Is friction in a system always 'sludge'?

Episode transcript

Amelia (00:16)
I'm Amelia.

Paul Wood (00:17)
I'm Paul.

Russell (00:18)
I'm Russell.

Amelia (00:19)
And this is Fin the Week. How's everybody doing? How are we this week?

Paul Wood (00:24)
Good, yeah, looking forward to some more sunny weather, which has continued since last week.

Russell (00:30)
Yeah, same. It's beautiful over here in Guernsey today.

Amelia (00:33)
Hopefully more of that as we head into the Easter break. And of course this week, a bit of a smaller panel, but we have got a cracking topic. We'll be asking whether friction is always a bad thing. But before we get into it, Paul, have you ever accidentally bought something with one click that you didn't mean to?

Paul Wood (00:52)
Yeah, I was thinking about this. I've definitely done it on Amazon more than once. They make the shopping process so seamless that most of the time it's a pleasure because you just don't have to think about it. But they've got that Buy with One Click button that's caught me out more than once. I've accidentally bought something that I did intend to buy eventually, but you've ordered it before you know, or you've ordered the wrong colour or something like that.

Or the thing I do most commonly on Amazon actually is I order something to the wrong address. Luckily, I notice it in time, and their system is well-built enough that you can immediately go back in there, change the delivery address, and there's no harm done. But then, as I was thinking about this question, I remembered recently I had to buy a new car insurance policy, and that process caught me out.

I went through a comparison site and, as part of the search settings I put in, I specifically ticked the box to remove the, what do you call them, telematic policies. So the policies that will put what they call a black box in your car, or that they use your own phone data to monitor your driving habits.

I deselected that option, so I didn't want those policies. When I was reviewing the quotes that came back, I found one that looked suitable, clicked through to the actual insurer's website to finish the process, and the insurer's website presented me with two different routes. There was the default route that was already selected, which was something called, I think they called it an app policy, and then the other route was a different named policy.

The fault I made was I didn't spend the time reading the page in enough detail. I just assumed that the app policy was a policy that was slightly cheaper because you manage everything via the company's app. You don't speak to their team or anything, you just sort of manage it yourself.

That's the way they pitched it on the site, to be fair. I certainly didn't read the small print, but the headline was that. So I thought, that sounds fine, I'm happy using an app. So I went ahead with that. And then a little while later, I got an email that said, thanks for purchasing a new policy. Once we've got this system set up and running, you can then install the app so that we can monitor your driving. And then, if you drive well you'll get a cheap price forever, and if you don't… it posed the idea that if you don't drive well your price won't be quite as good next year.

I just thought, that seems sneaky, because it really wasn't that clear when I went through the process. So I suppose that counts as an example where the process to purchase it was actually pretty smooth. But my natural habit is to just scan-read stuff as I'm going through these processes, and I didn't fully understand the product. So I managed to make the change afterwards but it certainly wasn't easy. So yeah, that's an example where what we're going to be talking about today, friction, might have been a useful thing.

Amelia (04:24)
I think it's an interesting one. You were saying about the address thing. I moved house just a few months ago and I cannot tell you the amount of things I've ordered online that have gone to my old house. There are some retailers that do it really well. As you say, like Amazon, it's quite easy to save it last minute, or even if you've put in the wrong address, you can then change your delivery address. But there's lots of websites I've used, lots of retailers who, there barely is even a step where you get to see your address. It's so quick. And even then when I've gone and used that retailer again, and I'm really aware of it, there doesn't seem to be a really obvious place to check your address. So that's caught me out so many times. So do you think it's a case that helpful tech is maybe becoming too helpful?

Paul Wood (05:06)
Yeah, I was thinking about this before we were talking, and I think helpful tech used on basic things is stuff that doesn't matter too much. So the Amazon example is one where, okay, if you accidentally order to the wrong address or you make a mistake, they make it pretty easy to fix that mistake. But when it comes to something more important, like the car insurance example, which could… because I've read stories of people using these systems where they'll be driving down a road that runs parallel to a motorway. Well, sorry, no, driving down a motorway that runs parallel to a lower speed limit road. And the app will think that they're on the lower speed limit road and therefore mark them down for speeding, and that then meaningfully impacts your future insurance standing.

So having friction in the process of making decisions like that is actually quite important, I would argue, because it's not really a simple mistake. It's a mistake that has consequence. So yeah, I think these systems — having them built to be smooth and easy to use is obviously important, but going too far with it can actually have some unintended consequences.

Amelia (06:41)
And yeah, it seems that as designers, developers, marketeers, the natural instinct is to reduce friction, make it easier, make it faster, remove those barriers. But sometimes slowing down the user, as we say, can cause more of a problem. Russ, as a designer, is your instinct always to make things as smooth as possible?

Russell (07:05)
Well, I suppose what you want to achieve in terms of user experience is often set by the business objectives. And I think it's the business objectives that really form the foundation for what you see at the end in terms of interface design. Amazon's a really interesting example because they've spent a long time trying to make that platform as easy as possible to use.

I think there's a famous statistic that Jeff Bezos invested, I think, 100 times more capital into customer experience than advertising at the beginning of Amazon. It's very much focused on trying to have the smoothest, quickest user journey as possible, which aligns with the business objectives because people are purchasing quickly.

They're probably thinking, is it better to have a one-click order and then deal with what happens afterwards, than have friction in that process and then have fewer orders? They've probably calculated that in terms of returns and administration afterwards, it's actually better for the business to do it that way around. Even if there are orders that are cancelled or those kinds of queries post-clicking this kind of one-click option. So you'll probably find that it's the business driving it as much as it is the designers.

In the case of a regulated space like finance, I guess that's where it becomes more complicated because there's certain questions people are going to need to answer. You can't quickly go through a mortgage, even a mortgage calculator form, without accepting various terms and conditions and asking questions, and especially if you can borrow an evaluation in that way I think they need to be quite careful in what they say and the information that goes into it.

Yeah, I think there's cases where it is just one click and you want that to be smooth. You just have to be careful with addresses. I also moved house a few years ago and I was having post turning up at our old address for years because of that very reason. And I think there's a big difference between, say, retail and finance in that respect, where you kind of want that friction at certain points to make sure that you're asking the right questions, that people are accepting the correct terms and conditions, that they can't just go through and change their car insurance or change around their pension pots.

And also, I think we probably forget in the world of design, we're not always designing for the end user sat at their desk just browsing on their computer as a customer or as a consumer. There's the whole process of the internal team using these systems as well and using these platforms. I know people who work in finance and they have to make huge payments, for example, using these systems.

And they're not simple processes to make, if you're, say, administrating a trust or something like that. Or you're onboarding a client or closing a client's account. There's digital systems that are used to do that. And that's more encouraged these days as well. They don't want their teams being swamped in paperwork.

It's all done through digital systems which need to be carefully planned out, so they look at areas like, well, they need to meet regulation around anti-money laundering and make sure very strict processes are followed. So those internal processes as well — you couldn't just have a one-click transfer of 10 million pounds or whatever into another bank account. So that's another example of where friction is really important. There's so many different scenarios where friction is important.

There's also a law that's worth looking up as well called Hick's Law, which is a design law, which — probably not going to explain it too well, but essentially it's if you give the user more choice, the decision making becomes harder. So you kind of want to remove choice and decision making to make the experience as easy as possible.

Paul Wood (11:43)
Speaking of similar things, I was really interested in behavioural economics for some time, and there's the idea of choice architecture and principles of choice architecture which are similar along the same lines of what you were just speaking about there. Yeah, if people are overwhelmed by choice, they find it very difficult to then choose because there's just too much available.

And then I was looking through the principles of choice architecture and one of those is to expect user error. So a good system will expect people to make an error and then do its best to solve that. And I think the stakes are probably getting higher now because, first of all, AI-enabled systems are making user journeys much smoother, quicker. You can achieve more in less time. Which obviously means you can reach a result more quickly, but it also means you can reach an error more quickly as well. So if a certain percentage of users are going to make an error, whatever you do, you're just speeding up that process to error. So I suppose, to user interface designers — do you think that in this brave new world of AI, we need to rethink how we design interfaces to expect that error and put blockers in the way, if you like?

Russell (13:21)
Is that an error that an agent would encounter or a user?

Paul Wood (13:25)
Well, I suppose user really, because the agent will be acting on behalf of the user, but ultimately it's the user who carries the can.

Russell (13:32)
Yeah, I think it's interesting in the context of when an agent would encounter the error on behalf of the user, also users encountering those errors, because an error is a point where it's a flag that they need to put in — whether it's a mandatory field or they've input the data incorrectly. It could be something quite simple, like putting in your card number and you've chosen the wrong type of card. So you might have chosen MasterCard and put in the wrong, you know, you might have a debit card, for example.

I think those errors are going to force the user to perhaps pay attention and to get things right in a world where things are just going to get faster and faster to complete. Because ultimately that's what users want, right? They don't want to be stuck at their screen filling in forms. That's the boring bit and that's the bit that people just want to be done as quickly as possible. I think most people probably enjoy exploring the internet and ordering online and chatting to friends and the social aspects of it. But when it comes to forms, that's probably fully the lowest down on people's list, especially if you go to older websites where they become even more difficult to use, you kind of give up.

And that's where people go to these big websites and comparison sites, because they take a lot of that stress away, a lot of that research away. But you don't want it too seamless, because everyone just wants to have that one click on everything. Because that's the bit that people don't want to be wasting their day doing. You don't want to be filling in a form and you don't want to be filling in those details when you're booking your holiday, or getting car insurance, or opening a bank account, whatever you're doing. The user wants that to be one click, but how do we stop it being one click to make sure that we capture all of the information that we need to in the right way? So yeah, it's a case by case basis.

We've been talking a lot about progressive disclosure recently and wrote an article on it. I think it's important as well when you're laying out forms, like the example that you gave, Paul, with insurance. Instead of being presented with loads of form fields, as you fill in the form it should change and it should give you that information as you go through, instead of it just being a set process for everybody. That's where progressive disclosure works really well.

I looked at every major mortgage calculator, well, no, not every one, maybe the top five or six in the UK, mortgage calculators a few years ago. And the best one used that progressive disclosure approach. So for example, if there were two people applying for a mortgage and you wanted an idea of what you could borrow, you wouldn't be presented with two sets of information to fill out, you'd choose whether there were two people first, and then the page would change if there were two people applying. Which sounds quite basic, but actually you'd be surprised how many forms and systems don't work in that way and really struggle to cater for different use cases. So yeah, there are techniques you can use, but going back to friction again, I think it is important to have in place instead of just flying through the process in one click when it comes to booking or ordering or applying.

Amelia (17:24)
And thinking about AI, when you're trying to strike that balance between speed and convenience and, of course, safety and comprehension, does AI make that easier or do you think it makes it harder?

Russell (17:37)
Well, I think it has to make it easier. The exciting thing about AI and all the ways you can adopt it is just making everyone's lives easier. It's giving us more time to do other things that we enjoy, or it's allowing us to multiply our output. I think if it isn't, then it isn't working and we shouldn't be using it in that context.

There's an interesting one — OpenAI have been working on something called Operator. And there's two AI agents that have been quite interesting out there, which is Project Mariner by Google and Operator by OpenAI. Project Mariner was more of a research tool when I first looked at it. Operator's really interesting and I'm sure it's embedded into their new browser as well a bit more. But it does carry out those end-to-end transactional journeys for you. So it would do your online shopping, it would book a seat at a local restaurant for you.

But it stops at the point where the user has to make a decision, which is a really key decision. It would ask if you — it wouldn't go ahead and book your reservation at a certain time and a certain place without then checking with the user. It wouldn't change your car insurance or your buildings insurance without checking with the user first, being like, is this actually what you want to do? I don't think we'll ever get to a point where it stops checking. It's going to be quite dangerous though. Because then what do you end up with? You can end up with anything.

Paul Wood (19:23)
I think there are people in the world though who, that is their ambition, isn't it? And I think there are people whose vision it is that you'll have an AI agent that will act on your behalf and make decisions within a set of parameters, I suppose. So the most basic example will be you've got a fridge that knows when the milk is running low and it'll go and order some milk so that you never run out of milk basically. But there'll be people pushing for that to expand into all different areas of life.

So yeah, the car insurance example is probably a good one because, certainly in the UK, for a lot of people, myself included, it feels like a burden. Every year you get a renewal quote, you have to remember who you're insured with, you have to go and check some other quotes. It's just something you have to do. I don't do it because I like buying car insurance. I do it because I have to buy car insurance. And so if you're like me, you naturally get a bit lazy with that process and you sort of get through it as quickly as possible, with the only thing in mind being I want to make sure I don't get ripped off, I suppose, is what I'm thinking.

And I suppose as I get older, I also think I want to make sure I've got a good deal, good coverage. I'm not buying something rubbish. But if AI starts to get involved in that sort of process, I suppose the risks are quite high because you won't necessarily know there's been a mistake made until you need to rely upon the car insurance, and then you make a claim and the claim then gets rejected because — who knows — you might not have given your agent the right spelling of your surname and it invalidates the insurance or something silly like that. I worry that that's the direction that we're going in. And, as designers of systems, I suppose we've got a role to play in setting the guard rails a bit and managing that.

Russell (21:46)
Yeah, essentially what we're talking about is designing the interaction with the agent more than the interaction with the agent and the firm at the end. And if there's an issue with your current insurance because of something the agent's put for you, then the agent's going to have that conversation with the firm and then be relaying back to you as well, and probably apologising. Does it at that point compensate you if it's made a mistake? Who's liable in that situation if the AI agent's made a mistake? Because I guess you'll need insurance for the AI agent to take out. That would be your main insurance package because the AI agent would be doing everything on your behalf.

Russell (22:42)
I think people are so busy, it feels like more than ever. And the things people don't like doing, they're always looking for technology to help them with that, whether it's their weekly shop and using subscriptions, whether it's chores. Everyone's dream, right, would be to not have to go shopping and have a robot doing all their chores, right? Because then you'd have time to do what you want to do, whether that's reading, or spending time with your kids, whether that's playing music. As you get older, it's difficult to find time for those things that you enjoy because you're so busy on just like maintenance of your life with all these different packages that you have on the go that need to be renewed.

So if technology can actually do all of that on our behalf, whether that's AI, whether in the longer term there's other technology that we can use, then I think people are going to start to adopt it because it gives us time, and ultimately time is the most valuable commodity and it's short on supply. So yeah, I think anything that makes people's lives easier, they will. But there's a level of risk, isn't there? Because I don't know about you, I wouldn't feel too comfortable with having an AI robot doing all the chores around my house. But that could be a generation gap. That could be, that might be the generation gap.

Paul Wood (24:20)
Yeah, because I can genuinely remember the time when people would say they don't feel comfortable buying something online. That was an actual thing. And you didn't look at them and think, "you old fossil," you just sort of accepted it because you're like, yeah, I can see where they're coming from.

Paul Wood (24:46)
It took a while before, I suppose Amazon was the first where you thought, okay, they seem to have a process here. I feel comfortable buying from them. And then over time that sort of spread out to independent online websites that you then bought from. And now you just don't even think about it, do you? And I think usage of AI agents could go the same way.

Russell (25:15)
Yeah, I think people will start to adopt it. They'll get more time back. They can do things that they enjoy. One thing I haven't checked out is the new OpenAI browser. Well, I did check it out, I haven't used it to the point where I would switch to it. And it's really interesting because it just shows how change is so slow. Mass adoption of browsers, or a lot of online shopping — to begin with, it takes a long time. Even if people get it right.

I think moving to this new way of living, really, where you're using agents to carry out these tasks, you'll have people at the cutting edge of technology probably already doing it somehow, and if the technology isn't out there they'll probably create something which can do it for them with a variety of different systems, or code it themselves. But for mass adoption commercially, it might take five, ten years before we see AI agents being used by people day to day and then people stop thinking about it, it just becomes a normal part of their lives. Maybe sooner, it depends on the impact it'll have.

Yeah, it's going to happen. It's just how we get that level of trust and how we get those right points in that process just to avoid it being complete chaos. And like you say, having all kinds of transactions being made on your behalf, having the wrong things turning up, having the wrong insurance, for example, having tickets to the wrong event, you know, everything happening all at once.

Paul Wood (27:10)
One of the things that led us to discussing this today was we've recently been talking about consumer duty and what they term sludge practices, which they define as excessive unnecessary frictions. And I think what the FCA is getting at with that is maybe the classic example of the firm that you're locked into that you can't get out of easily. So you could maybe sign up for something online very easily, but then to cancel it, you have to phone some office that's only open at a certain time and wait on the phone for ages. I suppose that would be an example of a sludge practice.

And then it got me thinking, because they're removing unnecessary friction — but is there such a thing as necessary friction, which I think we've established there probably is. But in your role as a designer and someone who has assessed user experience workflows across, say, private banking apps, for example, or mortgage journeys and things like that, have you ever had to stop and think we need to introduce some friction here? Is that a thought that ever enters your mind? Or is it just something you're naturally geared towards thinking about, this is a high-risk piece of work?

Russell (28:43)
Well, normally the scope of what's required to introduce the friction is set by the client. So they'll have a specification document, which will have various fields. And they'll have set processes for closing an account, let's say. But I think making it fair, it does come down to the patterns that are used.

In UX it's referred to as dark patterns, where it might be difficult to close an account, a link might be hidden or the text might be too small, which you probably see. Sometimes you can see it when you're going through even to purchase something that they're trying to upsell you a subscription or something else. And actually the link to going through to the checkout is really small, and it's really small text, which is actually a bit of a dark pattern in terms of design. That's what we've referred to.

I think what we do consider as designers is accessibility though. And that's just making sure that these systems can be used by everybody. So if you've got low vision, we look at the size of text and colour contrast. And if you're using a screen reader, making sure that the hierarchy of the page is labelled correctly and there's alternative text for images. If you can't see images, the alternative text explains what that image is. So in the view of consumer duty, that's good because it's fair and transparent services and products for everybody, no matter what ability you are.

So we consider that as a designer. But yeah, normally the actual process and the scope is set by the client because they have form fields that they need to use, which normally need to be input back into their own systems, their CRMs and their own databases to manage it in-house. So our role as a designer though, is to introduce that friction when those key steps are required. If it's saying that you're a board member, for example, or if you perhaps have business shares, those kinds of steps introducing friction at that point, perhaps in a company registration process, would be quite important. But yeah, normally that's set by the client. Our role is normally to work within that scope and make it as easy and as smooth as possible. So we're actually the ones probably trying to reduce friction, to be honest, and friction is introduced by the client because it's the client who wants all these various steps in place.

Amelia (31:55)
So it's ultimately kind of the aim to have that friction there, but ultimately you don't want the user to even notice. So for them, it's a really smooth process and they're not even really noticing that at those points you're slowing them down or you're adding in those extra steps.

Russell (32:10)
Yeah, we're talking more about forms. I feel like when we're talking about friction you always think of forms, but it isn't always, because it could be adding a product to your cart, whether that's in finance or retail. It could be — I think we've talked about investments before as well, which is a really interesting area — because moving your pensions, your pension pots around, for example, it could be as simple as introducing friction in the way of a pop-up, just saying, you want to do this? Instead of carrying out that action. For the end user, I think it's important for them to have some visibility of what's going to happen next. They could also click on the button by mistake.

Russell (33:04)
So the best practice is if it's an important decision, having a pop-up to say, you know, are you sure you want to do this? It could just be as simple as that, which I think in that case, it is the responsibility of the designer working with the developers to make sure that for those key interactions, there is a warning message which pops up. And maybe a couple, you don't want to overdo it. It depends. Even a preview step.

Often, with some of our bigger clients, we have four-eyes checks on even making simple content changes to a website, publishing an article, let's say. If it's written by an expert, then they might want internally to have that four-eyes check on content which is published. So it's then our job as the team producing that content publication workflow for them to make sure that there's a four-eyes check. And I think that happens a lot in regulated firms as well, where there's normally approval, whether it's by managers, to have that four-eyes check process. So a lot often that check process you could have run through into the website to avoid someone even clicking on a pop-up. You could take it to the next level, if it's a really key transaction, maybe for a large sum of money, you could then offset that risk even further by having an approval step after that. So yeah, I think there's ways to mitigate it. It depends how big the consequences are of that interaction.

Paul Wood (34:51)
Yeah, and it feels like — because I always like processes where you have to take one step at a time, because everything that's presented to you in a system, whether it's a website or an app or process, whatever, everything you see is a decision effectively. So on a website, the navigation is a decision to use it or not. And then the content, you decide whether to read it or not. All the buttons that are there, you have to decide which one you interact with. And so, it feels like the higher risk a process is, the fewer decisions should be available at any one time. So a perfect process, I suppose, if you ignore every other thing, is that it would just be one thing at a time that says, okay, here's the first thing you need to decide. And is that what you mentioned, I think progressive disclosure, is that the right term? Is that basically what that is? It's presenting the most pared-back decision that you can at any one time.

Russell (35:51)
Yeah, yeah, I'm a big fan of progressive disclosure because it doesn't overwhelm the end user, whether that's on a website, whether it's an internal system, with too much choice. Like the last thing you want is to visit a website and be presented with 50 form fields. If you can go through that process, so even if you are buying some new car insurance, building insurance, breaking that down into steps, so you progressively disclose that information — often when we see big forms as a designer our first thing we look at changing is breaking that up into steps, because it's a much neater and a much more flexible way to put that journey in front of the user. Then you can label the steps, you can categorise the steps, but you can also change them depending on how that first step was completed, and how they progress through that form ends up perhaps changing.

Paul Wood (36:53)
Forms are interesting as well, because we work with a lot of clients who have legacy systems or legacy processes where certain things that you need to do are done via forms that you effectively have to print out and fill in. And I always groan when I see that because I think, God, you could do that so much more elegantly through an online system, which they are generally — all clients are generally working towards that. But as we've been talking today, I've just thought, actually, is there something to be said for an old-fashioned print-and-fill-in form, simply because it makes you engage your brain and it makes you actually take care over the process? I don't know what it's like for everyone else, but when I have to literally sit down with a pen and fill in a form, there's a sort of nervous response that I naturally get which is like, God I hope I don't write the wrong thing here because I only get one go. So I suppose there's something to be said for that.

Russell (37:59)
Yeah, I recently went through the process of opening up a savings account and all the forms were printed out on my behalf to look at and fill in. And I think people will always still want to work in that way. It depends on your setup as well. And this is where consumer duty comes in. Not everyone has a nice big monitor, or a new laptop. We're kind of assuming that everyone has technology which they can easily use all these websites on. But not everyone can afford it or maybe they're just not interested in technology. And there's a lot of different use cases.

I often think of that as something that you need to constantly remind yourself of when you're designing websites — that we design websites on these shiny Mac machines with beautiful pixel resolution and colour contrast, and although technology is improving, if you watch an end user using it they're probably using it on a screen with their contrast right up, they're probably on Windows and a completely different setup, because they're not perhaps so technology-minded as the people working in the agency.

So yeah, this is where UX research comes in, or user research, because I think that sits between what we're presenting to them in terms of interface design. It sits between the business objectives and what we're presenting as the interface designer. You need to constantly be speaking to your users, sending out surveys, whether it's a pop-up survey on the website or a survey in an email, interviewing users, conducting usability testing studies. And the higher the stakes, the more important that becomes. So you need to be doing that on a regular basis and compiling all of this feedback, which is getting easier using AI as well. Analysing transcripts and crunching data to find out exactly what people are struggling with, but also how they're using your website and how they're using your system. Because it might be different to what the initial stakeholders thought, or how they used it when they were creating it all. So it's just really important to understand. And there might be steps of friction there which are just too difficult. So you might have to do it in another way.

So I set up an account recently where I had to call the firm to get an ID number so I could log into the platform. It just felt like a process of friction which wasn't necessary, and wasn't what you would expect — the process of obtaining your ID number to be digital. And it was difficult to call. I had to wait, I had to get a call back in the end because no one answered. They called back, I was asked some questions, then I was given an ID number, then I used that to log into my account. Now perhaps that had to happen because they had to verify my identification over the phone. But it felt like a bit of a — it probably took an hour or two because I had to wait for a call back. So after making the phone call, I was then left a little bit wondering what was going to happen next because I didn't know when they were going to call back, which wasn't great for customer service. And then yeah, it was a really nice, pleasant conversation. I got my ID number, I could log in, I could access my account. But that is a good example of friction which was kind of taken away from the digital process to something like over a phone call. So if I was the firm looking at that process, I'd be thinking, how could we improve that and improve the customer experience? So perhaps the ID number could be created online and ID verification could be done in a different way.

Paul Wood (42:01)
I think one of the most interesting areas, certainly in financial services, that is going to need to think about this whole process is pensions, because in the UK workplace pensions have been around for quite a while now. So everyone who's working will be, unless they've opted out, paying into a workplace pension. And that means people, as they move jobs, pick up different pension pots from different providers. That's then driving a growth in people consolidating pensions with one provider, moving pension pots around.

And that's quite a high-risk area, I suppose, because if you move your pension to the wrong place or you make a mistake with it, that's a meaningful thing. But at the same time, pension providers can't be seen to be trapping people, they have to be able to enable transfers. So I think there's a fine balance there between making the process smooth and easy, but not making it so easy that it's one click and, whoosh, your pension's gone somewhere else. I don't really know what the answer to that is and I don't know who is doing it well, but I think that's going to be a real area of focus — how do you balance that? Removing the sludge practices which trap people versus introducing good interventions that make sure people know what they're doing and why they're doing it.

Russell (43:39)
And that becomes even more confusing with using an AI agent, doesn't it? Because they're going to have to deal with the sludge. So the agents are going to be the ones trying to almost bypass all of this because they're going to be trying to act in the best interests of their master, let's call it, and finding the best deal for them and the best result for them.

So it's going to be really interesting because AI agents are going to be completely unemotional when it comes to being drawn in different directions through an interface design. They're going to be looking at it in a very systematic way. Our website is going to have to start introducing different ways to provide sludge to agents than humans because they're going to be able to get through processes faster and less likely to be convinced to stay, because they're not going to have any emotion at all. They're not going to have any emotional attachment to the company and brand loyalty and anything like that. So yeah, it's going to be quite interesting to see how companies retain customers when AI agents are making decisions, or maybe giving only one decision back to the user of them.

Russell (45:06)
It's going to be a whole new world in terms of user journeys. It's going to be one user journey for the human and one user journey for the agent to start thinking about.

Amelia (45:18)
I'm sure there's all stuff we'll explore further and touch upon in coming episodes. But for now, shall we play some Jargon Busters?

Amelia (45:27)
I can't remember how you guys did last week.

Paul Wood (45:29)
We got it. Yeah, we got there. I think Pat did quite well, didn't he? He sort of had a good explanation for it.

Russell (45:31)
It was a while, wasn't it? Yeah.

Amelia (45:39)
So this is where I put your industry knowledge, or industry term knowledge, to the test. I've got my list of those terms. I picked one at random. Today, I want you to have a go at explaining the meaning of the word spread.

Paul Wood (45:54)
Yeah, this one. So spread, I think I've heard spread in the sense of it to do with investments. And it's about — yeah, I am guessing — so it would be spreading your risk, I suppose, or spreading your exposure. So if you've got a pot to invest, you spread that pot among different investments. That would be my guess.

Russell (46:25)
Hmm, it's quite a broad term. I can't help but think of it in the context of betting for some reason.

Amelia (46:26)
What do you think?

Paul Wood (46:34)
Yeah, spread betting.

Russell (46:35)
But I could see why spreading your risk as well across different commodities and industries to reduce your risk. It might be related to that, but I'm going to go with betting, because I'm not really quite sure what is spread betting — when you do the same thing, but you do it across lots of different sports or, well, maybe not sports, but events.

Amelia (47:04)
So spread means this is the difference between the bid price and the ask price.

Paul Wood (47:11)
So we were nowhere near actually.

Amelia (47:14)
So we can see what you're going for.

Paul Wood (47:15)
Yeah, okay, so that's — because if it was what I thought it was I was thinking that's quite a logical name for it. But it's not. Spread. It is jargon still.

Amelia (47:30)
Yeah, it's not often logical really, is it? As we're learning.

Paul Wood (47:33)
No.

Russell (47:36)
It's investment again, it's investment time again. Yeah, the bid price and the ask price, all right, okay.

Paul Wood (47:42)
Price difference, yeah.

Amelia (47:43)
Yeah, so there we go. I mean, of course we will have more of this in the next episode. Speaking of the next episode, we're going to be discussing what happens when AI becomes your financial advisor. What happens then, Paul?

Paul Wood (47:57)
Well, who knows? Because this was all driven by — well, first of all, the term advice is really hotly protected. If you work with anyone in financial services, they'll almost always have had a story of using the word advice in a piece of marketing, or proposing to use it in a piece of marketing, and being shot down. And funnily enough, it happened to me this week. The word advice snuck into something and somebody said, whoa, careful, we can't say advice.

So yeah, this was the idea for the next episode — it was driven by some research carried out by Lloyd's Banking Group that suggests that, I think they said 50% of, sorry, 56% of UK adults have used AI to help them manage their money over the last year. So it's obviously happening right now. People are relying on AI to — I don't know whether they're getting advice per se, but they're certainly asking questions of AI relating to their money. So it's only a matter of time before AI is dishing out advice. So I think it'd be a good discussion to kick around — well, what does that mean? How do brands exist in that world and protect consumers at the same time?

Amelia (49:23)
I think it's definitely going to be an interesting discussion, as I say, that's coming up on next week's episode. But that is it for this week. Thanks both for joining us and thanks for listening. And as I say, we'll be back next week.