Stream-aligned
The default
- Owns a single customer value stream end-to-end
- Most teams should be this one
Org, ProdOps & Roadmap
Org design isn't the org chart — it's the fit between strategy, work, people and structure; cascade OKRs by rank and you quietly break it.
Empowered teams, subtractive ops, and a roadmap that speaks three languages — held together by congruence, not the reporting line.
One shared Objective · inherited verbatim
“New teams reach real value in their first session.”
Squad A
28%→42%
Activation rate
Squad B
9d→3d
Time-to-value
Squad C
61%→70%
Retention
The anti-cascade tell: if squad KRs read like sub-totals of one area KR, you've cascaded by rank — not empowered. Distinct KRs under a shared Objective is where empowerment becomes real.
Leaders own Objectives (what problems) · teams own Key Results (how measured)
Above the fold
The three org moves that matter most.
01
Give teams problems, not features.
The Product Operating Model turns feature teams into empowered ones — but the hard part isn't the diagram. Empowered teams need stronger managers and coaches, not fewer.
02
Share the Objective verbatim; author distinct KRs.
Leadership supplies stable inputs; squads set their own goals under a shared Objective. KRs that read like sub-totals mean you've cascaded down the org chart.
03
Diagnose the misfit before you reorg.
Performance follows congruence among work, people, structure and culture — not any one box. Leverage is the biggest misfit, not the loudest symptom.
Product Operating Model · Cagan/SVPG, Transformed
Seven elements turn feature teams into empowered teams.
Feature teams are told what to build; empowered teams are given problems to solve. Skip this and “be product-led” stays a platitude.
- 1 Staffing
- 2 Vision
- 3 Team topology
- 4 Strategy
- 5 Team objectives problems, not features
- 6 Discovery
- 7 Delivery
Calibration trap · empowered in name only
A team called “empowered” but still handed features underperforms a well-run feature team. Empowerment needs stronger managers and coaches, not fewer — and standing up platform or enabling teams before any stream team is starved for them is the classic premature-structure error.
Team topology · Skelton & Pais, Team Topologies
Four team types, chosen by what the team owns.
Platform
Lowers cognitive load
- Self-serve capabilities that lower other teams' cognitive load
Enabling
Temporary embed
- Temporary specialist embed — ML, security, a11y
Complicated-subsystem
Deep tech
- Deep tech: ranking, video, pricing
AI-era extension — the Shipyard 6-function squad (the Udezues, Building Rocketships 2025; hedged prediction) collapses EPD into one small cross-skilled pod where AI absorbs the handoffs. Treat as an option for greenfield AI teams, not a reorg mandate. Hedged prediction
ProdOps · Perri & Tilles, Product Operations
Three pillars — kept subtractive.
Hire ProdOps when scale (multiple product trios) makes insight and rituals leak — not before. It builds the system PMs run; it doesn't run the roadmap.
Business / Data Insights
Pillar 1The data and analysis layer product trios lean on to decide.
Customer / Market Insights
Pillar 2Shared research and market signal so every trio isn't re-gathering it.
Process & Practices
Pillar 3The rituals and tooling PMs run on. Keep it subtractive (Cutler) — kill rituals whose friction costs more than they return.
Boundary: ProdOps is not program management and not a PM-task dumping ground.
OKRs · beating the cascade trap (Cutler)
Don't cascade objectives down the org chart.
Leadership supplies stable strategic inputs — vision, a few bets, guardrails. Teams set their own goals against them. Under a shared Objective each squad commits a distinct KR; do not split one KR across squads.
Planning loop = Product Kata
PerriDirection → current state → next target condition → squads choose the problems → experiment.
Confidence scoring
decouple from comp5/10 baseline · 9–10 = sandbagging · ~70% hit = success. Decouple grading from comp to kill sandbagging.
Before you lock
DoshiPublish “What we're NOT doing this quarter”; pre-mortem the draft; gut-check candidate objectives against DHM.
Cadence
Set
Quarterly
Review
Bi-weekly
Score
Monthly
Roadmap design
One roadmap, three audience cuts.
A roadmap is a communication and alignment tool, not a dated commitment — outcome-based Now / Next / Later. Audience-segment it (Mironov); never hand sales the internal version.
Board
The bets
- Bets + outcomes
Sales
Rough timing
- Themes + rough timing
Internal
Full detail
- The detail — never handed to sales
Treat items as experiments, not promises; for AI bets run a capability funnel — many probes narrow to the few that clear the quality bar before earning a slot. External roadmaps carry a no-commitment disclaimer.
Layered roadmap architecture · Phaal / Probert / Farrukh (IfM Cambridge, T-Plan)
Under Now/Next/Later sits a why / what / how canvas.
Now/Next/Later is the communication format; underneath, three perspectives link on one timeline (now → short → medium → long → vision). Most PM roadmaps draw only the product layer. Very high
Know-WHYMarket / business
Drivers, strategy, needs — the demand / market-pull side. Reads top-down (pull).
Know-WHATProduct / service
Form, function, performance — the bridge between why and how.
Know-HOWTechnology / resources
Solutions, capabilities — the supply / technology-push side. Reads bottom-up (push).
Without the market-why above and the tech/resource-how below, you can't tell whether a feature ladders to a driver or whether the capability to ship it even exists. Align why / what / how before sequencing Now / Next / Later.
Org-fit diagnostic · Nadler-Tushman Congruence Model (1980)
Diagnose the misfit before you reorg.
Strategy is the input, performance the output. Between them, four components must fit each other — performance follows from congruence among the four, not from any one box being “good.”
Input
Strategy
WorkThe actual tasks / JTBD
PeopleSkills, motivation
Formal structureRoles, topology, process, metrics, incentives
Informal orgCulture, norms, real power
Output
Performance
Pairwise fits: work ↔ structure · structure ↔ culture · people ↔ work
Leverage = the biggest misfit, not the loudest symptom
Re-tuning incentives while work and culture stay misaligned just relocates the dysfunction (systems-thinking). Classic tell: a strategy change with no matching change to the four → execution stalls.
Aumayr org integration — in mature orgs split Strategic PM (market strategy, pricing, positioning) from Operational PM (backlog, delivery); PM can report to Marketing / Eng / GM / standalone (each a tradeoff); define RACI for every interface (Sales, Mktg, R&D, Ops, Finance, Legal); the advanced model runs PM as a profit-center with P&L ownership.