Stage 7 · Master
Platform Engineering Fundamentals
Platform as a Product
Apply product management practices to your internal platform: user research, roadmaps, versioning, SLIs, and developer NPS.
Why Product Mindset?
Without product management, platform teams become ticket queues. A product mindset shifts from 'what do stakeholders ask for?' to 'what do developers need to be productive?' It introduces discipline: roadmaps, versioning, deprecation policies, and success metrics tied to developer outcomes.
| Ticket Queue Mindset | Product Mindset |
|---|---|
| Reactive: wait for requests | Proactive: research, anticipate needs |
| Success = tickets closed | Success = developer productivity improved |
| No versioning, breaking changes anytime | Semver, 18-month support, migration tooling |
| Docs as afterthought | Docs as part of the product |
| No deprecation policy | Clear sunset dates, migration paths |
| Platform team = cost center | Platform = leverage multiplier |
Platform Product Manager
The Platform PM owns the platform roadmap, prioritizes based on developer impact, and acts as the voice of the developer. They don't write code — they write specs, talk to users, and measure outcomes.
- Conducts developer interviews, shadowing, friction logs
- Defines platform SLIs (scaffold time, API latency, CI queue time)
- Prioritizes backlog using RICE (Reach, Impact, Confidence, Effort) weighted by developer count
- Runs quarterly planning with platform team and stakeholder representatives
- Communicates roadmap, releases, deprecations via changelog and portal announcements
- Owns developer NPS survey and action plan
Start with 1 PM per 8-12 platform engineers. Before that, a tech lead acts as PM. The PM should have engineering background — they need to understand the technical constraints to make good trade-offs.
Developer User Research
- Friction Logs: New hires document every 'wait, how do I...' moment during onboarding
- Shadowing: PM sits with developers for 2 hours/week, watches them work
- Quarterly DX Survey: 10 questions on tooling, docs, speed, pain points
- In-App Feedback: 'Was this helpful?' buttons on portal, CLI, error messages
- Support Analysis: Categorize #platform-help Slack / Jira tickets — top categories = roadmap
- Usage Analytics: Which golden paths used? Which APIs called? Which docs viewed?
Roadmap & Prioritization
Platform roadmaps balance: new capabilities (golden paths, APIs), reliability (platform SLIs), developer experience (DX improvements), and tech debt. Use a framework:
Feature: Ephemeral Preview Environments
Reach: 50 developers (all backend teams)
Impact: 3 (high - saves 30 min/PR, reduces merge conflicts)
Confidence: 0.9 (proven at other companies)
Effort: 4 (2 engineers × 6 weeks)
RICE = (50 × 3 × 0.9) / 4 = 33.75
Feature: Custom Dashboard Builder
Reach: 10 developers (frontend only)
Impact: 2 (medium - nice to have)
Confidence: 0.6 (unproven)
Effort: 8 (2 engineers × 12 weeks)
RICE = (10 × 2 × 0.6) / 8 = 1.5
→ Prioritize Ephemeral Environments
RICE forces explicit assumptions and enables comparison across disparate features.
Platform SLIs & SLOs
Platform SLIs measure the platform's health from the developer's perspective. They are distinct from service SLIs (which measure user-facing reliability).
| Platform SLI | Target | Measurement |
|---|---|---|
| Scaffold Success Rate | >99% | % of platform scaffold commands that produce a working repo |
| Scaffold-to-Deploy Time (p50) | <30 min | Time from scaffold to first production deploy |
| Platform API Latency (p99) | <500ms | API gateway latency for mutating operations |
| CI Queue Time (p95) | <5 min | Time from push to CI start |
| Preview Environment Provisioning | <3 min | Time from PR open to preview URL ready |
| Golden Path Drift Detection | <24h | Time to detect template drift in managed repos |
| Platform API Availability | 99.9% | Uptime of control plane APIs |
Feedback Loops
- Daily: Platform team reviews error rates, latency, failed scaffolds
- Weekly: PM reviews top 5 friction log items, support tickets
- Bi-weekly: Platform demo to stakeholders, gather input
- Monthly: Platform SLI review, error budget burn rate
- Quarterly: Developer NPS survey, roadmap planning, deprecation announcements
- Annually: Platform strategy review, org structure assessment
Publish a monthly changelog: new features, bug fixes, deprecations, migration guides. Distribute via portal announcement, Slack, email. Developers should know what changed without asking. Example: 'v2.4.0: Added Redis capability to Go golden path; deprecated v1 Go template (sunset 2024-06-01); fixed CI queue timeout bug.'
Mark this lesson complete to store local progress and unlock a cleaner resume path the next time you visit.