AI Can't Fix What You're Not Tracking

A business owner recently asked me to help her add AI-powered personalization to her email marketing. Smart idea, on paper. Then I looked at her data setup and had to give her an answer she didn’t expect: not yet.

Her CRM had no reliable way to tell which customers had actually purchased versus just browsed. Her email platform tags were inconsistent. There was no clean record of purchase history tied to individual contacts. Adding AI personalization on top of that wouldn’t have created smart, tailored emails. It would have created confidently wrong ones, at scale, automatically.

AI amplifies what’s already there

This is the part that gets lost in a lot of AI marketing hype. AI doesn’t fix bad data. It executes on bad data faster and with more confidence than a human would. If your tracking is broken, AI-driven decisions built on top of it will be broken too, just delivered with more polish and less obvious hesitation.

Feed a model clean, well-structured data about what customers actually buy and when, and it can find patterns a person would take weeks to spot. Feed it a mess, and it produces a mess dressed up to look like insight, which is arguably worse than an obvious mess, because it’s harder to catch.

Garbage in, garbage out was true before AI existed. AI just made the garbage move faster and look more convincing on the way out.

What “foundation” actually means here

Before layering AI into marketing, the questions worth asking are less exciting than “which AI tool should I use” and more foundational:

• Is customer data structured consistently across every platform that touches it?
• Can you actually trace a purchase back to the marketing touchpoint that drove it?
• Is the data clean enough that a pattern found in it reflects reality, not just noise?

If the honest answer to any of those is no, that’s the actual starting point. Not because AI isn’t worth using. Because it’s not worth using yet, on top of a foundation that can’t support it.

What I told that client

We spent three weeks fixing her tracking and data structure before touching any AI tooling. It wasn’t the exciting part of the project, and it’s not the part that makes for a flashy case study. But once that foundation was solid, the AI personalization work that followed actually worked, because it had real, trustworthy data to learn from.

If you’re considering adding AI to your marketing and you’re not sure your data foundation can support it, that’s worth checking before you spend money finding out the hard way. Happy to take a look.