The BCG formula is uncomfortably simple: 10 per cent algorithm, 20 per cent data and technology, 70 per cent people, processes and culture. What this means in practical terms for your organisation, and where most people stumble at first.
April 2026 · Noted by · reads in ~2 minutes
BCG didn’t invent the formula, but it did popularise it: in AI projects, 10 per cent is down to the algorithm, 20 per cent to data and technology, and 70 per cent to people, processes and culture – these factors determine whether the project succeeds or fails.
I have tested this formula in around 30 projects. It holds true. In most cases, it’s even more pronounced: that 70 per cent feels more like 90 per cent.
People need to trust the system. Processes need to be adapted – not the tool to fit old processes. And the culture must allow for things to go wrong occasionally, without the whole approach being called into question.
That sounds soft. It isn’t. In every organisation where an AI project has really worked, I’ve seen a manager who has clearly communicated this: ‘We’re giving this a go. We’re learning. We’re adapting.’
With the prompt. It’s easy to write a good prompt. It’s hard to write a prompt that will still work six months down the line, when the tone of the inbox has changed, when new staff have been trained, or when a campaign has gone differently than expected.
That’s a process problem, not a technical one. And process problems take time, repetition and sometimes even uncomfortable conversations.
Before you write your first prompt: talk to the team. Not about AI. About the work. What’s the routine that takes up the most energy? What’s the part that’s genuinely no fun? That’s where the first lever lies.
45 minutes, one screen shared, an honest assessment. Not a sales pitch, I show you where I see leverage and where I do not.