Why does AI in sales achieve nothing if the CRM doesn't work?

AI in sales works with the data held in the CRM. If the CRM is maintained incompletely, outdated or dressed up, the AI analyses a fiction — and delivers recommendations that look precise and are wrong. A functioning CRM is therefore not a preliminary stage of AI implementation, but its condition.

The problem is old; AI only sharpens it. In many sales organisations, the CRM is a control instrument: the team enters what management wants to see. The customer shows green because green doesn't trigger questions. They cancelled anyway. As long as a human reads this data, their experience silently corrects the gaps. An AI has no such experience. It takes the data literally.

This leads to an uncomfortable chain: lead scoring on bad data prioritises the wrong customers. A forecast on dressed-up stages predicts revenue that never arrives. A next-best-action based on outdated activity recommends the next step for a relationship that's long dead. In all three cases, the AI isn't broken. It's doing exactly what it was told.

The solution isn't a data-cleaning project. Anyone who only cleans the data will have the same gaps again in six months, because the cause runs deeper: CRM isn't used if it gives the salesperson nothing. A CRM works when it's designed and led as a tool for sales — when the salesperson looks at it in the morning because it helps them, not because they're being monitored. Then usable data emerges by itself. And only then does AI on top of it pay off.

I introduced CRM systems at Dell and Cisco in sales leadership roles long before AI was a topic. The insight from back then holds truer than ever today: CRM is not an IT project. It is a leadership project. AI doesn't change that — it only raises the price you pay for getting it wrong.