · Social-impact crowdfunding platform
+20% tip rate
How it runs
Cohort teardown, funnel walk, and stakeholder interviews. Output: a written read of where retention actually leaks and why.
One bet, chosen against the diagnostic — not the loudest voice in the room. Spec, success metric, kill criteria, all written before code.
Weekly operating cadence with your team. I write the tickets, run the reviews, and unblock the calls. Instrumented before launch.
Read of what moved, what didn't, and the runbook your next PM inherits. If it worked, we talk about the second wedge. If not, we don't.
Questions people ask before the intake
Series A through early B, subscription revenue model, Performance Health category (fitness, longevity, mental health, nutrition, sleep). Product-market fit signal exists but retention is soft. Pre-Series A only if there's a paying subscription base to instrument.
Fixed monthly retainer, quoted after the intake. No hourly billing, no scope creep line-items. 1099 contract — no benefits, equity, or overhead loaded onto the number. Comparison against a loaded FTE cost lives on the /engage page.
The diagnostic takes two weeks. Shipping one real wedge takes six to eight. Anything shorter is a workshop, not an engagement. The notice window protects both sides — you can exit at any monthly boundary after month three, I get enough runway to close cleanly.
Complement, usually. If there's a Senior PM in seat, I take the first-90-days wedge and hand it back. If the role is open, I run product until the hire lands and then hand off. I don't want to be your permanent Head of Product.
Three options: extend for a second wedge (some do), transition to a lighter advisory cadence, or hand off clean and go our separate ways. The handoff doc is written whether we continue or not.
The scope
Included
Not included
Recent writing
All writing →Never let a model grade its own work. I gave my AI build pipeline a planner, an engineer, an adversarial reviewer on a different model, a design critic, a brand critic, and a QA agent. Here's what the setup caught, what it missed, and the rule that only exists because it missed.
Read →Hours of podcasts and feeds for one usable insight. So I built a scheduled AI task that does the extraction overnight and ships a six-section briefing to my inbox by morning. The build decision that mattered: a written spec, not a clever prompt.
Read →Built an AI prototype for aiEDU almost entirely with AI. The product thinking that held up, the bugs that didn't, and the loop that turned every break into a permanent rule.
Read →Not a fit if: no data infrastructure · outside product (marketing, ads, upper-funnel) · Shopify work.