When the client asked for customer onboarding training, the data found a process that existed in no document anywhere.
Tools: AI-Assisted Data Analysis · AI-Powered Solutions · Claude Code · Claude Design · Performance Consulting
A fast-growing SaaS startup was losing customers in the first 90 days and came to me wanting better CS onboarding training. The analysis revealed something different: a team that had scaled faster than the systems built to support it, and a customer onboarding process that existed only in the heads of two or three tenured reps.
Sample project. This is a fictional case study created to demonstrate my performance consulting process and instructional design approach. The client, data, and scenario are hypothetical.
THE PROBLEM
The ask was about training. The data pointed to infrastructure.
Streamlyft came to me with a clear presenting problem: customers were churning in the first 90 days, and the CS team needed better onboarding training. Before recommending anything, I slowed down and asked better questions.
What those conversations revealed was a team that had doubled in size in under a year, and had no shared process, no reliable product knowledge source, and no mechanism for passing on what its most experienced people knew.
From the discovery conversations:
Is there a documented customer onboarding playbook (defined stages, standard milestones, consistent touchpoints) that all reps are working from?
“No. We’ve talked about building one for well over a year. It just hasn’t happened. Everyone has been too busy keeping up with the volume to stop and build the infrastructure.”
Have you identified a point in the customer journey where you tend to lose customers most often?
“Weeks two and three seem to be the danger zone. Customers get through the kickoff call fine and then go quiet. Our working theory is that they hit friction during implementation, and reps aren’t checking in at the right moment to catch it.”
Has the team tried anything already to address this? What happened?
“We ran a half-day product training session about six months ago. It helped reps feel more confident talking about the product, but the churn numbers didn’t move. That’s part of why I’m not convinced this is primarily a training problem.”
WHAT I FOUND
Three structural gaps. None of them could be fixed with a training course.
No customer onboarding playbook existed
There was no defined process for what a CS rep should do from the moment a customer signed. No stages, no milestone touchpoints, no escalation triggers. Each rep improvised their own approach, and the data showed it.
New CS reps were set up to discover the job rather than learn it
After minimal shadowing, new reps were assigned live customer accounts and left to find their footing. The behaviours that drove retention, such as proactive outreach, anticipating friction, and knowing when to escalate, had never been documented or taught. They existed only in the heads of the most tenured reps.
Product knowledge had no reliable infrastructure
Product updates arrived through fragmented channels. The internal wiki was distrusted. 67% of CS reps surveyed said they did not know where to find reliable product information when customers asked.
61% of all first-90-day churn occurs between weeks 2 and 4 · 131 customer onboarding engagements across 6 months
Customers whose reps made 4 to 5 proactive check-ins in weeks 1 to 3 churned at 10.7%. Customers whose reps made 0 to 1 churned at 54.5%.
“First 90-day churn cut in half, at minimum. And a CS team that feels equipped and confident - not reactive and scrambling. Right now, we’re losing customers at exactly the point where we should be winning them.”
THE SOLUTION
Not a training programme. An infrastructure rebuild, in two phases.
Phase 1 builds the process and product knowledge infrastructure that the team is currently missing. Phase 2 activates it through training and real-time performance support.
Phase 1: Build the Infrastructure
Customer Onboarding Playbook
BUILT (AI-ASSISTED PROTOTYPE)
A stage-by-stage guide covering the full 90-day customer onboarding window: defined milestones, week-by-week touchpoints, escalation triggers, and success criteria for day 30, 60, and 90. Designed as a dual-purpose document - a companion workbook during training and an on-the-job reference guide afterward. First draft developed using AI, grounded in the discovery findings and root cause analysis.
CS-Friendly Product Guide and AI-Assisted Update Process
RECOMMENDED
A plain-language translation of the Product team's technical documentation, generated and maintained using an AI tool. When updates ship, the guide reflects them, and a Slack notification goes to the full CS team. New reps read through it in their first two weeks and complete a product knowledge walkthrough with their team lead before going live on accounts.
Phase 2: Enable and Support the Team
Rise-Style AI-Generated E-Learning Module: The Customer Onboarding Journey
BUILT (AI-ASSISTED PROTOTYPE)
A six-lesson scrollable e-learning module walking new CS reps through the full 90-day customer onboarding journey. Each lesson uses stacked content blocks: day-by-day action checklists, accordion call agendas, at-risk signal callouts, and scenario-based knowledge checks. Designed to be completed alongside the Customer Onboarding Playbook. Built using Claude Design.
AI Product Knowledge Chatbot with Live Assist Mode
BUILT
A Claude Code prototype giving CS reps instant product answers in two modes. Standard chat: rep types a question, gets an answer from the product guide in seconds. Live Assist: the browser microphone listens to the live customer call, transcribes in real time, and proactively surfaces relevant answers on the rep's screen - no typing, no hold time, no interruption to the customer.
Manager-Led Product Knowledge Assessment
RECOMMENDED
A 30-minute walkthrough conversation at the end of each new CS rep's second week. The team lead runs the rep through three or four product scenarios as if they were live on a customer call. No formal test - a practical conversation that surfaces gaps before they surface in the field.
DELIVERABLES
What was built
Customer Onboarding Playbook (AI-Assisted Prototype)
PDF · 16 pages · Dual-purpose training workbook and job aid
Rise-Style AI-Generated E-Learning Module: The Customer Onboarding Journey
Interactive module · 6 lessons · Built with Claude Design
AI Product Knowledge Chatbot with Live Assist Mode
Live prototype · Built in Claude Code · Standard Chat + Live Assist
GO DEEPER
Full case study
The complete narrative covers every discovery question and client response, the full AI-assisted data analysis with all three charts, and the detailed reasoning behind each solution component.
📄 Full case study narrative
PDF · Includes all discovery Q&A, mock dataset analysis, AI prompts used, and complete solution rationale
