The Staged Journey

Problem

Organizations try to jump from anxiety to ROI in one move

They don't. In the same department, one person hasn't opened the tool because they're worried about what it means for their job. Another uses it every day for small things. A third is building workflows nobody else understands. Treating them as one group means the plan fits almost no one: too fast for the anxious, too slow for the builders, and silent on how to get from one to the other.

The numbers show the spread. In PwC's 2025 survey of nearly 50,000 workers in 48 countries, 54% had used AI in their role in the past year, but only 14% used generative AI every day. And the gap matters: 92% of daily users reported real productivity gains, compared with 58% of people who used it only occasionally.

Then comes the messy middle. Usage is uneven, early wins are small, and the returns leadership was promised haven't shown up. Without a map of the stages, that looks like failure - so the program gets cut, rebranded, or replaced with the next tool, and everyone starts over.

I lived this building this library. Before I was working from patterns, producing a single pattern took all day. Once I was practicing them - briefing the AI, interviewing myself, red-teaming, running work overnight - the same work took a couple of hours. Same tools, same person. The difference was the stage.

Research on change has said this for decades, in two ways. Everett Rogers' work on how innovations spread shows that people start at different times - a few early, most in the middle, some late - not all at once. The Concerns-Based Adoption Model, developed for school change in the 1970s, shows that people need different help at each stage, as their concerns move from "what is this?" to "what does it mean for me?" to "how do I manage it?" to "how do I make it better?" A plan that ignores where people are gets resistance, not adoption.

The messy middle isn't failure. It's a stage. The failure is not knowing which stage you're in - and nobody owning the move to the next one.

Value

Every team gets the next step that fits where it is

The messy middle gets read correctly. Leaders can tell normal progress from a real problem, instead of cancelling a program that was working. The tell: in a healthy messy middle, usage is spreading and early wins are repeating. In a failing program, usage is flat and the same few people are doing everything.

Moving a stage changes the economics. Building this library, one pattern took all day before I worked from patterns, and a couple of hours after. Same tools, same person.

Fewer people get left behind - or held back. Individuals, not just teams, sit at different stages, so one team can run at two speeds: anxious beginners get safety and small wins while experienced users get room to build. Nobody waits on a one-size plan.

Investment follows readiness. Money and attention go to the teams ready to use them, not spread evenly across teams that can't yet.

Value expectations are honest. Each stage has its own measure of progress, so early stages aren't judged by late-stage ROI.

Every next step has an owner. Each move has a named person and a date, so the plan is a commitment, not a slide.

Measure it: share of each team at each stage, reviewed quarterly; time for a team to move from one stage to the next.

Context

When this pattern applies

It applies at any size: a department, a PMO, a whole enterprise. It matters most where people start from very different places, which is almost everywhere.

It is not a maturity model for scoring teams against each other. The stages describe where a team is so it can get the right next step - not a ranking.

Solution

Name the stages, place each team, and give each the next step

Step 1

Expect the messy middle - and say so up front.

Tell leadership before launch that there will be a stretch of uneven results between experimenting and integrating. Naming it in advance turns it from a surprise into a milestone - and keeps the program from being cancelled while it is working.

Step 2

Name the stages.

Use a small number of plain stages everyone can recognize:

AnxiousUnsure what AI means for their work; not using it, or using it in secret.
ExperimentingTrying it on small personal tasks; results vary.
PracticingUsing a few patterns repeatably on real work.
IntegratingAI built into team workflows, with records and review.
LeadingShowing measurable outcomes, and helping others move.

Some stages are concerns (Anxious) and some are behaviors (Integrating); that's deliberate - people move through both. How long a stage takes varies by domain. Content creators and coders don't move at the same pace, because the work, the tools, and the risk of a wrong answer differ.

Step 3

Place each team - and each person - honestly.

Ask, don't assume. Use a structured assessment; for example, the GO Framework's Personal AI Readiness Assessment scores people across five dimensions. Expect different people on the same team to sit in different stages.

Step 4

Give each stage its next step, with an owner and a date.

The next step, not the final goal - and a named person who delivers it by a set date.

Anxious and ExperimentingSafety and one small win. Ways to get there: pair a power user with an anxious user for one task a week; embed an expert's judgment so beginners get trusted guidance (Pattern 14, Embed the Expert); start with an interview so the AI works from what they know (Pattern 05, Avoid AI Slop with an LLM Interview).
PracticingShare what works (Pattern 08, Assemble the Team) and find first real value in invisible work (Pattern 13, Automate the Invisible).
IntegratingRecords and governance (Pattern 12, Ghost in the Workflow).
LeadingMeasures that hold up (Pattern 11), and helping the next team move.

Step 5

Measure each stage by its own yardstick.

Don't judge an experimenting team by ROI.

Stage	What to measure	What good looks like
Anxious	Who has tried it; stated confidence	Most people have tried one task and know it's allowed
Experimenting	How often people use it; what for	Use is weekly and spreading beyond the first few
Practicing	Repeatable patterns in use on real work	Early wins repeat, and people share what works
Integrating	Workflows with records and review	AI assets have owners and records (Pattern 12)
Leading	Cycle time, quality, value	Measurable outcomes leaders can see, and other teams adopting

Notes

Signs you skipped a stage

A mandate with no training. Training with no follow-up patterns. Workflows built by a few people that nobody else understands (ghosts - Pattern 12). ROI questions asked of teams still experimenting.

The people side of the journey. Redesign the Stream, Not the Role (coming soon) redesigns the work; this pattern paces the people through it. The two work together.

Related. Pattern 11 covers what to do with freed time at the later stages; Pattern 12 keeps records at the Integrating stage; Pattern 09 helps people move up a stage; Pattern 06 - people move at different speeds.

Sources. Everett M. Rogers, Diffusion of Innovations (Free Press, 1962; 5th ed. 2003). The Concerns-Based Adoption Model: Hord, Rutherford, Huling-Austin and Hall, Taking Charge of Change (ASCD, 1987). PwC, Global Workforce Hopes and Fears Survey 2025 (49,843 workers, 48 countries and regions; released November 12, 2025).