Roadmap Example — Industrial
Annual roadmap planning at Northfield Automation for the NFA-500 control platform.
Why hardware changes roadmapping: tooling has lead times measured in months, certification windows are fixed by a third party, channel partners commit inventory a year out, and the previous generation is still under service contract. The roadmap has to carry the old product's wind-down alongside the new one's growth.
Example: NFA-500 annual roadmap
Context: Second year of NFA-500 availability. The NFA-200 replacement play is underway — End of Sale is set for March 2027 with service obligations running to December 2028. Eight channel partners resell both.
Phase 1 — Gather Inputs
- Business goals: NFA-500 to 60% of new orders (from 35%); free the NFA-200 manufacturing line by Q2 2027; hold service margin through the wind-down
- Customer problems: field diagnosis time; firmware lag across the installed base; integrators carrying three controller families because the range doesn't cover their bid spread
- Technical: bootloader rollback for remote update; I/O module range gap at high channel counts
- Regulatory: UL 508A reassessment required if remote update alters the listing — assessment pending, gates everything downstream
- Channel: partners need last-time-buy quantities for the NFA-200 by Q1 2027 to plan inventory
- Constraint nobody could move: injection tooling for a new module housing is a 14-week lead time, and the tooling vendor's next open slot is August
Phase 2 — Define Epics
- E1 — Remote firmware update (fixes installed-base fragmentation; gated on UL)
- E2 — On-unit fault isolation (the diagnosis-time problem; validated in discovery)
- E3 — High-channel-count I/O module (closes the integrator range gap; needs new tooling)
- E4 — NFA-200 End-of-Sale execution (not a feature, but it consumes real product and channel
capacity — see
eol-checklist) - E5 — Integrator configuration reuse (reduces commissioning time; software only)
Phase 3 — Prioritize
- Ran
prioritization-advisor. RICE was abandoned within ten minutes: it kept ranking E3 highly while ignoring that missing the August tooling slot pushes it a full year - Switched to constraint-first sequencing — order by what has an immovable external date, then
prioritize what's left:
- E3 tooling decision must be made by June regardless of its RICE score (14-week lead + August slot). Decide early or lose a year.
- E1 blocked until the UL assessment returns. Cannot be scheduled, only reserved.
- E4 has a fixed external date — EOS March 2027, and partners need numbers by Q1.
- E2 and E5 are software-only and genuinely reschedulable. They are the only real flexibility on the board.
Phase 4 — Sequence
| Quarter | Committed | Notes |
|---|---|---|
| Q1 | E4 (partner last-time-buy, EOS comms) · E2 (fault isolation) | E4 is date-driven; E2 is the validated customer win |
| Q2 | E3 tooling decision + order · E2 ships | Tooling ordered by June or E3 slips to next year |
| Q3 | E1 (if UL clears) · E5 | E1 reserved, not committed |
| Q4 | E3 module launch · E4 (EOS execution) | Tooling arrives ~14 weeks post-order |
- E1 is shown as reserved, not committed. Sales was explicitly told "reserved" means it may not happen this year
- E5 is the designated shock absorber. When something slips — something always slips — E5 moves first, and everyone knows that in advance
Phase 5 — Communicate
- Internal: the constraint logic, not just the sequence. Engineering needed to know why the tooling decision outranks a higher-scoring epic
- Channel partners: E4 dates and last-time-buy windows, 25 days before public notice. They plan inventory on this
- Customers: E2 and E5 by quarter; E1 as "in development, no date," because promising a date gated on a third-party assessment is how you lose credibility
- Explicitly not communicated: E3's tooling risk. Internal decision, no customer-facing commitment until tooling is ordered
What this example teaches that the SaaS roadmap can't
- A scoring framework was tried and abandoned, on the record. RICE ranked E3 highly while blind to the fact that a June decision date controls a Q4 delivery. When external lead times dominate, constraint-first sequencing beats scoring — and knowing when to drop the framework is the skill.
- Some items can be reserved but not committed. E1 depends on a regulatory answer nobody at Northfield controls. Putting it on the roadmap as committed would be fiction; leaving it off would hide real planned capacity.
- The retiring product is on the roadmap. E4 ships no features and consumes real capacity across product, channel, and service. Roadmaps that only show new work systematically under-plan the quarter a sunset lands in.
- One epic is designated the shock absorber, publicly. Naming E5 as the thing that moves first means slippage becomes a known plan rather than a renegotiation.
- Different audiences get different roadmaps, deliberately. Partners get dates they plan inventory against. Customers get "in development, no date" for the gated item. Nobody outside hears about the tooling risk. That isn't spin — it's matching commitment level to what each audience will act on.