Building a customer maturity scoring system that maps where the customer base stands on product adoption — informing product strategy, AI roadmap priorities, and long-term vision for intelligent service delivery.
I built a scoring system that assesses the maturity and AI readiness of TOPdesk's active customer base (~4,900 customers across 11 branches). The system scores each customer across multiple adoption and capability dimensions, assigns maturity stages, and surfaces patterns in the data.
The primary purpose is to inform product strategy and AI roadmap decisions — understanding where the customer base actually is today so the product team can prioritize what to build, when to ship AI-powered features, and which capabilities need to mature before new functionality will get adopted. The output is an interactive dashboard with benchmarking, filters, and what-if simulators.
TOPdesk is investing in AI-powered features — chatbots, automated workflows, intelligent routing — but needed to understand whether the customer base is actually ready to adopt them.
No way to know which customers had the prerequisite maturity (KB content, automation, integrations) for AI features to actually work.
Product decisions about what to build next lacked data on what the customer base could realistically adopt today vs. in two years.
Unclear which capability gaps (knowledge, automation, integrations) were blocking progression — and therefore which platform investments would have the most impact.
No external context for how the customer base compared to published industry benchmarks — making it hard to set realistic ambitions.
I built a scoring framework using behavioral platform data to create a repeatable assessment.
Identified which platform metrics meaningfully differentiate customer maturity levels — usage patterns, feature adoption, integration depth, and self-service behavior.
Developed a multi-dimensional scoring model with configurable thresholds. Each customer receives scores across 6 capability dimensions (channel usage, knowledge management, automation adoption, integration depth, contextual data, and experience management), producing an overall maturity stage from 1 (Fragmented) to 5 (Ambient).
Used sensitivity analysis to test how threshold changes affect stage distributions. Thresholds are configurable in the dashboard so stakeholders can adjust them.
Built cross-cutting comparisons by product line, region, ARR tier, organization size, and sector.
Built interactive Streamlit dashboards with configurable filters, drill-downs, and what-if simulators.
Prepared presentations and summaries to communicate findings to different audiences.
The maturity model assesses customers across six dimensions, each measuring observable adoption behaviors rather than self-reported perceptions.
| Dimension | What It Measures | Data Sources |
|---|---|---|
| Channel Architecture | How customers distribute intake across channels (portal, email, API, forms) | Platform analytics |
| Knowledge & Content | KB depth, article quality, deflection rates, content freshness | Usage metrics, content analytics |
| Automation Adoption | Workflow automation runs, action sequences, automated routing | Automation logs |
| Integration Depth | Connected systems, API utilization, data flow maturity | Integration registry |
| Contextual Intelligence | Asset data richness, caller history utilization, relationship context | CMDB & asset data |
| Experience Management | Measurement practices, feedback loops, continuous improvement signals | Proxy indicators |
Disconnected channels, no self-service, reactive only. The starting point for most organizations.
Portal exists, basic KB in place, but limited deflection. Form-heavy, minimal automation.
Active self-service, meaningful deflection (15–30%), some automation. Conversational interfaces emerging.
Context-aware, 30–50% deflection, agentic workflows, proactive for specific use cases.
Proactive, self-healing, invisible IT. Largely theoretical in 2025 — the north star for the framework.
The measurement program delivers insights through interactive dashboards designed for different audiences — from executive overviews to detailed account-level drill-downs.
The scoring system provides the evidence base for product strategy and AI roadmap decisions, and also supports customer-facing conversations.
Shows what percentage of the customer base has the prerequisite maturity for specific AI features — informing when to ship and how to sequence the roadmap.
Weakest-link analysis identifies which platform capabilities need investment first — e.g., KB tooling before chatbot rollout, because content is the bottleneck.
What-if simulators model the effect of adoption improvements: "if we lift KB adoption by X%, how many more customers become ready for guided experiences?"
Industry benchmarks (Gartner, HDI, SDI) ground internal targets in external reality — setting realistic timelines for AI feature adoption.
Account-level scores can support targeted conversations about readiness gaps and what would help a customer progress.
Segment-level views (by region, ARR, size) surface where adoption patterns differ — and why.
The project involved working with people across different functions to make sure the scoring was meaningful and the outputs were useful.
Presented maturity analysis to internal stakeholders. Translated scoring methodology into plain language for non-technical audiences.
Discussed scoring thresholds with colleagues who have regional account knowledge to check whether the model made sense against what they see on the ground.
Built dashboards stakeholders can use independently — with filters, drill-downs, and simulators — rather than creating static reports.
Synthesized industry benchmarks (Gartner, HDI, Ivanti, SDI) to give internal data external context.
Measurement framework design, scoring methodology, threshold calibration, sensitivity analysis, behavioral data analysis.
Python, Pandas, Plotly, Streamlit, interactive dashboards, what-if simulation, statistical analysis.
Executive storytelling, audience-appropriate narratives, stakeholder presentations, enablement materials, benchmark contextualization.
Scoring based on what customers actually do (not surveys) produced more reliable signals that didn't require survey participation.
Making scoring parameters adjustable lets people explore "what if" scenarios — making the tool more useful than a static report.
Showing relative performance (vs. peers or industry) prompts more interesting conversations than absolute scores alone.
Insights are more useful when they connect to existing workflows and conversations rather than living in a standalone report.