AI decisioning in CRM

Make sure AI is learning the right things.

The data you give AI shapes every decision it makes. Here’s what needs to be in place and why fixing it helps your programme right now.

Download the AI Readiness Checklist

The AI risk in CRM isn’t where most people are looking.

 

The debate often centres on platforms, tools and autonomy.  Discussing which AI will make decisions, how much control to hand over, and which model is most sophisticated. These are real questions… eventually. But they’re not where the risk is.

The risk is in the data. AI doesn’t know whether the signal you give it is good. It will work with whatever it has and do so with complete confidence. Which means the questions that actually matter isn’t “are we ready for AI?,” it’s “is our data good enough for the decisions it would make?”

For most organisations, the answer is: not quite. But closer than you think.

The same work fixes two problems.

The foundations that make AI decisioning possible are the same ones that make your CRM programme better right now. You don’t have to choose between solving today’s problems and getting ready for AI.

  • Better identity resolution means less wasted spend on already-converted or lapsed customers today
  • Fresher data means your decisions reflect what customers are doing now, not three weeks ago
  • A complete exposure history means your models aren’t working from gaps
  • Causal measurement means you know what’s actually driving results, not just what correlates

That’s where this starts.

Four foundations that fix today’s problems and get you ready for AI.

These are the data foundations that determine whether your CRM programme makes good decisions today and whether AI decisioning is possible tomorrow. You don’t need to perfect all of them at once. But ignoring any of them creates blind spots that compound.

Customer identity resolution

Who is this customer across every touchpoint and channel?

Data latency & freshness

How current is the data driving your decisions?

Event & exposure history

Does your data show what customers have been exposed to, not just what they did?

Causal measurement

Can you prove what your campaigns actually caused, not just what happened near them?

Reduces wasted spend on already-converted, lapsed or misidentified customers today. And without a solid identity layer, any AI model is working with fragments and will be very confident doing so.

Campaigns respond to what customers are doing now, not last month. A model trained on old data will optimise for a customer who no longer exists.

Prevents your measurement from crediting activity that was going to happen anyway. Without this context, every model is learning from an incomplete picture.

Makes the trading meeting more credible and budget conversations stronger. And without incrementality, AI optimises for patterns rather than impact which is the difference between a tool that reports and one that learns.

Start wherever you are.

Whether you want a quick read, a self-diagnostic, or something to go deeper with, there’s a starting point here for every stage of the conversation.

The Decisions Behind the Decisions

Start here if you want to understand why we think most of the AI in CRM conversation is focused on the wrong thing. A straight read, no commitment.

Read the post →

AI Decisioning in CRM — Full Session Recording

The full recorded session from our recent event series on AI decisioning in CRM. Covers the four foundations, the causal measurement argument, and a worked example of how learning-led decisioning delivered 40%+ uplift in incremental revenue.

Watch the recording →

Data Foundations in CRM

A practical guide on the four foundations that determine whether your data supports better decisions today and AI decisioning tomorrow. What good looks like, what gets in the way, and what to prioritise first.

Download the guide →

AI Readiness Checklist

Five minutes. Four foundations. How many warning signs does your team recognise? A self-diagnostic tool with what good enough looks like, the common warning signs, a sanity-check prompt for each, and one pragmatic improvement you can make now.

Download the checklist →

Creating the Learning Layer

The measurement framework that fixes today’s CRM problems and builds the learning AI depends on. Covers causal measurement, how target and control actually works in practice, and the case study showing 40%+ uplift in incremental revenue.

Download the eBook →

Machine Learning Playbook for CRM

For CRM teams who’ve moved past the ‘what is AI’ conversation and are thinking about how to actually build and deploy models within their existing programme. Practical, practitioner-level, no transformation project required.

Download the playbook →

40%+
uplift in incremental revenue

A large multichannel retailer used a causal learning layer to reprioritise their campaigns based on true incremental value, identifying which customers genuinely drove additional revenue and suppressing activity that was being credited for outcomes that would have happened anyway.

The result wasn’t from sending more. It was from making better decisions about who to contact and when. That’s what the right measurement infrastructure makes possible.

See how the Learning Layer works

Want to talk about how this applies inside your organisation?

If you’re thinking about where your programme stands across these foundations, we’re happy to spend 30 minutes talking it through. No agenda, just a conversation with practitioners who think about this every day.

Get a fresh perspective