Signal Loss launch: why Google Analytics isn't as useful as it used to be

Episode transcript

Amelia: I'm Amelia.

Paul: I'm Paul.

Pat: I'm Pat.

Russ: I'm Russ.

Amelia: And this is Fin the Week, a brand new episode. Welcome back everyone, and the first one in Q4.

Paul: Yeah, can you believe it? We're now on the downhill slope to Christmas, aren't we?

Pat: You dropped the C-bomb. Can't believe it.

Russ: Paul, you were saying that this is the 26th episode? Is that correct? Wow.

Paul: Yes, consistent for 26 weeks now. I'm sure there's a stat out there that says if a podcast makes it past that point, then it's got a good chance of surviving.

Amelia: Not bad.

Pat: We need some sponsors now, don't we?

Paul: Yeah, give us a call.

Amelia: And if anyone's listening and they think, "26? I've only listened to a couple," get yourself back to episode one and work your way through. There's plenty of material there.

So this is episode one of a new quarter, which means it's time to launch the next edition of our series. Throughout Q3, we focused on what we called the trust reset: the idea that trust in financial services is no longer something you can claim, but something you have to prove. We started with Consumer Duty three years in and tore it down sector by sector, then trust versus credibility, AI bias, accessibility, pensions, AI-powered fraud, and last week we spoke about the bots quietly blocking your site from search. Regulators, customers, fraudsters, crawlers, and a whole quarter on who gets in. Which leaves one question hanging: once you've decided who you trust, can you still see what's working? And that's where we pick things up this quarter. Paul, what are we looking at in Q4?

Paul: So the edition for this quarter we've called Signal Loss. It's about how the demand for accurate information has stayed consistent, or perhaps even grown. There's more demand now than ever for accurate information about your marketing campaigns, your business's performance and whatever tools you're relying upon. But access to that accurate information is possibly more difficult than it's ever been.

We'll get into the detail of what we mean by that over the course of today's episode and throughout the coming weeks. But in summary, tracking online performance has become difficult, and now feels like the right time to be discussing that because we're moving into the final quarter of the year. This is the time of year that firms will be looking back on what's worked well so far this year, and they'll be thinking about what 2027 looks like.

I think there are a couple of reasons why there are challenges at this time of year. Firstly, firms will be planning on arguably less stable ground than they ever have done before. Even compared to 2025, if you cast your mind back, AI was much less prevalent than it is now. It's hard to imagine that, but we as a firm use AI far more in our day-to-day work and in our research than we did this time last year. So the ground has moved there. And then equally, it's time to plan for regulatory reports. Again, more so than ever, regulators are relying upon clear information. They want to see evidence of how things are going. But I would say gathering that evidence is trickier than it's ever been, for reasons we'll get into.

So each week this quarter, we're going to be taking one place where a signal that marketing, operations or compliance rely on has gone quiet. That could include rankings, AI visibility, consent, conversions, and the information that's relevant for a board pack. We'll be exploring why that signal has gone quiet, what firms can do to replace it, and how they can react in the world we find ourselves in now. And just as importantly, what can you still honestly say to a client or a board without over-claiming? Because it's very easy now to look at a stat and read too much into it. So we'll be trying to unpack where the truth lies. That's the theme for the quarter ahead.

Amelia: So that's the quarter ahead, but this week we're going to set the scene by taking a look at some of the key challenges facing marketeers when it comes to web tracking and attribution. So where do we start, Paul?

Paul: It's probably worth mentioning that we had a chat before this, and Amelia, you mentioned a lot of this goes over your head because on the face of it, it's quite technical. So hopefully we can ask some basic questions and unpack what it all means, because I think it's something that affects everybody, but a lot of people might not necessarily understand the detail of it.

To set the scene, the situation we find ourselves in right now is that marketers have never been under more pressure to prove the worth of what they're doing. But getting the evidence is harder than it's ever been since the web became the thing to focus on. Picking out just one stat from the CMO Survey, which is a US-based survey, this spring's edition spoke to over 300 CMOs, and the most common response to the pressure to prove value was to develop stronger marketing performance tracking. 86% of the respondents said their response to that pressure is to just get better at tracking.

But within that is an issue, because there's this concept of signal loss, where tracking is harder. The perfect place to start with that, and we alluded to it last week by chance, is the difficulty with tracking website performance. In the past, everybody used Google Analytics. You would install it on your website and it was pretty reliable, you could argue. There have always been caveats to that, but generally speaking, you could rely upon it to give you a pretty clear picture of how many people came to your website, what they looked at, and where they came from. You could even see specifically which keywords they typed into Google search to send them to your website. Slowly but surely, that usefulness has been eroded. They removed the ability for you to see the keywords that people had typed in, and different features affected the value of the signal you were getting.

But the big change was with cookie compliance. You hear a lot about it, and as you browse the web, you'll see cookie warning messages asking you to accept certain terms. Actually, a question for you, Amelia, because you might not know the ins and outs of what cookies are. What's your reaction when you're browsing the web and you see all of these pop-ups before you look at anything that say "accept"? What's your gut thing to do?

Amelia: I just think, oh, that's annoying. Just click on something, get rid of it.

Paul: Yeah, and that's probably what most people do.

Amelia: So how would you describe that in its simplest terms? What is a cookie?

Paul: Pat, take it away.

Pat: Have you got half an hour? One of my favourite topics. Cookies were designed at the beginning of the web, really. They are effectively a little bit of information that a website leaves on your machine when you visit that site. So you go to amazon.com, and the Amazon website can leave a cookie on your machine. When you next go to the Amazon website, your browser will pick up that cookie and send it back to the Amazon website. So it's like a cookie trail, and Amazon can see that you've been there before.

It can store information in this cookie. It can store products you've looked at, it can store a key so that you can stay logged in, it can store other preferences that you may have saved on the site, and it allows you to maintain a session from one browsing session to the next over multiple days or weeks. When a cookie is set, it has an expiry date, so after 24 hours, three days or 90 days, your browser will automatically discard that cookie. This is why when you clear your cookies, you'll generally get logged out of all the sites that you use, because cookies are used to maintain login sessions. That's the fundamental thing they're useful for, and that's what they were designed for: to maintain login sessions, to maintain carts, to make a website more usable.

But quite early on, they started getting abused, primarily by third-party providers. A third-party cookie works like this. If you go to the Amazon website, they might have a script on that site that's downloaded from a third party. It might be cheaptracking.com. That script, because it's running in the background on your computer, can also save a cookie. Every time you visit the Amazon site and your browser loads that script, it will send the cheaptracking.com cookie to the cheaptracking.com website rather than to Amazon. Cheaptracking.com is an advertising provider, and that script might appear on a thousand different websites across the internet. So when you visit Amazon, cheaptracking.com will pick up that the cookie has been saved through the Amazon website. Then you'll go on to sportsshoes.com and that cookie will be updated. Suddenly cheaptracking.com has a picture of all of the sites you visit across the internet. That's third-party tracking, and this is what has effectively been shut down as a result of the ePrivacy Directive and cookie law.

Another way to describe how invasive this is, is to imagine it in a physical sense. Imagine you're walking up the high street and you pop into WH Smith to buy a newspaper, and there's a gaggle of annoying-looking individuals in the corner with clipboards. One of them comes up to you and says, "Hello, what's your name?" Then you leave the shop and this individual follows you and you can't get rid of them. You go into the next shop and the individual's just there, writing notes on everything you're looking at. You go into the doctor's surgery, you do all your errands, and the individual is following you about and making notes this whole time. When you finally get home, the individual leaves you alone, but he's waiting outside your house, and the next morning he follows you again. He follows you around for three months and you can't get rid of him.

What this individual is doing is of no benefit to you, but he's taking all this information and selling it to advertisers. And then suddenly you have a load of people knocking at your door day in, day out. "I saw you visited WH Smith on the 18th of February and you looked at the Guinness Book of Records. Did you know the new Guinness Book of Records is coming out next week? Would you like a discount?" They're taking all this information without your permission, they're building up a profile of you in great depth, and they're using that to make huge amounts of money selling advertising back at you.

This is the abuse of cookies that happened from the mid-nineties through to the early 2000s and up to about 2010, I suppose. It was a complete Wild West, and it created an environment online where you could literally track anyone in great detail everywhere they went. From a marketeer's perspective, it was absolutely golden, because marketeers went from throwing 50 grand at a motorway advertising banner and hoping people would look at it, with very little actual ability to track, to spending pence on a highly optimised pay-per-click ad that would generate 58 clicks, of which 23 converted, of which 14 were above-average revenue. You got all this amazing data. It spawned an entire industry, but it was spawned on shaky ground, because underneath this whole industry was a huge abuse of privacy. That's what the cookie law aims to address, which was the ePrivacy Directive. I can't remember exactly when it came out, maybe 2012 or thereabouts.

Paul: And is that what led to all of these cookie banners that we now see?

Pat: Yeah. What that law dictated was that a user needs to opt in and provide explicit consent before a site sets or reads any non-essential cookie or similar tracker on a device. An essential cookie is the cookie that keeps you logged into the website, the useful cookies that the website needs in order to perform its primary purpose. A non-essential cookie is like Kevin with the clipboard, following you around the internet.

Amelia: And would you ever opt in, Pat? Are we always ignoring Kevin and his clipboard?

Pat: You're always ignoring him. And then in 2016, things became even more challenging for tracking with the introduction of GDPR. The ePrivacy Directive focused mainly on cookies rather than personal data, and at the time it wasn't really implemented that widely or properly. Most websites just put up a notice saying "we use cookies, OK?" In recent years it's become stricter and you have to have an approve and a deny button, clearly labelled.

GDPR added extra regulation over personal data. That meant that unless the user opted in, you wouldn't be able to track their IP address, what they purchased, or their name. The user needed to opt in and fully understand exactly what you're doing with that information before that data could be processed. Then you had the rise of browsers like Firefox and Safari that completely disable third-party cookies by default, which rules out any form of tracking of that nature. Chrome is actually the only mainstream browser now that enables third-party cookies by default, but that's mainly because Google's primary business model is selling advertising based on cookies.

Paul: That makes me think of a question I've never thought about before. If you're using, say, Firefox, which rejects them out of hand, and you arrive at a website, you see a cookie banner, and the clear button is "accept all". A lot of people will just say, as you mentioned, Amelia, "get out of the way, I want to carry on," and click accept all. Would that then enable third-party cookies for that site, or would Firefox still say no?

Pat: Firefox will still say no. And actually, I think the whole ePrivacy Directive has been implemented in completely the wrong way. I don't think the onus should be on the website to control when cookies are set, because it's created this heinous UX disaster across the internet where every website has a different implementation of a cookie pop-up. Most of them are awful. You have to approve or reject, some websites do it properly, a lot of websites don't do it properly, and some websites don't do it at all and track you anyway. It's just not a nice thing. It's a step back for how the internet should be.

My view is that the browsers should be controlling this. The law should dictate that if you release a browser into a jurisdiction where a privacy directive like this exists, then when you first open that browser you should be asked: do you agree to third-party cookies? Do you agree to cookies in general? What are the sites you agree to cookies for? So you can maintain a whitelist of sites that you trust, and for the vast majority of sites you go to, you wouldn't see a cookie pop-up because cookies would just be disabled by default. That's how it should work. I think there is legislation in the works that's moving towards that kind of solution, and there are a lot of browser plugins you can download that surface that kind of behaviour anyway and disable cookie pop-ups. I'm going to stop there, because I risk going off on a tangent.

Paul: That's an interesting point, because if you look at the market aside from Chrome, it has probably gone down the route you've just described anyway, where you control it at the browser level. But Chrome is the market leader by a long way. I'm just having a quick look: they're losing a little bit of market share, but they're still at 66%, so two-thirds of the browser market, and they're the ones who don't set it at the browser level.

That's a nice segue into the tracking issues we started with, which is that Google Analytics, or GA4 as a lot of people will refer to it now, is affected by all of this. If people reject these cookies, GA4 doesn't work, basically. That's the gist of it, isn't it, Pat?

Pat: Yeah. GA4 relies on cookies to operate. It's quite a sophisticated tracking tool and cookies are a fundamental part of how it works. Even though they're not advertising cookies, they are still third-party cookies and they are still tracking cookies, so they're not essential and you have to opt in to them.

Over the last few years, we've seen a real increase in the proportion of cookie pop-ups that are implemented correctly, which means there is a clear approve button and a clear deny button. Most people, when they see a deny button, will click deny. That has resulted in Google Analytics effectively becoming useless as a tool for measuring actual traffic. It's still useful for measuring trends, but if you want to say, "Today I received 50,000 visits on my site compared to 45,000 yesterday, therefore I've seen an improvement," it's just not any good for that.

Amelia: So website analytics don't actually describe what is happening day to day. Is that what you're saying?

Paul: Yeah, basically.

Pat: For Google Analytics, yeah, for sure. We use an alternative tool called Plausible quite a lot, which tracks website visits in a completely anonymous way. It's GDPR compliant, it's cookie law compliant, and it just doesn't use cookies at all. All traffic is summarised and crunched in a way that's completely anonymous and then saved in an anonymous fashion. Plausible actually gives you real numbers, but it's not as sophisticated as Google Analytics, so there's a lot of stuff that it doesn't do.

On the flip side, over the last year we've seen a massive increase in the amount of bot traffic hitting sites as well. These are search engine spiders, AI spiders looking to train models, but also traffic from people using agents to carry out research on their behalf. Arguably that is actual traffic, but it's a different kind of traffic and you can't differentiate between them. So you've got all these different factors poisoning the quality of this core data.

Paul: Amelia, are you any clearer on what cookies are and what those banners mean?

Amelia: I think so. Don't test me, but I've got a clearer picture. Paul, you're going to tell us a bit about some stories of client data being decimated. You've got some examples here.

Paul: I won't obviously mention names, but to be honest, it's affecting all clients. As Pat mentioned, if a company implements the correct cookie policy and has a compliant cookie banner on their website that enables people to opt in or stay opted out by default, then you see the level of data coming into Google Analytics being, well, decimated is the word.

We've had multiple instances where a client will have the conversation with us about cookie compliance. We'll have identified that their site isn't currently compliant, we'll put something in place that gets signed off, and you communicate the fact that this will impact their analytics, but nobody ever seems to be prepared for the level at which it impacts. I was looking around for some stats and I found one from a firm called Clickport, which said that with a compliant banner that lets you reject or accept, practitioners typically see 20 to 30% of visitors decline. I would say from what we've seen, that's a much lower level than reality. We've seen sites where you probably get 20 to 30% of visitors accepting at most, so your traffic could drop by 80% overnight when you implement the right thing. It means your analytics is either useless or much less useful than it used to be.

And then it compounds. I don't have any stats for this, but my gut instinct has been that it's not a random allocation of people who accept or reject the cookies. What you probably get is a certain type of customer who just accepts and a certain type of customer who never accepts, so the data you do get isn't necessarily clean, because it's probably a segment of a certain type of person.

Amelia: What type of person do you mean, Paul?

Paul: How technologically savvy they are, their age maybe. It depends what type of company you are, but we're obviously working a lot in the wealth management space, and that might be a good example. There might be a certain portion of that company's customer base which is older and less comfortable with technology, not because they're old, but because they just aren't that interested. There might be a certain segment that just doesn't care about it and therefore accepts more often, and you're skewing the data. Whereas there might be a younger portion of your audience that's inheriting wealth and is much more switched on to this stuff. These are just random examples, but you're probably getting an unbalanced view of your audience. You're not getting 20% of every segment; you're probably getting way more of one segment and way less of another.

So when you say, "This page on our website is getting more visits, therefore it's good content," you're not necessarily able to say that, because the audience might be looking at another page on your website, but they're the portion that's not tracked. It's a real murky area. What I've concluded from most of the examples I've seen is that for some clients, analytics is still useful, because it still gets a decent share of traffic and it's good for seeing things like how efficiently the site loads. Generally speaking, it lets you see what parts of the website people go to. Maybe not specifically which articles people read, but just knowing people go to the articles section is useful.

And then the other thing is that you have to rely upon a suite of tools and not just one. I think that's where there's probably a conversation around a cultural shift, because marketing departments will have been used to using Google Analytics as their source of truth. I expect a lot of firms have now realised that it can't be that any more, but I wouldn't be surprised if there are some out there who still believe in it. It's probably worth having a look at how you've got it configured, because it's probably not as accurate as you think it might be. It's a big change.

Russ: From a UX perspective, we work with data quite a bit as well. If you're thinking about signals, we have a set of what we call UX signals. In order to see whether an interface is usable, we would set positive success metrics and negative success metrics, and then formulate some signals off the back of that. For example, depending on the website, you might want average time on page to be short, because it isn't an article page, it's a page that you want people converting on, so you want people to make that decision quickly. You'd go through the pages and see what would be positive or negative UX signals.

It's quite interesting, because this restriction of data does affect that process of looking at a page and seeing whether it's usable or not. In terms of visits, that perhaps isn't so important, because you can still have a statistically significant sample size to measure something like average time on page. A thousand people would be quite a good sample size to see how long people are spending on that page, even if that isn't all the traffic. But it's interesting to hear: who are those thousand people on the page? Is it a mixture of traffic, or is it all people from one segment? It isn't giving a true representation.

One trick we use, coming back to the source of truth, is to try and triangulate the data. If analytics is telling us one thing, say average time on page is a lot higher than it should be for that page, we could then look at screen recordings in another platform to see whether people are scrolling a lot and what they're doing. We'd look at scroll maps and heat maps to match up that data set. And then it's good to have a third source on top of that, which is interviewing users, speaking to them and seeing what they're struggling with on the page. If you can triangulate that data, you've got a sweet spot in the middle where you can make some clear judgement calls, because it's really difficult to look at one source as the source of truth. It sounds like that's also what we're talking about here: multiple sources, and finding the truth in the middle.

Amelia: I wonder, from your point of view, how do you feel about cookie banners being the first thing that a brand presents? What challenge does that present in itself?

Russ: I think it's a big challenge from a UX perspective, to be honest. First impressions count, don't they? If you're visiting a website for the first time and the cookie pop-up design hasn't been thought about. We were auditing a website recently for a client, and on mobile the cookie pop-up covered 75% of the screen. I was using it again this morning, and when you click on "adjust settings" you can't actually use it at all on mobile, and it's difficult to close. That might put someone off accessing the site at all. I'm sure if they really wanted to speak to that firm, they'd figure out a way to close it or find another way to contact them, but it's created quite a bit of friction at the beginning of that journey.

It's important that cookie design isn't just an afterthought in the design process. You think about the colours that you use, you want the typography to match the brand, and you make sure it doesn't take up too much of the screen. These aren't difficult things to get right, but it feels like a lot of these cookie pop-ups are templated or third-party and not always easy to customise. And as you mentioned at the start, Amelia, you just want to get rid of it and get to the page as quickly as you can. That browser-level setting would transform the web, really, because almost every site you go to is littered with these cookie pop-ups.

I was also looking at it from a regulatory perspective, in terms of what websites should be doing. I think the ICO in the UK explicitly looks for equal weight between accept and reject buttons: equal emphasis, equal colours. We shouldn't have the primary button being accept and the reject button being secondary, which is what you often see, and I'm sure a lot of websites are guilty of that. The French regulator actually fined American Express for this kind of deceptive design pattern. I think it was a 40,000 euro fine. There were three things that formed part of the reason they were fined, and this was one of them: the imbalanced choice presentation. The wording of this I thought was quite nice. The regulator views the design asymmetry as an impermissible nudging technique, a form of deceptive design or dark pattern. So what they're saying is, if you've got these buttons and they're not symmetrical, and you're visually nudging the user towards accept, that's a breach of the regulation.

I'm not sure how often it's been enforced, but there are two things here. There's the branding of these cookie pop-ups, so they match your brand and it isn't too jarring when you land on them. And there's actually a regulatory angle to implementing these as well, where you can potentially get a fine. So it's more than just a tick-box exercise for the web team. They need to get it right on a couple of levels.

Paul: Your point about triangulating data, that's the ideal way to put it. I think the incentive right now is that a lot of brands think, "We want to be able to track traffic to our website, therefore we want people to accept," and that, subconsciously at least, pushes people to put in place cookie banners that try to drive you in that direction. I get so frustrated with the ones where you click reject, or the option isn't even reject, it's "customise", and it gives you a long list of radio check boxes to decide what you want and what you don't want. I know a fair amount about it all and I don't even care to understand all of that. It doesn't surprise me that firms do that.

But if firms instead decided that Google Analytics isn't going to be the one and only source, it will be one of them, and alongside it they have something like Plausible, which we use, as you mentioned, Pat, which doesn't rely upon cookies and therefore is much more reliable, and maybe Google Search Console as well, then you've got three pretty good sources that will all tell you a similar story. That's where you draw your conclusions from.

When it comes back to the central conversation about the board coming to you and saying, "How many visitors did we get over the last quarter?", I think there's an education piece there. Absolute visitor numbers aren't worth trying to get, because you never will agree upon an actual number. You'll get in the ballpark. So get yourself in the right ballpark, understand the trends, and always compare like-for-like data sources. The trouble with Google Analytics is that once you implement the correct cookie policy, it becomes impossible to compare September this year with September last year, because last September you didn't have the right policy in place. You have to get comfortable with that fact and start to rely upon this triangulation approach. Make sure everybody understands we're looking at trends and direction of travel. We're not talking about how many people came through our site down to the single user, because you won't get that number any more. I don't believe that exists anywhere.

Pat: Numbers like conversion rate are a bit meaningless these days, aren't they? But there are still loads of really useful KPIs that you can track. The number of actual conversions, obviously, whether that's orders on your e-commerce store or enquiries through your web form, is something you can definitely measure. You've got a wealth of data in Google Search Console, which you mentioned, Paul. That's not user data, because it's data compiled from how Google has indexed your site and literally where Google is sending users that have used its search engine, so that's still really accurate. You've got tools like Plausible that give you actual numbers, but those numbers are often inflated by bots. Plausible does a reasonable job of filtering out a lot of the bots, but not the AI research agents and the AI agents that are crawling for content rather than for training. We talked a bit about Cloudflare last week as well, and how Cloudflare can be used to prevent bots, but then you also need to allow some bots in because they're working on behalf of actual people who might buy actual services. So there's still a wealth of data out there.

It was almost too good to be true, those heady days from the mid-nineties to the early 2010s where you could have everything you wanted as an advertiser. I think the advertising industry needs to accept that cookie-based tracking has ended, and adjust itself to the actual metrics we've got at our fingertips now rather than trying to figure out how to introduce alternatives to cookies.

It's quite interesting to go through Google's various attempts to work around the cookie law over the last 10 or 15 years. I won't spend too long on it. In 2019, Google announced their Privacy Sandbox, which was an alternative to cookie tracking. It promised that fingerprinting, which is something we haven't really talked about today and is an alternative way of tracking users that's pretty much completely illegal, was wrong. In January 2020, Chrome committed to phasing out third-party cookies within two years. In March 2021, Google launched an initiative called FLoC, Federated Learning of Cohorts, which tried to anonymise users by grouping them into cohorts with different preferences. That was prevented by GDPR and other privacy-related legislation. In January 2024, they blocked cookies for 1% of users, realised it was going to decimate their business model, so delayed it again. In July 2024, they abandoned the cookie switch-off completely. In December 2024, they U-turned on their fingerprinting position, effectively allowing advertisers to fingerprint if they wanted to, even though it's arguably illegal. Then they got quite a big fine in 2025 over Gmail ads and cookie consent. In October 2025, the Privacy Sandbox APIs were retired, and it feels like they've given up on trying to remove cookies and are just continuing to run with them. In a more concerning development, they've actually started to use IP addresses as a base measurement for targeting ads now, which is personal data and something they said they would never do.

Google are really, really trying their hardest to maintain ad targeting, because it funds their entire business model. I don't blame them, really. It must be challenging for an organisation of that size, with so much revenue, to have this legislation looming over them that they can't manoeuvre around.

Paul: The clinging to cookies thing feels almost like they're fighting a losing battle, doesn't it? Whilst they control the browser market share at the moment, another stat I was looking at was that in the UK, 45% of mobile browsing is on Safari, with people using iPhones, and Safari will block those types of cookies. So how useful is it going to be? The share of people using browsers that block this out of hand is only going to grow, and then you've got ad blockers.

Pat: The industry just needs to abandon cookies. They're never going to be more useful tomorrow than they are today. It's only going down. Every day it's less useful than the day before. The industry needs to accept that, adapt, suffer the pain of change, and find a new normal.

Russ: The other thing is, people are visiting websites less, aren't they? They're finding out information through AI search. How long is it going to be before you can complete a transaction easily through Claude or ChatGPT and you won't have to visit a website at all? That's going to throw this whole cookie muddle even further into the dark ages.

Pat: Then on the flip side, you've got pretty recent news of tools like Meta's Muse agent and OpenAI's agents. These are AI agents that effectively go out on your behalf and do stuff on the internet. They're like personal assistants. If you look at Meta's history of the use of personal data, there's no way I'd ever entrust any of my personal data to their Muse agent. These agents open up a whole new realm of privacy abuse and advertising possibilities. So whilst cookies are under control now, there's this huge risk of big tech using AI agents to continue personalised advertising by other means.

Paul: Yeah, I think there are loads of subsets of things we could discuss. But to bring it all back to the essential thing we raised at the start, which is that your analytics probably isn't as useful as it used to be, I feel like the big takeaway from today's discussion is that idea of triangulation: putting in place a model that lets you rely upon multiple data sources and drawing conclusions from the combination, not from a single source. That's what I think will help people when they start the process of planning for the year ahead.

Amelia: That seems like a good place to leave it. A really interesting discussion to kick off Q4, and I'm sure lots of these topics will rear their heads in the coming episodes. Before we go, should we do some Jargon Busters?

Paul: Yeah, hopefully we've got another strong one this week.

Amelia: So this is where I test your knowledge of industry terms to see how you get on. This week's term is "junk bond". Do we know what this is?

Paul: I've heard it before.

Pat: I've heard the term, but yeah, I'm unsure.

Paul: The thing that's popping into my head is in The Wolf of Wall Street, when near the start he starts getting involved in trading and he's selling those penny stocks, or whatever they call them. It reminds me of that, but no, I don't have a definition of junk bond, actually.

Pat: A bond is effectively where a company or a government will sell a bond to an individual or an organisation for a fee in order to raise money. We talked about bonds last week, didn't we, in the context of quantitative easing. You sell a bond and then you pay the owner of that bond a fixed fee every month over a set period of time. So you might sell a bond for 100 quid, you get a cash injection of 100 quid, and then you pay that person a £1 a month fee for 110 months, so the person who bought the bond gets a rate of return of 10%. As a wild guess, a junk bond is probably a bond that's sold by a firm that hasn't got the financial stability to maintain those payments, perhaps.

Amelia: Fair guess. Are you going to have another go, Russ?

Russ: I don't think I've been yet.

Amelia: Oh, sorry, Russ.

Russ: No, I shouldn't have said anything. I should have just waited for you to say a few more words before I went. But yeah, I understand that from an investment perspective, bonds are quite safe. People move their money into bonds, perhaps depending on what the market is doing, for that security, but you perhaps wouldn't make as much money from them as you would from a more risky investment. That's my understanding of them, or of government bonds anyway. So I guess it's not a government bond. "Junk" means something that isn't good, that's rubbish, so it's a bond that perhaps doesn't give you any return on your investment.

Amelia: So a junk bond is a high-yield, high-risk bond with a low credit rating. That is what it is.

Pat: So kind of what I said, but I didn't allude to the high-yield, high-risk part of it. You might buy a bond for 100 quid from a company that's struggling, and they might agree to pay you 30%, but there's also a risk that the company might just go under and not be able to repay you.

Amelia: There we go. Well done, a good stab at that. Of course, we will have more Jargon Busters next week on a brand new episode. Looking forward to getting into it. We will see you next Friday. Take care.