Organisation: ECITB (Engineering Construction Industry Training Board)

Sector: Engineering construction, including nuclear, renewables, oil and gas, water and pharmaceuticals

Scale: 300 organisations in scope, 190 live on the platform, 30,000 users a year, 60,000 course completions a year

ROI: Over £2 million saved for industry last year. E-learning grant payouts cut from £1.3 million to zero.

The Challenge

ECITB is an arm’s-length public body. It collects a training levy from engineering construction employers and grants it back when they train their people. Before working with Filtered, most of that levy funded classroom-based learning. When the pandemic closed training providers overnight, ECITB needed a way to deliver learning online, at scale, across some of the most safety-critical industries in the UK.

The bigger problem sat underneath that. ECITB’s team could see thousands of pieces of content across the market, from SCORM packages to YouTube videos to industry blogs, but they had no fast way to work out which pieces were relevant to a given skill or standard. Every request for a learning pathway meant days of manual searching and cross-referencing before anyone could act on it.

The Solution

ECITB started with a proof of concept: an algorithm-built course, assembled from video and web content against a training standard in four days. A course like that would normally take 12 to 20 weeks to build. In this way, Filtered assembled a ‘Digital Passport’ for 12 ECITB customers at speed. That result gave ECITB the confidence to build a full business case for a different way of working: the full digitisation of its learning offering. 

From there, the relationship grew alongside what became Filtered Intelligence:

Ingest Content. Filtered’s algorithm searched and assembled learning content from across ECITB’s own catalogue and partner content purchased through Filtered’s marketplace, breaking training standards into the key points a learner needed to cover.

Map Skills. ECITB is now taking this further, breaking job roles down into tasks and the skills each task needs. This work feeds their national occupational standards and gives them a live map of what a role like pipe fitter or rigger actually requires, rather than a generic job title.

Signal Quality. ECITB built their own content hub, called the Content Search Index, and asked Filtered to expose the relevance score through an API. Every piece of content in their system now carries a red, amber or green rating, so users can see at a glance what’s worth their time.

Connect AI. The next phase already underway is conversational search. Instead of an admin searching a meta description, an end user will describe the skill they need and content will build a pathway in response. ECITB plans to put this directly into the hands of every worker, not just system administrators.

The Results

• 12-20 weeks cut to 4 days. The build time for a full training-standard course in the original proof of concept.
• 3-7 days cut to instant. ECITB’s old manual curation turnaround, before their API integration with Filtered put relevance scoring directly into their own platform.
• £1.3 million grant to £0. ECITB paid out £1.3 million in e-learning grants in 2020/21. By year two of working with Filtered, they had stopped paying e-learning grants altogether because the content already existed and was easy to find.
• £2 million+ saved last year. ECITB’s outgoing costs stayed under £1 million while they saved industry over £2 million, an independently audited result. They are forecasting £3-4 million in savings this year, with costs holding flat.
• Team size held steady. ECITB’s learning team has not grown despite the platform now serving 30,000 users a year, because the work that used to need three or four extra people is handled by the platform instead.

In ECITB’s Words

“We couldn’t have got to where we’ve got to without what you’ve been a part of. That’s true for everybody who’s part of this journey.”

“I honestly think it would have took probably three or four different people to do the job of what the algorithm was doing.”

Steve Rick, Head of Learning, Design and Technology, ECITB

What’s Next?

ECITB is pushing further into task-level skills data, rebuilding national occupational standards with input from employers across the sector, and preparing to give every worker direct, conversational access to the skills search that used to sit with a small admin team. As Steve puts it, teaching and learning is just the vehicle. Skills are the goal, and understanding the tasks behind them is what makes AI adoption possible in industries that cannot afford to get training wrong.