Episode 92 - How to actually deliver one-to-one CRM

Welcome to episode 92 of The Retention Blueprint.

CRM has made the same promise since the day it was invented.

The right message, to the right customer, at the right time.

It's on slide three of every strategy deck. 

Every vendor has said it. 

No one has ever actually delivered it.

I want to talk about why that’s finally changing. 

Because for the first time in my career, with agents, it is finally buildable. 

What "one-to-one" actually means

When I say one-to-one, I need to be precise, because the phrase has been worn out.

I don't mean a first name in the subject line.

I don't mean personalised journeys.

I don't mean cohorts.

I don't mean a few personalised content blocks driven by a handful of attributes.

Those are all fine. 

They're also all still group-level. 

And they get built into journeys that are the same for everyone. 

To be truly one-to-one means:

1. Every customer gets a decision made for them, at every moment. No more fixed journeys. Individual-level decisions. 

2. When or whether a message gets sent is determined at an individual level. 

3. What you send for that customer at that moment. The right next action to move this customer's behaviour wether it is to reactivate, deepen, reassure, or reward. A drifting customer and a stuck-but-loyal customer need opposite things.

4. The moment of truth. Intervening at the individual moments that decide retention. The failed payment. The onboarding stall. The first session after a big win. The first sign of disengagement.  

5. Real-time attributes. Content assembled from the customer's state at the moment of send. What they did an hour ago, not what the overnight batch knew about them last Tuesday.

6. Timing and channel. The right send-time and the right channel for this individual: email, push, in-app, SMS, on-site coordinated, so your channels stop competing for the same person's attention.

Do all six, per person, automatically, and you've moved from running clever campaigns to running a system that quietly makes a hundred thousand right calls a day.

Why is this possible now (and wasn't two years ago)? 

Short answer: Agents. 

How this works

The agent carries pointers; it reads the customer’s state and attributes. It picks an approved action. It emits a short, structured instruction: this customer, this action, this channel, now. It never stores content. It never writes content. It selects.

Where all the content comes from

If every customer gets something tailored, where does the content come from? You can't hand-write a thousand variants every minute, every day.

And you can't let a model free-write them at send time; that's too high-risk. 

The answer is what I call a second brain.

It's a governed knowledge base that the AI always reads from first: your CRM and retention principles, behavioural science, and your own IP,  brand voice, product and catalogue knowledge, offers, tone, and compliance rules.

Because every piece is grounded in that brain rather than invented from nothing, you can produce enormous volumes of content that's on-brand, principled and genuinely good, and every variant traces back to a principle and a source.

One brand I've worked with recently went from a single image-based email a week, the same creative to lapsed and active customers, with only the CTA changed, to over 2,400 personalised communications across multiple cohorts and moments for the World Cup. This was achieved in under two months. With my two-person team, plus the client.  

The part your ExCo actually cares about

Now, the bit that decides whether any of this survives contact with your ExCo. 

The first question you'll get is: "So, how do I track who got what? And why did it message Sarah six times?"

In CRM 2.0, the journey on the canvas was your audit trail. In CRM 3.0, that artefact is gone, because the path is different for every customer. If you try to eyeball a million journeys, you'll wrongly conclude the whole thing is unauditable.

But you're just auditing the wrong layer.

Stop thinking of it as a campaign. Think of it as a trading desk. You audit the limits they can't breach, the record of every trade, and the surveillance on the whole book. Same here:

• The envelope: a per-customer frequency cap and fatigue budget that lives outside the agent and which it physically cannot exceed. The question stops being "why six messages?" and becomes "what was the cap, was it sensible, can we prove nothing breached it?"

• The ledger: an immutable, versioned record of every decision: customer, time, action, channel, the signal that triggered it, the checks it passed, and the version of everything. You can reproduce exactly what anyone received on any date.

• The distribution: the system is designed to message people unequally, so the real audit is statistical. You watch the shape of contact across the whole base, and you catch both failure modes: hammering your most responsive segment, and going silent on one you've quietly abandoned. In a regulated business, that second one is critical for compliance.

Underneath all three: a governed objective, a permanent control group, a circuit breaker, and a human-in-the-loop on the edges.

Governance, done this way, is the thing that lets you put your foot down.

Isn't this just Optimove or Bloomreach? 

No, but it works perfectly well with them. The execution platforms are getting genuinely good agentic features. 

But in most orgs, to truly deliver 1-1 CRM, your agents need to be able to read unstructured data, spread across multiple platforms, all while operating within regulatory constraints.  

So yes, you still need Optimove or Bloomreach or whatever you use, that's what you execute from, and you can use their AI features to make things work better. 

But the agent lives on your cloud platforms and informs your execution platforms. The reason: total access to all of your data that exists across all your diverse systems, even if it's unstructured. 

So that the agent makes a call to your execution platform every moment and builds one-to-one journeys on the fly, all while being fully governed using approved content. 

Gartner predicts that 60% of brands will be doing CRM this way by 2028.  

And the role that follows is a better one.

Less time in internal debates about priority and frequency.

More time defining moments of truth, setting guardrails, shaping the objective, owning the edge cases.

Same team. Different job. Bigger impact.

This is exactly what I'm teaching,  live and free on Thursday 25 June: Activating Agentic CRM 3.0: how to build one-to-one CRM you can actually govern. The architecture, the second brain, and the governance model, with live Q&A. It's open to anyone now.

👉 Save your seat 

One last thing: if you want to keep seeing this in your inbox, just click a link or hit reply with a hi or a question, even a question about the event. I read and reply to every message, and it tells your email provider that this is useful. If it's not for you, unsubscribe; no offence taken.

Until next time,

Tom

—

Tom Burrell is a retention strategy consultant with 26 years' experience, formerly SVP at DAZN with senior roles at Manchester United and Digitas. He works with ExCo teams in streaming, iGaming, sports and subscription on structural retention strategy and agentic CRM design. After several successful client engagements, Tom is fully launching his agency, Retention AI, this summer. More to follow.

P.S. What did you think of this episode?

😎 It was great!

🤷 It was ok

🫤 Could be better

Book a call: When you are ready, reach out to discuss your specific agentic CRM ambitions.