How to sequence a Claude rollout across multiple teams
Rolling out to 10 teams at once is a recipe for chaos. Here is the sequencing strategy that works — who to start with, how to expand, and what to do when a team is struggling.
If you have been asked to roll out Claude company-wide, the instinct is often to do it all at once — provision everyone's accounts, set up all the Projects, send the announcement, and call it done. This approach reliably produces mediocre results: some teams get meaningful value, most don't change how they work, and after three months someone asks why adoption is low.
The alternative is sequenced rollout. It takes longer up front and produces much better results.
Why sequencing works
Every team's Claude Project needs a system prompt, uploaded documents, and configured Skills and Connectors. Writing good system prompts requires understanding what the team actually does. That understanding takes time to develop.
If you roll out to all teams simultaneously, you have to write all the system prompts at once — which means writing them quickly, before you understand the work, based on assumptions rather than observation. The result is Projects that kind of work for everyone and really work for no one.
Sequential rollout lets you go deep with one or two teams, get their Projects right, learn what actually works, and carry those learnings into the next wave.
The sequencing framework
Wave 1: Your highest-value, clearest use cases (weeks 1–3)
Start with the teams that have:
- High-volume, repetitive text work (CS, marketing, or operations typically)
- A clear, specific use case (not "use Claude for everything" — "use Claude to draft ticket responses")
- A team lead who is genuinely interested, not just compliant
Go deep with 1–2 teams. Set up their Projects properly. Run the first use case well. Measure the time savings. Get testimonials.
Wave 2: Adjacent teams with similar patterns (weeks 4–8)
Once wave 1 teams are using Claude consistently and producing results, expand to similar teams. If CS worked, HR and operations typically have similar patterns (high-volume, document-based). If marketing worked, sales and comms usually follow.
Critically: use wave 1 learnings. The system prompt pattern you developed for CS probably applies to HR with minor modifications. The documents structure that worked in marketing applies in sales. You are not starting from scratch.
Wave 3: Complex or specialised teams (weeks 9–16)
Engineering, legal, finance — teams where Claude use cases are real but require more careful configuration and stronger guardrails. Use the credibility from waves 1 and 2 to have those conversations from a position of demonstrated results, not theoretical promise.
Who to start with in wave 1
The three questions to evaluate each team:
- What specific task would take meaningfully less time with Claude? (Can they name one?)
- Does the team lead believe it will work? (Skeptical leads will undermine adoption)
- Can you measure the impact? (Even roughly — time per task, volume handled, revision rounds)
Teams that score well on all three are your wave 1 candidates. Teams that can't name a specific task are not ready yet — spend the time helping them identify one, then return.
What to do when a team is struggling
At 30 days, if a team is not using Claude or reporting poor outputs, diagnose before making changes:
Low adoption usually means the workflow is unclear. Talk to users, not the manager. Ask: "Show me the last time you tried to use Claude and it didn't work." The answer will tell you what to fix — usually the system prompt, the onboarding, or the use case definition.
Poor output quality usually means the system prompt is wrong or the Project is missing key documents. Compare what Claude is producing to what you would want. The gap between those is the gap in your configuration.
Abandonment after one bad experience needs individual follow-up. Find the person who tried it, got a bad result, and stopped. Walk through it with them. Show them what the right approach looks like. One successful session after a bad one often recovers the relationship.
The measure that matters most
At 90 days, the question is not "how many people have used Claude?" It is "which teams have meaningfully changed how they work, and what did that change deliver?"
Aim for two or three teams with clear, measurable before-and-after stories. Those become the internal case studies that drive the next wave of adoption better than any company-wide announcement.