Agentic CRM:
The Five-Step Method:
How the Best Brands Turn Their Customer Data Into a Growth Engine

Let’s get the AI part out of the way first. We’ve spent eight years working shoulder to shoulder with the best engagement and data technology in the world. Braze , Hightouch , Snowflake and Databricks. They are extraordinary companies moving faster than anything we’ve seen in this space.

Here’s what that means in practice. Agentic capability will be baked into every one of them. Agents that draft campaigns, agents that build segments, agents that optimise send times. Within a year or two, none of that will differentiate anyone, because everyone will have it. When every team has the same engine, the engine stops being the advantage.

You won’t win by picking the right platform. AI has made output easy. Outcomes are still hard. That’s the part nobody is selling marketing teams, and it’s the only part worth worrying about.

This is why so many CRM programmes disappoint. Teams judge them by the visible layer, the emails and the push notifications, and invest accordingly. But the visible layer is downstream of everything else. If the data underneath is wrong, every campaign built on top of it is wrong too, just faster and at greater scale. Most agencies send the emails. The real work is growing customers, and that work starts several layers below the inbox.

For eight years this has been the engine behind 200+ CRM implementations for some of the largest global enterprises in the world, across the US, Europe, APAC, and LATAM. We never wrote it down, because it was how we won. Now we’re putting it in the open. We’ll publish the whole method, step by step, one piece at a time.

Five steps. Fix the data: get a single, trustworthy view of the customer into the warehouse, because nothing downstream works without it. Fix the reports: agree what a customer is worth and how you’ll measure change, before you send anything. Plan campaigns that make money: start from the revenue opportunity in the data, not from the content calendar. Run the campaigns: ship them properly, with the testing discipline to know what actually worked. Show the money: report incremental revenue in language a CFO accepts, so the programme funds its own next phase.

It’s a loop, not a funnel. Step five feeds step one, because every campaign you run teaches you something about your customers that makes the data richer and the next plan sharper. The same method applies whether you’re building from scratch, migrating platforms, or scaling a programme that already works. Only the entry point changes.

Here’s the part most of the market has backwards. Agents don’t replace the method. They run it. In a traditional programme, humans execute every step of the loop: pulling the data, building the reports, planning, shipping, measuring. The loop turns as fast as your team’s capacity allows, which for most brands means quarterly at best. In an agentic programme, the humans set the strategy and the guardrails, and agents turn the loop continuously. A lapsed-customer opportunity that used to wait for next quarter’s planning cycle gets spotted, actioned and measured in days.

That’s why the method matters more in the agentic era, not less. An agent executing a sound process compounds it. An agent executing a broken one just breaks things at machine speed.