Case study · Operating model · Whole portfolio · 2023–2026
One front door for demand: owning intake across a ~$10M portfolio, setting commitments with the CTO and delivering eight annual roadmaps in a row
Head of Product · company board member · 30-person product org at peak · 2023–2026
- 8consecutive annual roadmaps delivered
- 30people in the product org at peak: 23 engineers in 3 teams, 4 PMs, 3 analysts
- ~$10Mportfolio revenue, 100–120K orders a year
Situation
A ~$10M portfolio of three client-facing products and the two-sided freelance marketplace that supplies them, with 100–120K orders a year from the US and Canada. Sales, Operations and engineering all brought requests to the same three engineering teams.
Complication
Demand always outran capacity, and requests arrived through every channel at once. Without one place to compare them, priority went to whoever escalated, and a change in one product could quietly move another: a client-side price change shifts what freelancers will accept.
Decision
One front door, one ranked list, one plan with engineering. Every request from Sales, Ops and engineering came through a single intake and was placed on the same impact-effort matrix, then prioritised, deferred or declined. Delivery commitments were set jointly with the CTO, so what product promised and what engineering could build were the same plan.
Exhibit
Every request came in through one door and left with a decision
↺ Front-line feedback: from Roadmap back into One intake
- Requests: Sales · Operations · Engineering · Front-line feedback
- One intake: Every request placed on impact × effort
- Decision: Prioritise · Defer · Decline
- Commitment: Set jointly with the CTO
- Roadmap: One ranked roadmap for the portfolio
- Front-line feedback loops back into One intake.
What I did
- 01
Owned demand intake. Requests from Sales, Ops and engineering were prioritised, deferred or declined in one place. Each request was a one-page brief (the problem, the evidence, the metric it should move and a rough engineering estimate), placed on an impact-effort matrix and triaged weekly.
- 02
Set commitments with engineering. Delivery commitments were agreed jointly with the CTO, and I sat on the company board with the founders and department heads.
- 03
Fed the front line into the roadmap. Chat-log bottleneck analysis, a sales QA system, client surveys and interviews turned field problems into roadmap items.
- 04
Pushed prioritisation down as the team grew. Once each product had its own PM, the matrix set the portfolio-level bets and PMs ranked features within their product with RICE, so the same logic ran at two levels.
- 05
Kept the portfolio shipping through a restructuring. When a 2026 restructuring removed the product team, I ran the whole portfolio and its roadmap alone, working directly with engineering, design and data, until the same restructuring closed my own role in September 2026.
Exhibit
Quick wins shipped first, big bets were planned with engineering, and anything costly with little impact was declined
↑ Impact→ Effort
High impact · low effort
- 6% price increase
- First-order discount
- Quarterly reactivation campaigns
- In-app upsells
High impact · high effort
- Quality pre-checks and verification layer
- Specialist supply in worst-hit subjects
- AI-report add-on
- Repositioned subscription
Low impact · low effort
- Post-purchase offers
- Refund caps and per-operator limits
Low impact · high effort
Requests that landed here were declined or deferred, with the reason written down
- Quick wins: 6% price increase (low x, high y)
- Quick wins: First-order discount (low x, high y)
- Quick wins: Quarterly reactivation campaigns (low x, mid-high y)
- Quick wins: In-app upsells (mid-low x, mid-high y)
- Big bets: Quality pre-checks and verification layer (high x, high y)
- Big bets: Specialist supply in worst-hit subjects (high x, high y)
- Big bets: AI-report add-on (mid-high x, mid-high y)
- Big bets: Repositioned subscription (mid-high x, mid-high y)
- Fill-ins: Post-purchase offers (low x, mid-low y)
- Fill-ins: Refund caps and per-operator limits (mid-low x, mid-low y)
Impact
- Eight consecutive annual roadmaps delivered.
- The results in the other cases came through this system: refunds 15% → 7%, client LTV +15%, fraud under 10 cases a month.
- Every request got an answer, prioritised, deferred or declined, with the reason written down.
Learning
A roadmap is mostly a record of what you’ve agreed not to do.