Why do AI projects in sales fail?

AI projects in sales rarely fail because of the technology. They fail because AI meets a sales system that wasn't working in the first place — unclear processes, empty CRM data, leadership without a governance model. AI makes a good system faster and a bad system fail faster.

The typical sequence looks like this: a company buys AI licences. Expectations are high because the demo was impressive. After three months, ten per cent of the team uses the tool. After six months, nobody talks about it anymore. The licence keeps running.

The cause is almost never the tool. It lies in three prerequisites that nobody checked beforehand.

First: the data. AI works with what's in the system. If a third of the pipeline has shown no activity for months and closing dates have been pushed back three times, the AI is forecasting on fiction. The result looks precise. It isn't.

Second: the process. AI can accelerate a sales process. It cannot replace one that doesn't exist. If every region maintains its own templates and every salesperson has their own method, AI automates the chaos — just faster.

Third: leadership. An AI tool changes workflows. Anyone who expects behavioural change without leading it gets training certificates instead of usage. The goal isn't knowledge. The goal is the ability to act in day-to-day work.

That's why an AI project in sales doesn't begin with tool selection. It begins with a question: does our sales system — people, processes, data, governance — work well enough that acceleration helps? If yes, AI is a lever. If no, the sequence is wrong.

I've experienced CRM rollouts at Dell and Cisco in sales leadership roles and today accompany AI implementations in sales. The pattern has stayed the same for 25 years: technology fails where the system behind it doesn't hold up.