I recently headed to Nudgestock 2026 in London for a day of behavioural science, brand thinking, and a lot of conversation about trust. It was a natural follow-on from the latest EmailWeekly newsletter I wrote around proof over promises. We spend a lot of time in CRM talking about journeys, platforms, personalisation, and content, but underneath all of that sits a very human question: why should a customer believe you? Here are the five takeaways that stayed with me.

1. Proof beats promises

Professor Michael Muthukrishna made the point that people believe what organisations do, not what they say. That came through again in a panel with Octopus Energy, Cancer Research UK, and Lloyds Banking Group. The examples weren’t big shiny brand campaigns. They were things like making sure customers can speak to a real person, being honest about what you can promise, and adding friction when it protects someone from fraud.

For CRM, that feels significant. Advertising can make a promise. CRM is often where the customer finds out whether you actually meant it. Every confirmation email, service update, preference choice, and follow-up is another small piece of evidence. Does the brand remember what I told them? Does it use my data in a way that feels useful? Does the experience after the click match what the message promised? Trust is rarely built in one impressive interaction. It’s built through lots of small ones.

2. Friction is not always the enemy

We often treat better digital experiences as shorthand for less friction. Usually that’s right, but not always. Lloyds talked about deliberately adding steps to protect customers from fraud, while Rory Sutherland warned that too much verification can make every customer feel like a suspect.

The better question is not simply ‘how do we remove friction?’ but ‘which friction creates value, and which just creates work?’ An extra confirmation before moving a large amount of money can add reassurance. Asking a customer for information they have already provided you three times does not. Strong journey design is not frictionless for the sake of it. It understands how someone is likely to feel in that moment.

3. AI needs human ownership

Google shared research suggesting that many organisations are stuck between fear of missing out and fear of messing up with AI. One finding stood out: marketers encouraged to use their own judgement alongside AI developed more confidence and ownership than those told to defer to the model.

We hear a lot about keeping humans in the loop. I think the more important point is keeping humans responsible for the loop. AI can find patterns, generate options, and speed up production, but someone still needs to ask: is this useful, is it appropriate, and would we be comfortable explaining why the customer received it? The more AI can do, the more valuable judgement, context, and empathy become.

4. Make value concrete

The Avios session gave one of the most immediately useful CRM examples. Instead of telling members they could multiply their Avios by ‘up to four times’, the team showed a worked example of what their existing balance could become. That change increased conversion by more than 17%.

Simple, but powerful. Customers should not have to do the maths themselves. Brands are often so familiar with their own products that phrases like ‘4x’, ‘5% back’ or ‘unlock more rewards’ feel completely clear. But understanding something is not the same as immediately seeing its value. There is a personalisation lesson here too: the opportunity is not simply to demonstrate that we know something about someone. It is to use what we know to make the experience easier to understand, more relevant or more useful.

5. Looking credible is not the same as being credible

Possibly the most uncomfortable demonstration of the day came from Phill Agnew. For nine months he and Dan Bennett had been publicly referencing a behavioural principle called the ‘stocked gun effect’. The problem? They made it up. More than 10,000 people had apparently encountered it without questioning whether it was real.

It worked because it sounded plausible, fit the language of behavioural science, and came from someone who appeared authoritative. That feels particularly relevant now. It has never been easier to create something that looks convincing. A polished answer is not necessarily a correct one, and a confident recommendation is not automatically a good recommendation.

For marketers, that makes scrutiny more important, not less. Where did this insight come from? What evidence supports it? What are we assuming? And what would make us change our mind?

What does this mean for CRM?

The thread I took away from Nudgestock was not really ‘how do brands make people trust them?’ It was much simpler: how do brands behave in ways that earn trust?

For CRM teams, that means being clear about the value exchange when we ask for data, using personalisation to improve experiences rather than simply show capability, understanding when friction reassures and when it frustrates, and using AI to support better judgement rather than outsource it. Every message is another opportunity to affirm that the brand behaves the way it says it does. Which is why the idea of proof over promises continues to stick with me. We shared more from Nudgestock on the day – you can catch up here.

At ActionRocket, we look at CRM through both the customer and business lens: understanding the behaviours we want to create, the barriers that get in the way, and how strategy, technology, and creative can work together to improve the experience. If you are looking at how to build more useful, trusted customer journeys, get in touch and we can help you work out where to start.