Skip to main content

Project roadmap

This page is about how I'd run a content program for a NimbusWiz-shaped product, and the framework I'd use.

info

This is a method demo, not an actual operating plan.

Treat documentation as a product

Documentation has users with jobs to be done. My first step is to define these segments precisely enough that content decisions can be argued for or against.

For NimbusWiz, three segments use the docs differently:

SegmentJob to be donePrimary concernHow they read
DevOps EngineerGet a system through the pipeline without it failing in productionRisk: what can go wrong, what the rollback options areSkips to the relevant stage; uses the API reference as a working document
System AdministratorKeep the fleet healthy and catch problems before they escalateVisibility: what the metrics mean, when to actReturns to the same pages repeatedly; uses the KB during incidents
Technical ManagerMake and defend the upgrade-vs-replace decisionDecision support: confidence scores, tradeoffsReads overviews; arrives via the Dashboard

Each segment implies different content investments. By naming these segments, I aim to avoid over-investing in whichever audience my writer-avatar most resembles.

Prioritize on effort and impact

Score every backlog item on two axes:

  • Effort is relative to the existing baseline: low (one page or a section rewrite), medium (a new multi-page section or structural change), or high (a new doc suite or major IA change).
  • Impact is estimated from the analytics signal driving the action: ticket volume, search failure rate, drop-off rate, or feature-usage-vs-doc-views gap.

The scoring isn't precise, but names what each item is for. See Content Analytics Insights for worked examples of the items with impact signals.

Instrument what to measure against

The NimbusWiz program is organized around outcomes, so three things are planned or in place before the first content investment:

  • Page-level analytics on every doc page (Plausible, GA4, PostHog, take your pick).
  • Search analytics on the doc-site search bar, including zero-result queries.
  • A funnel metric that ties docs to product outcomes. For NimbusWiz, that would be Time-to-First-Success: time between first docs visit and the first system completing the full Register-to-Monitor pipeline.

The point of the funnel metric is how the number moves when content investments ship, not the absolute number.

Phases, not features

The docs program for NimbusWiz runs in phases, each defined by what the prior phase made possible.

PhaseGoalDefining work
BaselineKnow the numbers before optimizing for themImplement analytics, set search baselines, publish the dashboards
Critical gapsFix the content with the loudest signalsAddress the top unresolved searches and the highest drop-offs
Persona optimizationTailor by segmentSplit entry points by audience; add contextual links from product to docs
Retrieval-augmentedLet users ask questions instead of searching for pagesRestructure the corpus for AI consumption; ship a RAG chatbot with an evaluation framework

Each phase produces a measurable change in the funnel metric. A phase that doesn't move the metric is a signal to recalibrate the plan.

What I'd do first in a real engagement

Spend a few weeks in instrumentation and observation before a single content change ships. It is crucial to know what you're optimizing for before you start optimizing.