Data, Analytics & Reporting

Turning raw data into actionable insight for learning design, operational improvement, and stakeholder reporting.
Richard Williams at his desk with the IKON Academy Moodle learning platform visible on screen, working on digital learning data and analytics

Power BI · PowerShell · Selenium · SurveyMonkey · Moodle Analytics · Arlo TMS · SharePoint · Kirkpatrick-Phillips Evaluation · Cambridge Spark Level 4 Data Analysis

Power BI Learner engagement dashboards at UCL and IKON Training
Selenium UCL IT-approved automated Moodle data extraction pipeline
KP L1–L4 Kirkpatrick-Phillips evaluation applied at IKON Training
Level 4 Cambridge Spark Data Analysis qualification

Overview

Data work here is not about dashboards for their own sake. Every piece of data work described on this page was driven by a real question: Are our learners actually engaging? Are the automations working? Is our training changing behaviour? What does the feedback tell us that we aren't already hearing?

The tools vary - Power BI, PowerShell, Selenium, SurveyMonkey, Moodle analytics - but the purpose is consistent: to make data visible to the people who need it, in a format they can act on, without requiring them to be data analysts.

Analytics work has been carried out in two distinct professional contexts - IKON Training (2022–24) and UCL STEaPP (2024–present) - with different tools, data sources and access permissions in each. These are documented separately below.

UCL STEaPP: Moodle Engagement Tracker (2026)

UCL STEaPP Analytics Channel logo - the Power BI reporting channel for the STEaPP Moodle Engagement Tracker

At UCL, access to the Moodle API via web-token is restricted to the central IT team - meaning direct API-based data extraction is not available to departmental staff. With the approval of UCL IT, I built an practical IT-approved alternative: PowerShell scripts using Selenium-based web automation to scrape learner activity and completion data from Moodle course pages. This data is written into Dataverse, where it is modelled, stored and connected to a Power BI dashboard for programme teams.

The Architecture
  • PowerShell scrapers with Selenium extract activity completion and last-access data from Moodle
  • Data is written to Dataverse with a structured entity model (learners, modules, enrolments, activities, completions)
  • A model-driven Power App provides an administrative interface for managing module configuration
  • Power BI reads from Dataverse to provide a refreshable, stakeholder-facing dashboard

See the UCL STEaPP Moodle Engagement Tracker page for full system detail.

Power BI dashboard showing STEaPP module completion bar charts and engagement gauge

The Power BI reporting layer - module completion rates and engagement overview for UCL STEaPP programme teams.

IKON Training: Attendance, Learner Analytics & Feedback (2022–24)

At IKON, I held site administrator access to Moodle, meaning learner data could be extracted directly via API calls - without requiring web scraping. I also built a Canvas App and a lightweight CRM to manage learner data, though both were superseded once the Arlo TMS was fully implemented, which centralised all learner management in a single platform.

Attendance & Engagement Monitoring

A Power BI and PowerShell-based attendance monitoring system was built to automate data extraction and visualisation. PowerShell scripts handled scheduled data extraction, feeding into Power BI dashboards that gave operational staff a real-time view of attendance patterns and engagement metrics.

Key Features
  • Automated data extraction via PowerShell on a scheduled basis
  • Power BI dashboards providing visual summaries of attendance and engagement
  • Alert flags for attendance patterns requiring action
  • Stakeholder-facing views requiring no technical knowledge to interpret
  • Drilldown from programme level to individual learner level
Attendance Dashboard
Attendance monitoring dashboard showing attendance data across a training programme with visual indicators

Attendance and engagement overview - designed to surface patterns and flag learners requiring follow-up, without the viewer needing to interrogate raw data.

Learner Analytics & Time-on-Task (IKON, 2022–24)

Learner Time Tracking
Chart showing average time learners spent on the platform, broken down by pre-course and post-course activity

Average time spent on the platform, broken down by pre-course and post-course activity - used to identify engagement patterns and inform course design decisions.

What the Data Told Us

Tracking time-on-task within Moodle revealed patterns not visible through completion rates alone. One learner spent significantly longer than the cohort average on a particular section - which prompted investigation and ultimately led to a redesign of that module's navigation and content structure.

Pre-course and post-course activity patterns were tracked separately to understand whether learners were using the platform for preparation, revision, or both - informing decisions about where to place key resources and how to sequence content.

Using Analytics to Drive Design Decisions

Data was used prospectively to improve design, not retrospectively to judge learners. High dwell times in specific sections prompted specific, targeted improvements to content clarity, navigation and structure.

Feedback Analytics & Kirkpatrick-Phillips Evaluation (IKON, 2022–24)

Redesigning Feedback Collection

Standard end-of-course surveys produce data that is easy to collect and hard to use. Working with the IKON training team, the feedback survey was redesigned using the Kirkpatrick-Phillips methodology - moving beyond satisfaction scores (Level 1) to capture learning (Level 2), intended behaviour change (Level 3), and perceived organisational impact (Level 4). This involved presenting to weekly team meetings, proposing changes to survey question design, and implementing the redesigned survey across the full course portfolio.

The result was a feedback instrument that supported ROI conversations with clients, not just internal quality review.

SurveyMonkey Integration & Data Flow

SurveyMonkey was integrated into the broader automation ecosystem, with survey data feeding into the admin reporting workflow via Power Automate - using both the native connector and a custom API integration. Learner feedback data was used to monitor trends across the course portfolio, identifying dip-points in learner satisfaction and feeding those findings back into content and delivery design cycles.

Automated Data Extraction: Selenium & Web Scraping

Where platform APIs were unavailable or insufficient, automated data extraction was built using Selenium - a browser automation tool commonly used for testing but equally applicable to scheduled data extraction from web-based systems. This approach was used at both IKON (where some data sources lacked API access) and at UCL (as the practical IT-approved alternative to direct Moodle API access).

Data Pipeline Design Principles
  • Extract once, use many times: data extracted for one purpose was stored in a reusable format for other reporting needs
  • Prefer native connectors over custom extraction where they meet the requirement
  • Build in failure handling: extraction scripts that fail silently produce worse outcomes than ones that fail visibly
  • Keep humans in the loop for data that informs consequential decisions about learners or clients
  • Document the pipeline: a dashboard that no-one can maintain is a liability, not an asset

Interactive Visualisation: Feedback Analysis (IKON)

An interactive Bokeh visualisation was built to explore learner feedback patterns over time - enabling exploration of IKON training feedback data by date, course type, trainer and satisfaction dimension in a way that static charts cannot support. This visualisation was built as a project for the Cambridge Spark Level 4 Data Analysis qualification, demonstrating applied data analysis and visualisation skills in a real professional context.

Note: The automation that refreshes this chart with new data was discontinued after leaving IKON Training. The visualisation below shows a snapshot of data from that period.

For a detailed case study of the full UCL learning analytics pipeline, see the UCL Moodle Engagement Tracker. For the automation infrastructure that generates this data, see Workflow Automation & Solution Engineering. For AI integrations within the data pipeline, see AI for Learning & Assessment.