The experience question organisations keep skipping.

For the last year or so I've been sitting with leaders who are genuinely trying to get AI right, people who are thoughtful, well-resourced, and working hard at a question that nobody has completely figured out yet. And there's a question that almost never gets asked early enough, even in the most considered conversations I've been part of.

Not which tool to use, not how to build the business case, not even how to bring people along, though that last one is at least getting closer to the point. The question that I keep finding myself wanting to redirect conversations back to is this: what experience are you actually trying to create?

For your customers. For your people. For the person on the other end of whatever your organisation does in the world.

I think about this constantly through the lens of human connection, because that's what's ultimately at stake when we talk about where technology belongs and where it doesn't. There are moments in any service, any relationship, any interaction where what the person on the other end genuinely needs is efficiency, the right information, instantly, without friction, delivered in a way that respects their time. In those moments, AI is genuinely extraordinary. It can put exactly what someone needs at their fingertips in seconds, personalise a response to their specific situation, remove steps that were always unnecessary, and do it consistently at a scale no human team could match.

But there are other moments, and in most human services, these are the moments that define what the organisation actually is, where efficiency is precisely the wrong thing to optimise for. When someone is frightened, or confused, or navigating something that doesn't fit any template, what they need is to feel heard. They need someone to understand that their situation is specific and that they as a person matter more than the process designed to handle people like them. They need empathy, not accuracy. Human presence is not a technically correct response. And in those moments, the most sophisticated AI in the world is not just insufficient, it's the wrong tool entirely, and deploying it sends a message about what the organisation values that no amount of good intent can undo.

The organisations getting this right aren't distinguished by the sophistication of their technology or the scale of their AI investment. They're distinguished by the clarity they have about which moments in their service are which kind, and by the deliberate choices they make about matching the right tool to the right moment, rather than assuming that more automation is always better or that the efficiency case is the only case that matters.

The workshop that brought this to life for me

I was working with a leadership team recently, a service organisation that had grown significantly through word of mouth and reputation, but was starting to see something troubling: their growth was bringing cost pressures, their operational efficiency wasn't keeping pace, and the quality of their service delivery was beginning to slip in ways their customers were noticing. They came in with a clear agenda, which included a list of things they wanted to automate and a strong conviction that AI was going to be a significant part of the answer.

We didn't engage with the list straight away. Instead, we walked them through a different set of questions, about the kind of organisation they wanted to be, what great looked like for their customers at each point in the journey, and what their team needed to experience in a day of doing this work in order to do it well. By the time we got back to technology, the conversation had completely shifted. The question was no longer which things could be automated, but what experience they were actually trying to create, and how they could make sure that every decision about process, technology, and people was in service of that rather than a detour from it.

The light bulb moment came toward the end, when someone in the room said something to the effect of: so before we work out what AI can do for us, we need to work out what we're actually trying to be. And once we know that, the roles and the structure and the processes need to reflect it, because that's where most of our inefficiency actually lives, and no amount of automation fixes a structure that wasn't designed well to begin with.

That moment, the shift from individual technology decisions to something much more foundational, is the conversation I wish more organisations were having before they start their AI journey, rather than six months into a plateau.

The sequence that actually works

Experience design first: what are you trying to create for the people you serve and the people doing the serving? This isn't a marketing question, it's an operational and cultural one, and it requires real honesty about the gap between what you intend and what your customers and team actually experience right now.

Operating model second: do your roles, your accountabilities, your decision rights, and your ways of working actually support the experience you're trying to create? Because this is where most organisations discover that the inefficiency they were hoping technology would solve is actually structural, it lives in unclear ownership, in decision pathways that are too slow, in roles that have drifted from their original purpose. Technology applied to a poorly designed operating model doesn't fix the model. It makes the problems faster.

Technology third, and this is where AI belongs, not as the starting point but as the accelerator. Once you know what experience you're designing for and once your operating model is shaped to support it, AI becomes genuinely powerful. It can take a well-designed process and make it faster, more consistent, more personalised, and more scalable than any human team could manage alone. But it can only do that work if the foundation underneath it has been built deliberately.

This sequence is almost never how organisations actually approach technology decisions. They start with what the technology can do, build a business case around efficiency gains, and then discover somewhere downstream that the experience they created isn't quite the one they intended. The sequence matters not because it's philosophically correct but because it's practically faster, organisations that do this work upfront spend less time undoing decisions made in the wrong order.

What experience are you designing for? And is your operating model ready to support it?

#ServiceDesign #AI #OperatingModel #Leadership #HumanCentredDesign #ForPurpose