Tracking Consumer Duty outcomes using Google Trends

Episode transcript

Amelia Hi, I'm Amelia.

Paul I'm Paul.

Russ I'm Russ.

Amelia And this is Fin the Week. Welcome back to a brand new episode. How is everyone?

Paul Pretty good, yeah. I was just saying I can't believe how fast this week has gone. It's a cliché, but this one in particular seems to have gone very fast.

Russ Yeah, using AI in my workflow seems to make me more efficient, but also much more consumed in my work. Time just flies by because you can do so much with it.

Paul Yeah, I find it sort of sucks you in. You almost do way more than you ever used to, but it doesn't mean you're finishing faster. It means you go much deeper into a piece of work than you ever used to, so before you know it, hours have passed.

Russ Yeah. I think planning your day and planning your week has never been more important. It's good to experiment with these tools, but the danger, like you say, is losing yourself in them and then getting to 5pm and thinking, I haven't actually managed to do what I set out to do today. So yeah, planning is important.

Paul Yeah.

Amelia It's interesting to see how it just changes the way you're working.

Paul Yeah, it changes the ambition, I find. With a normal piece of work, I now sit there and think, but I could do this extra layer, and then another layer, and another layer, and before you know it you've built this incredible thing. But it's way more advanced than I would have done in the past. It doesn't really save you time, it just makes it a more interesting piece of work, actually.

Russ Some of the output now, and the different skills it gives you, is just really unbelievable. We're not talking about it today in the episode, but if you have an idea now, it's a great time to be using a computer, because you can execute that idea in so many different ways, so quickly, using AI, and bring it to life more than ever before. It's really exciting.

Paul In fact, that does actually tie quite nicely into what I've done over the last few days.

Russ Yeah, actually it does. I'll take it back.

Amelia So today we're jumping into something that we touched upon last week. Paul, tell us a bit more.

Paul Yeah, so as I alluded to just there, last week we were discussing and reacting to the FCA's annual public meeting. During that, I mentioned that the FCA CEO, Nikhil Rathi, in passing appealed to the audience to come forward with imaginative ways, as he put it, to really test the outcomes that consumers are receiving. As we were discussing it, it occurred to me that one of the imaginative ways you could do that would be to use Google Trends to analyse consumer search behaviour and understand how people are finding things at the moment.

So after the episode last week I went down a bit of a rabbit hole and actually built a reporting system. I leaned heavily on AI to help me deliver it much faster than I ever could have done before. Essentially, it's a reporting system that uses Google Trends data to analyse the financial services sector. So I thought it'd be a good idea for us to run through that today.

Amelia I think it's a really interesting approach. Before we get into it, give us a brief explanation of what Google Trends actually is.

Paul So Google Trends is a tool delivered by Google, of course, that lets you explore what's being searched for on Google as well as YouTube. A lot of people will be familiar with keyword research processes, where you get a search volume number out of it. That type of research comes from Google Ads, basically. It's how Google sells ads, and it lets you see what people are searching for.

Google Trends is slightly different. It's a much broader way of carrying out research. If you put a keyword in there, it will give you a score out of a hundred. You can almost look at it like a percentage, where a hundred is the peak. So if you set a time period of, say, five years, anything at a hundred is where the peak of that time period is, and anything lower is relative to that peak. It's not an absolute number, like this is how many people searched for something. It's more relative.

An example I use in an article I've written about this is the World Cup. If you set the date range back as far as it will go and put World Cup in as the search term, you'll see a graph with spikes that correlate with when the football World Cup tournaments are, and smaller spikes for when the Rugby World Cup is, for instance. You can also see things like the peak at a hundred coming at the most recent World Cup tournament in North America, and then how that compares to previous World Cups and what people searched for. So it's a good way of getting a barometer for how in demand a topic is.

Amelia And you mentioned last week, Paul, how Google Trends has been used to track flu symptoms. That's really interesting, isn't it?

Paul Yeah, it's something I'd heard about in the past. People have used Google Trends to experiment. There was a study in Taiwan, for instance, that wanted to understand whether searches for flu-like symptoms correlated with people visiting emergency departments presenting with those symptoms. The study found that it did correlate. It wasn't a nailed-on "here's how you track flu symptoms", but it was a potential early warning system to give you a week or two of warning that a wave is coming.

It's probably worth noting that Google also had its own flu tracker tool to track outbreaks, which it ran for a period of time using this same data, but they shut that down after it overestimated the flu season in 2012 to 2013. So it's not infallible.

Amelia Bearing that in mind, how trustworthy is it?

Paul I think it's one source of information among many others. I don't know too much about the history of Google's tool, but I would imagine the reason they shut it down was because it could mislead, in the sense that people might think it's to be a hundred percent trusted. In actual fact, it was probably built to be one way of potentially monitoring what a flu season was going to be like, but you should also use other methods of research. So maybe that was why it got shut down. Regardless, what the other studies have shown is that it can correlate with hospital visits, it can be an early warning system, and if nothing else, it's quite interesting just digging into what people are searching for.

Amelia Definitely.

Russ Yeah, I was just having a look. It's one of those data sources that's been around for a very long time, but it's often forgotten about because there are so many other ways to look at website usage. So it's really interesting that you picked up on it again, Paul.

I was just looking at UX as a term in there, because it's quite interesting to look at words that people are searching for that are perhaps used a bit more in modern language than they were in the past. UX, it's told me, has increased by 80% in the last five years compared to the previous five years, which is quite interesting. It's validation that that area of specialism is something people are looking for and is more relevant now than it used to be.

So it's good validation for a number of things. If you're looking for a job, perhaps even investments. From an entrepreneurial perspective, if you're looking for an opportunity, it might be quite a good way to look for things that people are searching for, certain industries or certain technical words, and you could create product concepts around those. Or, thinking about financial services, validate what people are actually looking for. For example, if the search keyword for a certain investment has nosedived over the last few years, you'd probably think it's less popular than it was before. So many uses. It's really interesting, Paul, to hear that you're looking at it in relation to Consumer Duty.

Amelia Yeah, let's get into that. How did you actually use Google Trends to look at Consumer Duty? Where did you start?

Paul The first step was to build what's essentially a database of search terms. It's a spreadsheet of around ninety search terms, and each of those is categorised into one of the four Consumer Duty outcomes: products and services, price and value, consumer understanding, and consumer support. Then there's an additional category that's cross-cutting against all of those, which is vulnerability. So essentially what you've got is a list of ninety-odd search terms that represent things people might search for against those five categories.

In addition to that, I built in an industry benchmark layer that takes the temperature of search in general for financial products, so things like mortgages, savings and credit cards, so that we can see a baseline of what the market average is. Are more people searching for these things at the moment, or fewer? The reason that's important is it lets us understand what is growing faster than the market, or faster than the baseline.

The report has been built to use an API to track those ninety-odd search terms against Google Trends, dating back as far as 2021, and it can show us, for each of those search terms, where they're at on their growth journey. Are they up or down in terms of popularity? Then we can look at that alongside the industry average and say, okay, these search terms are above the average for the industry and these are below.

There's no brand-level data in there, it's all industry level. We're potentially going to add in brand-level data in a next version, because I think that would be interesting as well. For instance, you could look at the top ten UK banking brands and what people are searching for in relation to those brands. For example, people having trouble with their apps. That might give you an indication as to which brands have an issue with the user experience of an app, or something like that. So there's loads you can get from it. But in the case we've got, we've been able to use this data to say, what's happening with the Consumer Duty outcomes? What are people searching for more than the average across the market? So it started life like that, to see what it can tell us, really.

Amelia What were the findings?

Paul First of all, it's probably worth outlining what it can and can't tell us. It can show us what is growing in demand in search, and from that we can make guesses as to why that might be. It can't tell us for sure that people are struggling with this or that, but it can usually give us a "this or that" type scenario where we can say what's worrying consumers right now. I think that's useful, because search is the first port of call for consumers when they have a worry, an issue or a question they want answered. They'll often go to search before anything else because it's low friction. You get a quick direction to go in. So I've been thinking of it as an early warning system.

As for the actual findings, there are loads in the report. Quick plug: I've published an article this morning with a breakdown of what's in there, with a link to the full report. So I won't go through every single finding, but I'll highlight the key ones.

If we start with price and value as a Consumer Duty outcome, searches are growing faster than the market to the tune of 24%. That includes search terms like "loan interest rates", which is up 163% year on year. When I say year on year, that means the most recent three months versus the same period last year. "Bank switching offers" is up 140%, "easy access savings" up 117%, and "fees and charges" up 58%. So they're all growing ahead of the market. I should mention that the market as a whole is 31% higher than it was at the same time last year. So that's the baseline.

Another interesting thing we can look at is case studies. In the price and value category, we looked at the 2022 UK mini-budget, which was that disastrous period where everything went wobbly for a couple of weeks. You can see in the data that there was a spike in the numbers around that time relating to mortgage rates and things, and then that settled down quite quickly when the market settled. But in 2026, switching offers, easy access savings and loan rates have all passed their 2022 levels and have stayed there for months. So the data is able to show us what happens when there's a sudden event that prompts a load of search activity, versus what is more of an ongoing trend. And there are signs of ongoing trends right now in that area. So that was the first interesting area that stood out.

Then vulnerability as a category is also growing faster than the market. Debt and arrears searches have more than doubled, so 116%. "Credit card debt" is up 66%, "debt consolidation" 61%, and "mortgage arrears" up 33%, so that's only slightly above the market.

But searches for the term "cost of living crisis" are down 90% from their peak in August 2022, which shows you the relationship that search has with the news cycle, because I expect back at that time the cost of living crisis as a term was much more widely reported in the news, and people's search habits probably mirrored that. We also track things like searches for "food bank", which is also down, but by a shallower 39% from a peak in December 2022, and it does seem to be ticking up again. So what that shows is that searches for food bank aren't as high as they were previously, but they're still relatively high and ticking up again. That's another type of insight that could be useful to regulators and brands.

Then there was another nice reminder of how the news cycle impacts things, because searches for "bankruptcy" have recently had a spike. That correlates with a news story about a football pundit, where there was talk of bankruptcy related to that person. Having said that, bankruptcy searches are generally up from where they were previously. So there are lots of findings there as well.

Scam checking is at a five-year high. This is a really interesting one, I thought. Searches for "investment scam" are up 199%, "crypto scam" 107%, and "is this a scam" up 77%. This is a good demonstration of where there are two readings of the data. One reading could be that searches for scams are on the rise, so does that mean people are being exposed to scams more frequently? Alternatively, does it mean people are just more aware of scams, and therefore are taking more caution and educating themselves? This data can't answer that question. This is where you then need to do your own research and build your own understanding. I would imagine the truth sits somewhere between the two scenarios. I'd imagine there are more scams people are exposed to, and I would also imagine people are more aware of it and searching for themselves.

Actually, a question I wanted to pose to the group, and particularly from a UX perspective, Russell, is what would be the ideal user journey if somebody was searching for "is this a scam" in relation to a particular brand? How does a brand use that information? It's a tricky one.

Russ Yeah, I mean, hearing you talk about everything you found, it just sounds like a real goldmine of information to help not only the FCA but also brands, if their brand name is mentioned. I almost visualised having a real-time monitor, a grid of monitors looking at all of these keywords, and if there was a spike you could act quite quickly on it that day. That would be such a powerful form of monitoring for your brand, or for the FCA, to quickly investigate something that might be happening and become quite a high-profile scam before it's brought to their attention.

In terms of the actual data source, it forms part of the bigger picture to understand exactly what's happening. So people are searching for something. Let's say a brand is mentioned, and a scam is mentioned, and perhaps how to get support or contact that brand. Then I think it would be up to the brand to look at their website and look at where that might be happening. Perhaps understand why people can't contact them directly as well. Why are people using search instead of going to their website, or going to their relationship manager if they have one, or using the facilities on the website? Maybe that isn't working. Maybe that's a flag that there's a problem with the complaints route on their website.

So they could do a UX review if they have time, or put it in the pipeline to do a UX review of the complaints route on their website. When we did the teardowns earlier in the year and looked at twelve firms, ten out of those twelve firms actually had the complaints route missing or buried in the footer. So if a brand is seeing a lot of activity on search, that might be a flag that their digital channels aren't working for them. I'd definitely look at that.

But yeah, like you mentioned about using different data sources, I think it just tells you one thing. Then you need to align that with reaching out to people and finding out a bit more information. You could implement a survey very quickly. Email response rates might not be the best, but you could have a pop-up on your website, you could send out a survey via email, you could perhaps call individual customers if they're happy for you to call them. Once you understand what people are searching for, I think understanding why then becomes the important thing to do from a UX research perspective. I talked about data triangulation perhaps the week before last on the podcast: finding different sources of information, speaking to people who are actually searching for these terms, and then you find the truth somewhere in the middle.

Amelia And Paul, were you surprised by any of these findings? Do any of these stand out to you, or is this what you would expect?

Paul I think I went into it without any idea of exactly what it would show. I don't know whether I was surprised. I was more excited, in a funny way, about what it can help you track. For example, searches for "food bank" and that type of thing are really interesting, because that's not a term that a financial services brand would research necessarily, unless you're directly involved in that world. But it's such a useful barometer for what's on people's minds.

And then the data on scams stood out. I don't think I expected that level of growth, because when you look at the report and the graph of searches for "is this a scam", you can see the curve it's on, and it really has shot up over a relatively short period of time. I don't know exactly how many people search for it, but a lot more people are searching for it today than were a few years ago.

There's so much of that type of information in this report. Like I said, you can use it in various different ways. I wouldn't use it as the single source of truth for something, but I would certainly use it as an ingredient in a communication strategy, for example.

Amelia And how do you think you would dig deeper with that? Because as you're saying, if people are searching "is this a scam", is it because more people are being scammed? Is it because people are just more aware and more cautious? How can you use that information with further research to actually get to the bottom of what's going on?

Paul So there will be statistics out there and research into how many scams have actually been reported to authorities, so you can combine it with that information to plot how that's growing. You could then look at your own internal data as a firm to understand whether you're getting complaints, questions or confusion from your customers about certain things. Are customers reporting that they feel uncomfortable with certain parts of your process?

I've had scenarios where you receive a phone call, or even make a phone call, and they ask verification questions, and you start to second-guess yourself and think, hmm, do I want to be giving out this information?

Amelia Yeah.

Paul Even though I phoned you, you might have somehow got a phone number to me that I didn't know about. So there's stuff like that. Firms probably get a sense of that through their customer support teams. Russ mentioned data triangulation, so using this in combination with other sources of data to then draw conclusions.

As a brand, you could just know that scams are on people's minds, and therefore build your user experience ecosystem to be ready for that, and make sure there are proper ways people can validate things, help people trust things that are trustworthy, and help them spot things that aren't.

For me, having always worked in search marketing, this type of data has always been fascinating, because it's almost like you're surveying a wide part of the population, but you're not asking them what they think. You're actually just able to observe what they're searching for. You're observing behaviour en masse. And it's not even in a creepy way, because it's at the global level. It's not an individual. We're talking about the market as a whole here. So I think this is one of the really fascinating parts of the world we work in.

Amelia And do you think it's used enough? Do you think firms are using this enough? That marketers are using it enough?

Paul I don't know. One of the things I was going to do as a follow-up was look at whether the FCA, for instance, actually uses this type of data, because I don't know. As for firms, certainly the firms I've worked with and spoken to over the years, I have always been surprised that what you can find using keyword research tactics is always new to them. It's only my own experience, which is limited. I don't have a view of the whole industry, but I would be surprised if many are doing this type of thing.

Amelia We touched upon how we can use this sort of information as part of the UX journey, but in general, if, hopefully, lots of people are going to go and read your article, Paul, and explore your findings, how can firms use this information moving forward?

Paul I think first of all it can provide context. An example I've picked out is, say complaints about your fees are up by a certain percentage, say 20%, but then you find in this data that searches for fee-related things in general are up by 58% across the market. That can give you some context through which to assess your own complaints backlog. It's not a get-out-of-jail-free card or anything, but it would help you in analysing the types of complaints you're getting and understanding that this seems to be not just specific to you, but a general market move in that direction. It could help you with your communications approach, or it could help you with setting fees.

I think it can be used for communications testing. One of the things with rising demand is "what is a credit score". That type of thing gives you a clear idea of what people want to know, or what people are confused about. If that's the part of the financial services world you work in, that's where you can put a bit more effort into educating your customers and making sure they understand what they're buying.

Amelia I think that's quite interesting, Paul, especially when it comes to financial services. Obviously all sorts of industries can use this, but I feel like this is the sort of situation where there are things that you're like, I don't understand that. If you're dealing with something with your bank, or maybe getting a mortgage for the first time, and there are terms and things that you don't understand, you will run straight to Google, won't you?

Paul Yeah.

Russ I just had an idea for it, and this might already be possible in the feature set, Paul, so correct me if it exists. I think it allows you to type in search terms or keywords and see what the trend is for those. Now, there'll be a wealth of underlying data here for what people are searching for. So could it help you find things people are searching for, if that makes sense? For example, could you type in something like "help me with", then an asterisk, then "scam", and a question mark, and Google Trends would replace that asterisk with what people are actually putting in there, and give you the top ten things people are searching for related to scams?

Because whenever I've used it before, with my basic understanding of it, I would see what the trend is for, I don't know, "making complaints", and it would give me a trend for someone typing in that specific text. But can it give you trends on related search terms and tier those, so you can find out what people are searching for outside of what you think they might be searching for?

Paul Yeah, there are a couple of ways it can. The first thing is that built into the way we've done this report is that it will automatically run every month, and part of that process is to look out for what Google Trends calls breakout terms, so things that are suddenly growing a lot. It also looks at generally new, growing terms. In the scam example, it will query "is this a scam", but it will also spot anything that Google Trends flags as a breakout term relating to that. So it could be "is XYZ scheme a scam" that's suddenly growing out of nowhere, and you would discover that. So it can spot things like that.

Then it's a good idea to combine this with a keyword research tool, whether you use Google's own tools or something like Semrush or Ahrefs, or various other tools. You can do the same research using those, and they will recommend related searches as well. So again, in the world of scams, it would then show you what else people are searching for against that topic, and you can loop that into this Google Trends data. So it should form part of a keyword research process, really.

Russ That's really powerful for, like you mentioned, communication strategy. I'm thinking about the website's content strategy. You could be quite specific for that brand. If a lot of their customers are making complaints about scams, you might even warrant a pop-up for a bit, if that's really prominent in search. So you could choose when to display that information, and an alert at the top of your website, which is persistent there, related to a specific scam. From a UX and a content perspective, it's a really valuable data source, isn't it?

Paul Yeah, and related to that, about communications and timing, another mini case study within the report relates to the financial redress scheme regarding car finance payments. That process has gone through various stages, and at each stage the FCA provides a formal update. What you can see from the data is that every time there's an update, search follows. There's a peak in search activity straight after an actual update.

So you can look back at historic cases like this. PPI, for instance, was similar. And you can say, well, the next update for the car finance redress scheme is due, I think there's something happening in February next year, and the next update is probably going to be around April time. So based on the data we can see from the previous updates, we can fairly safely say that consumers are going to be searching a lot at that time of year next year. From a comms perspective, it means get your information ready. Start working on your website landing pages in relation to that stuff, because that's what people are going to be searching for. So you can get ahead of it as well. You can think of it like an early warning system, in a way.

Amelia And is there anything else that firms can take from this, do you think?

Paul Like I say, I treat it as an early warning system in a lot of ways. One of the other examples I made a note about is that it could tell you when people are searching for things like being in arrears, or looking for debt consolidation, and you would get the search data before you would get the official data on how many people are in arrears. So you can use it as an early alarm to say, okay, this looks like a wave that's coming. When the formal data lands, which is usually later, because you don't get immediate stats on this sort of stuff, this search data can give you a direction of travel.

And then, like I mentioned earlier in the discussion, we also want to start looking at firm-level data to understand where consumers are struggling with individual brands across a sector, so banking, insurance and so on, and understand whether there are strengths and weaknesses within that as well. So I think there's loads you can dig into.

Amelia So if firms aren't using this, there are lots of ideas there. And as I say, we've got your article that people can find. I'm assuming that's on the website and on LinkedIn as well.

Paul Yes, yeah.

Amelia So we've all got lots of reading to do for the weekend.

Russ I was also thinking, because we haven't really talked about LLMs here, and this is specifically Google Search, I guess a lot of people are now turning to LLMs, ChatGPT, Gemini, Claude, to ask these questions. So I think seeing trends across models will also be quite valuable in the future, if that information's available, and especially for vulnerable customers. Someone in financial distress might feel more comfortable not calling someone, and instead finding out how they can be helped through articles they find through Google Search, or responses they get through an LLM, rather than actually calling their provider and telling them they can't meet their mortgage payments, for example, or that they're in a situation where they've lost someone. So it would be interesting to see trends within the LLMs as well as Google Search.

Amelia I'm assuming that's not accessible now? And do you think it will be soon?

Paul Not really, no. I'm not sure, is the answer.

Actually, that ties back to another of the findings, which was that searches relating to consumer understanding were growing slower than the market. You could read that as either fewer people are struggling, and therefore don't need to search as often, which is a good thing. Alternatively, it could be that people are turning to AI assistants rather than Google, which would tie into our signal loss stream. That would be a place where you'd lose the data. Currently there isn't a way to understand trends relating to what people are asking AI assistants, so that is a blind spot. I suppose the thing that works in favour of getting this data is that Google is pushing hard to be one of the leaders in the AI assistant space, so that keeps that data source open. But yeah, that's going to be a challenge as well, understanding what people are asking.

Russ What would be really good to see, and data protection would come into play, is a toggle on Google Trends between Search and Gemini. It would be fascinating to see and compare the results in a comparison chart, especially for certain keywords where perhaps having a dialogue with an AI system might be more valuable. And like you say, seeing that comparison between the two journeys that users can take might help answer the consumer understanding question as well. So over time, I think seeing the trends in the models will also be very valuable outside of search, and like you say, Google is well placed to provide that, especially with Gemini.

Amelia It's a case of watching this space to see where it goes. As always, another really interesting discussion. I'm sure we'll touch more upon this in the coming weeks. Before we wrap up, shall we have a game of Jargon Busters?

Paul Yeah, let's give it a go.

Russ Yeah.

Amelia I feel like you've both been doing pretty well recently.

Russ Really? I feel like I always come away having to look up what you've asked and thinking, I didn't answer that particularly well.

Amelia No, not at all. You both did very well last week. So if anyone's never caught this before, Jargon Busters is where I put the guys' industry knowledge to the test. I've got a list of industry terms, and each week we see if the guys can explain their definition. This week's term is compound interest. How are you going to explain this one?

Paul That's the concept of your interest earning interest. The simple model is you put money into an investment, that earns interest, which means your pot grows, and then the full pot continues to earn interest, so it compounds over time. Basically, your interest is earning interest as well. That's the way I'd describe it. I don't know if that's the formal explanation.

Russ Yeah, numbers aren't my strong point, but I think it's interest on interest. For example, if you have an investment pot and you continue to put in £50 a month, say, then you'll also get the interest on the £50 and the interest on the pot, and they combine together, and over time you get this compound effect. Which is why it's good to keep topping up things like pensions and investment pots, because of this process.

Amelia That was pretty good from both of you. The definition we've got here is: the interest calculated on the initial principal and also on the accumulated interest. So yeah, exactly what you both said. Well done!

Paul That felt like a fairly simple one.

Amelia You did well this week. But of course we will have more Jargon Busters next week. We'll be back for another episode next Friday. Thanks so much for listening, and we'll see you next week.