Before you buy an AI tool, clean your CRM

Everyone wants an AI strategy. Very few people want to clean their CRM first.

I understand why. One sounds like the future. The other sounds like a Friday afternoon of deleting duplicates. But in every business I've looked at, the second one decides whether the first one works.

AI is only as good as what you give it. Give it messy customer data and it gives you confident answers built on mess.

AI tools read your data, not your mind

Most AI tools for business work by reading information you already have. Your leads, your customers, your emails, your past jobs. Then they summarise it, sort it, reply from it or predict from it.

If the CRM says a customer is "John Smith (2)" with no phone number, no source and a status last updated in 2023, the AI doesn't know that's wrong. It just uses it.

That's how you get a tool that follows up people who already bought, misses people who are ready to buy, and reports that your best lead source is "Unknown".

What a messy CRM looks like

You probably have some of these:

• The same customer entered two or three times.

• Leads with no source, so you can't tell which marketing found them.

• Deals sitting in "In progress" for months.

• Notes kept in personal inboxes instead of the record.

• Fields nobody fills in, and fields everybody fills in differently.

None of this is unusual. It's what happens when a system grows without anyone owning it.

A one-week clean-up

You don't need a big project. Spend a week on it:

1. Merge the duplicates. Most CRMs have a tool for this. Start with the obvious ones.

2. Close the dead deals. Anything untouched for 90 days is marked lost, with a reason.

3. Fill in the source on recent leads. At least the last three months.

4. Delete fields nobody uses. Fewer fields means more of them get filled in.

5. Agree the rules. Who creates records, which fields are required, and what each stage means.

The last step matters most. A clean CRM gets messy again in a month without rules.

Then decide what AI should do

Once the data is clean, the AI conversation gets much more useful. You can see where the real gaps are. Slow replies. Leads that go cold after a quote. Customers who never get asked to come back.

Pick one of those and point a tool at it. One problem, one tool, measured properly. That's an AI strategy. It's smaller than the one in the slide deck, and it actually works.

The take

AI doesn't fix messy data. It speeds it up.

Clean the CRM, set the rules, then bring in the tools. The boring week is what makes the clever part pay off.