.png)
Insights, Marketing & Data: Secrets of Success from Industry Leaders
Insights, Marketing & Data: Secrets of Success from Industry Leaders
ASK BOSCO - JOHN READMAN, CEO/ FOUNDER. Mastering marketing attribution; lessons from AI-driven forecasting; integrating multi-platform data; educating CEOs on budget allocation; the journey from the RAF to digital marketing innovator.
Imagine you’re buying a new bike….you’ve been to multiple websites, you’ve seen several ads, you’ve clicked on some but not others, read reviews on different sites and…eventually you buy the bike. However, how does the bike seller understand the contribution of each marketing channel? And where should they invest next for other customers?
This is just one of the questions, which John Readman founder of Ask Bosco addresses in the latest Futureview podcast. Originally recorded in the summer of 2024 (hence some of the references) we cover a great range of topics including:
How to educate the CEO and CFO on how marketing spend really works
- Integrating marketing data to create a level playing field
- Ai tools to forecast marketing effectiveness
- The importance of brand, creative and being brave to stand out
- The journey from Dragons Den and the RAF to the world of marketing
- Lessons from John’s Dad
All episodes available at https://www.insightplatforms.com/podcasts/
Suggestions, thoughts etc to futureviewpod@gmail.com
and and ultimately, what we're trying to do by doing this is right is to open up the recognition of the contribution that the marketing is having further up the funnel. Because if you just focused on last click, you would stop spending money on branding, you'd stop spending money on display, you'd stop spending money on facebook advertising, and then that might work in the short term because, oh, we've spent less money and we're still getting all these google pay-per-click text. But maybe in two months or three months time, if you've got a considered purchase, you might not be getting as many google clicks because nobody was aware of your brand further up the time yeah, yeah, it's kind of like about some demand generation, isn't it like yeah?
Speaker 2:and we see that a lot. We see that a lot people going oh well, we turned off facebook or we turned off display marketing or we turned off retargeting, and it didn't have an impact.
Speaker 1:And then, three months later, that's when the impact will happen welcome to future view and an exploration of an angle we don't often cover in this podcast. Hence part of the reason why I was interested, namely direct response and how you analyze effectiveness of marketing allocation and, indeed, creative indirect response to sales. John Redman is the founder and CEO of a platform that does exactly that, and he'll explain all. John is also charming and a really interesting guy with a great backstory. He's from Yorkshire, so for Americans he sounds a bit like Jon Snow from Game of Thrones. He's also the brains behind an amazingly brilliant charitable endeavour where he's trying to cycle to Australia. Yes, that's really true. Listen in to find out more. So, John, welcome to the podcast. Delighted to have you on.
Speaker 2:Thank you very much for having me, henry, looking forward to it.
Speaker 1:Now you probably know where I'm going to start A little bit of an icebreaker. What's one thing? In fact, there are probably quite a few things, because we've chatted before that most people wouldn't know about you or they might find surprising.
Speaker 2:Well, I do have a story about being out on a night out in Vegas with Tom Jones, but that's probably best kept off a podcast.
Speaker 1:Oh, I don't know. Now, what?
Speaker 2:now I want to know but the the one thing not many people know is I've actually been on dragon's den. So I went on dragon's den, uh, to pitch the cycling event business that I set up with a friend of mine, so, um, and we went on just to get on telly, not to get the money, um, so yeah, if people wanted to have a google of john redmond dragon's den ride 25, they can.
Speaker 1:They can see why peter jones and thinks I was taking him for a ride, uh, so he wasn't very positive about cycling the other thing you mentioned to me that maybe more people do know, when we first met is you very modestly described yourself in a very self-deprecating way as fundamentally a failed RAF pilot. So what's that all about?
Speaker 2:Yeah, so I grew up right next to an airfield in Yorkshire and I always used to look up thinking I want to do that.
Speaker 2:And my father was a British Airways pilot at the start of his career and I always wanted to be a pilot and I went to air cadets and then school and university and joined the university air squadron and they taught me how to fly and I had an opportunity to join the air force, but not necessarily in the job I wanted.
Speaker 2:So I took my bat and ball home and went to London where the streets were paved with gold. But I've got friends who also didn't get in and then went back and are now very senior officers running running the air force and I've still got a very streets were paved with gold, but I've got friends who also didn't get in and then went back, uh, and are now very senior officers running running the air force and I've still got a very fond connection with the air force. I think it taught me a huge amount and I've still gotten a passion for flying and, um, all things aviation and military. So, um, yeah, it's, um, yeah, I, I, who knows what would happen. I think I'd have probably got kicked out because I'm not very good at being told what to do, so I can't believe that for a second.
Speaker 1:So, moving to a different type of flying or possibly actually to the next business that you created, as I mentioned, we're going to do something slightly different in this episode by exploring some broader marketing concepts that I think Consumer Insight listeners will have heard of but maybe not that familiar with. So it's latest trends in SEO, attribution marketing and that type of thing, but possibly within the context of the business that you've set up, Ask Bosco, and what that does. So could you tell me what does Ask Bosco do and what problems are you looking to solve with it?
Speaker 2:So what Ask Bosco is fundamentally trying to do is help brands and agencies consolidate and connect all their marketing and performance data into one place, and then it enables them to have a single view of the truth, hopefully all in one place, with all their different channels all stacked up next to each other in a very neat and friendly sort of dashboard.
Speaker 2:We've, then, now built out some AI and natural language sort of questioning within that, because I think it's fair enough having lots of pretty dashboards, but if you don't know which dashboard you need to go to to find the information you need, you're then just lost in loads of data.
Speaker 2:So we've now added in some AI so you can ask the dashboard questions in natural language so like what was my best seller on Facebook, or how does Facebook marketing compare to Google marketing and it'll then produce the graphs for you. And then, once we've organized all that data together, we then have a a sort of third party benchmarking tool within Ask Boss Go that lets you benchmark yourself against your peers. So the sort of dashboards and insights part gives you here's what's happened, and then, if you're having a bad month, it'd be good to know, or good months, it'd be good to know well how's the rest of the market doing compared to us. So we benchmark you against your peers and and then you can see well, actually everybody's having a bad month. It's not just us, or actually, well, everybody else is. We're having a good month and they're not.
Speaker 1:We must be doing something really well and john, if I, if I back up just like a little bit sorry to interrupt, but it's primarily around retail focused marketing, is it? I mean? Or if we backed up a little bit in terms of the types of companies you're working with and the types of data that you have within the dashboard in terms of marketing channels, it's primarily digital marketing channels, isn't it? As opposed to? It's primarily digital marketing channels, isn't it? As opposed to it's not TV or outdoor or radio or that type of thing.
Speaker 2:Not yet. So, basically, we support anybody who spends on Google, facebook, tiktok, amazon, all those sort of digital marketing channels and predominantly you're right Henry it is e-commerce, because they tend to be the bigger spenders who are trying to look for those sort of incremental opportunities and gains within their data and understand where to spend their money and which channels performing best but what it could work for anybody. So we've got big. Some of the biggest spenders online are actually b2b companies or travel or insurance or finance, so anybody spending money in multiple places would benefit from getting better visibility or comparing channel by channel next to each other. And then I suppose the clever bit, then, is our own AI and machine learning that we've built that basically enables people to do scenario planning and forecasting of where should they be spending their money and in which channel.
Speaker 2:So in the past past you've maybe had an in-house team or a marketing agency who says let's test this or let's do that or let's move some money around here and see what happens. We can run scenarios and forecast exactly what we believe will happen, or we can run the forecast automatically and say, right, we've got this budget, this is the outcome we want. Where exactly should I spend my money Down to a sort of campaign in each individual channel level. So, yeah, it's hopefully enabling junior marketers within an internal team or marketing agencies all to move quicker and make better decisions. And all these ad platforms and we're becoming more and more a black box. Trying to do is extract that data out of the black boxes and let you have a play with it, using our, our tech, uh, to hopefully enable you to make better decisions. So, and we we provide this as a sort of online sass product for brands directly, uh, but also agencies use it to help speed up the process of looking after their brands.
Speaker 1:I see, and so how do you prove that it works, as you said? So you're making these forecasts. Are you doing like geo tests, or maybe it's not geo tests, given the do you do one campaign that's based on I don't know Facebook's recommendations that's based on I don't know Facebook's recommendations there's two ways to test it.
Speaker 2:What we normally do is we go back in time and remove. So we go back into a period of time. So let's say, go back six months, remove what happened and then ask Bosco to predict, based on what they were doing with their spends, what do we think would happen and how close is that prediction to actually what did happen? So we can go back and validate it very quickly like that, and most times that's north of 90% accurate. There's always things you can't predict. So if we've got a lot of sportswear manufacturers who, and if the team they sponsor wins something, the demand goes up and we weren't able to predict that they were going to win something.
Speaker 2:Um, but it takes into account seasonality. It takes into account sort of um, all the different annual peaks and troughs within trading and that could be different for different verticals. So, like fashion and everything, christmas is a big one but say, for instance, cycling, this time of year coming into the summer is their peak. So it will. It will factor in those sort of historical trends and it tends to look at two years worth of historical data and then we pull a lot of the demand data from all the different platform apis and then how's the forecasting ai bit working from a an overarching perspective, being not not too technical, uh you're not in any danger there.
Speaker 2:So I'm more founded with the idea than the than the phd data scientist um. But fundamentally we're looking at a sort of cost revenue model and that sort of sits at the top and uses a lot of underlying sort of bayesian um statistics too, and we can look at, first of all, an overall marketing cost versus revenue. Then we look at an individual channel and we can look at first of all an overall marketing cost versus revenue, then we look at an individual channel level, then we look at different campaign levels and within that you'll get loads of different graphs which have sort of normal distribution curves of where's the point of diminishing return and where's ultimately the optimal point, which may be different, and this. I suppose all of this could be done by a really clever person with a really big spreadsheet. But the challenge and hard work if you've got multiple campaigns and multiple, or if you're an agency with multiple clients, it's a really good thing to do, but it takes a huge amount of time and effort and often by the time you've done it it's wrong because you've moved on.
Speaker 2:So really, what we're using rather than using, we do use generative AI to create lovely dashboards, but also using sort of the more machine learning side of AI to process huge volumes of data just means we can do this a lot quicker. But if you're saying you want maximum traffic, that might be a different question. Well, we want maximum sale volume, or we want maximum revenue or we want maximum profit. So you can quickly run different scenarios to see the different outputs right.
Speaker 1:And then what do you train? So I am geeking out a little bit here. What do you train the forecasting models against? Because the world I'm usually working in, like a lot of the consumer insights world agencies, really, really struggle to get like the end data to train their models.
Speaker 2:Yeah, against but, but it sounds like you're able to get the end data so you're able to train the model so the current setup which is about is going to be improved actually we're training it on two years worth of historical data from your um, from your trying all your impressions, clicks, transactions and moving you and everything and spend, and then we also can see the demand data in sort of like search impressions within certain keyword clusters within the different platforms. But what we've also done now is because there's a big issue around measurement, because you've got Google saying they're claiming a million pounds, facebook say they're claiming a million pounds, but you've only got one and a half million pounds in utilities.
Speaker 1:So so we're talking about attribution modeling here. So should we? Just should we, or maybe we're not?
Speaker 2:Sort of all. I say I think the challenge here is, were not, but I mean, maybe we sort of are. I say I think the challenge here is, uh, understanding um where and getting it all into one place. So this this, fundamentally, is complicated and hard. So what we're actually doing which is the bit I was just gonna say is we're now also pulling the data in from the e-commerce platforms, so you've actually got every single real transaction that happened. So then we can start looking for why, why is there a discrepancy and what is an acceptable discrepancy, and this is just getting harder and harder. So, with people like buying things on TikTok shop or in different marketplaces which may never turn up in your Google Analytics. So yeah, fundamentally, this all leads into some sort of attribution of like where should we be spending your money?
Speaker 1:And so I guess part of the problem playing it back a bit and correct me if I'm wrong is that you've got the various Google platforms, the meta platforms. Everybody will be reporting back their data, but you don't get an overview of the whole market and, in addition, you tend to get some overstatement around trying to claim the credit for the sale from those platforms overstatement in certain platforms is an understatement?
Speaker 2:um, yeah, it's. And there's a whole question over should you let google's ai spend your money or not, because are you putting the alcoholic in charge of the off license and are they going to spend it in the right way? And we've done lots of testing and actually it's pretty good, but you need to keep your eye on it. But fundamentally, you're absolutely right, henry.
Speaker 2:If you went for a meeting with Google and I'm sure there's people listening today who are either brands or agencies who've spoken to their Google rep and the answer where should we spend more money will be spend more money with google. And then a meta are going to say, oh yeah, spend more money with us, and tiktok would say the same. So, fundamentally that the hard question we're trying to answer is where to spend your next pound or dollar or euro? Um, because we're platform agnostic, we really just want to help the client demystify all of this and and hopefully reduce what they spend and increase what they're trying to do, whether that be inbound leads to their website or whether that be revenue from e-commerce. So that's the game we're trying to help people understand.
Speaker 1:Got it. Attribution modeling, which, as I think insight departments start to merge more with marketing, and that's a different subject. It's kind of in itself, I think, professionals in that space becoming more familiar with these ideas. But could you just give a quick primer, for instance, on last click attribution, what the issues are around that multi-touch attribution and how it's all getting more difficult?
Speaker 2:so it all started, this whole world of marketing measurements really sort of hotting up, and there's a lot of it. There's two main reasons, the first one being around privacy and around tracking. So the sort of removal of the third part of cookies and and everybody moving to different measurement platforms is making it harder, and I suppose also with apple and the ios 14 issue with facebook meant you couldn't track half of facebook traffic. Uh, so, fundamentally, measurements becoming harder, so so that's the sort of overlying problem. Before it was super easy you spend a pound on google, you make 10 pounds, it's working, let's just keep spending money. And I think that also gave a false illusion of how easy this stuff is, or isn't. Um, and then last click, basically what we're saying is uh, the last measurable action that someone took online that we can measure that link to either a conversion whether that be a form fill or or a sale, is what we're going to give 100 of the credit to. So, and then I'll explain why that works, but also why that could be flawed.
Speaker 2:So let's take, for instance, I don't know if I'm buying a new bike. Um, I might research best road bikes for long distance cycling and see two or three different brands, I might land on their website. They may cookie me with their own first party cookie. They may stick on my retargeting cookie, which they're still allowed to do now. I then may narrow it down and start putting actually best specialized versions and then get into the detail and over a period of two or three months, I might sign up for the newsletter. I might look at their Facebook page, I might click on some Facebook ads, I might read their newsletter and click back through to the website. I might watch a video on YouTube about a review of the bike and then, when it gets to payday and I've managed to convince everybody that it's a good idea that I need a new bike I might go on and do a brand search and click on a Google ad, and the Google ad will get 100% of that credit for that new bike sale, whereas actually there's a lot of media that was involved in that process to get me to that point of my decision.
Speaker 2:And, depending on what you're selling, I think a last-click model is probably flawed. What you're selling, I think a last-click model is probably flawed Because if you've got more than, say, four or more touch points in a transaction that you can measure, and that's the key thing now, I suppose, henry, is actually what can you and what can't you measure? I would recommend people should try and understand if, first of all, they need an attribution model and then, if so, what could that look like? And some of the common ones is, say, the u? U-shaped uh model, where they put 40 on the first interaction and give some credit, 40, that 20 then split out across all the other interactions, less maybe direct website visits, and then 40 on the last one. That knocked it over the line.
Speaker 1:And John, can I ask a question about that? I mean, that's very simple to understand and it makes sense at an overarching level. I assume there's some type of science behind that model, though it's not the type of thing I just make up on the back of an envelope and go.
Speaker 2:No, no, no, Right. So the way to do this uh, guessing into it is you need to take a statistically significant number of transactions where you have all the touch points. You then need to run them through. Um, well, what we use gives a given to some clever people to run them through their systems, where we're trying to get to actually henry's build in into the asposko platform, almost like it does this automatically for you, says you don't need one, or yes, you do need one. And here's three recommendations based on this logic um, first of all, you need to see how many transactions over a certain period have had multiple touches, right? So, yes, you do need one.
Speaker 2:And then, if you do analyze, well, what's the best fit?
Speaker 2:And I suppose the best example is attribution, is a, is a, it's flawed, there isn't a perfect fit. Uh, it's the best, least wrong answer, if that makes sense. Um, because, unless you sell one thing to one type of person all the time who buy it in exactly the same way, you're trying to get almost the line of best fit to give you a better and and ultimately, what we're trying to do by doing this is right is to open up the recognition of the contribution that the marketing is having further up the funnel. Because if you just focused on last click, you would stop spending money on branding, you'd stop spending money on display, you'd stop spending money on facebook, you'd stop spending money on display, you'd stop spending money on Facebook advertising. And then that might work in the short term because, oh, we've spent less money and we're still getting all these Google pay-per-click checks. But maybe in two months or three months' time, if you've got a considered purchase, you might not be getting as many Google clicks because nobody was aware of your brand. Further up the tunnel.
Speaker 1:Yeah, yeah, it's kind of like about some demand generation, isn't it like yeah?
Speaker 2:and we see that a lot. We see that a lot people going oh well, we turned off facebook or we turned off display marketing or we turned off retargeting, and it didn't have an impact. And then, three months later, that's when the impact will happen yeah, and would you say?
Speaker 1:would you say, john, sorry to interrupt again. Uh, if you've got a more considered purchase, of which a high-end bike clearly is it's not a 99p something or other you just click on the attribution.
Speaker 2:Modeling would be more important for those types of purchases, I would imagine, as opposed to… Absolutely, I think the attribution modeling is if you've got something that, yeah, if I'm trying to buy I don't know a jumper, right, and it's just a quick 30 quid jumper because I need a new jumper, that might be a one or two click process. However, if it was a 250 pound jumper made out of I don't know merino wool and this, that and the other, and it was a really technical thing for a technical and you've got to research it and it's considered purchase. So cars, holidays, and one of the other challenges around attribution is if the consideration is over a long period of time. Sometimes we don't have all the tracking because some of the cookies maybe haven't got that cookie window, uh, to actually look back. So then it's about having almost like a closed loop tracking model where you're you're tracking all these interactions within your own database or your own crm, uh, so that then you can actually have have the the tracking.
Speaker 2:But you're absolutely right, considered purchases, high value purchases, um, they, they're more important to look at these models. But the other thing to say to sort of anybody listening is marketeers often get this, as do often data people. The problem is educating the CEOs, cfos and the board about why we're measuring the numbers in a different way and about why this is important to unlock the budgets and actually spend the money in the right channels in the right way, and often that's an education piece, and so it seems, without being rude about it, it seems quite obvious the way you've just described it.
Speaker 1:So why do CEOs and CFOs have a problem with it, is it?
Speaker 2:they've kind of got this ingoing perception I think they've all got used to the instantaneous gratification of last click wins uh, okay so spend a hundred thousand pounds in june, get a million pounds in june of sales, whereas Whereas if you're looking at your attribution model over multiple months, you might spend a load of money further up the funnel and you may not get that instant gratification of the sale immediately. So that's why you need to sort of take everybody on a journey and educate them to what we're trying to do.
Speaker 1:I see, and I guess a lot of this also speaks to the importance of brand and demand generation, which actually is what a lot of consumer insight companies are doing.
Speaker 2:Yeah, I think this over the last 15 years because it's been easy to set up on a cheap printer cartridgescom. It ranks really well for printer cartridges. You can buy some traffic, you do no branding, you have a very basic website and you could probably win online. For the last however many years and I I think it's now becoming harder to stand out and it's quite interesting that a lot of companies don't invest in proper branding or proper, I suppose, reach um and maybe wouldn't even consider tv and whereas tv is now becoming as easy to buy as pay-per-click advertising to a lot of the dv, the digital tv channels, and I think represents a huge opportunity and potentially a good value opportunity, but most brands probably think it's outside of their reach yeah, yeah, yeah.
Speaker 1:Again. Also an interesting point. I mean, I think that the growth in connected tv is projected to be huge. Um, just going back to the modeling bit, this is basically what I do on this podcast, if I take stuff I don't really understand and I ask clever people like you to explain it to me. So we did last clip, we did the u-shaped one, and then what are there? Another couple that people should know about?
Speaker 2:there's loads of different models, uh, and some people might even load it even more to the front. So where are we getting the demand? And I I think people shouldn't get obsessed for which model fits my business, which model should I take and force onto my data. They should analyze their data and go right well, what's my data telling me? And often there's two or three that could work. And then what we would normally do is run.
Speaker 2:So we take the test sample over a period of trading and we go right, here's what the numbers look like on a last-click-wins basis. That's what it says you spent in Google, and that's what the numbers look like on a last click wins basis. That's what you, it says you spent in google, and that's what happened. Your spends would remain the same, but it's how you attribute the success of those. And then we take it and say here's what the u-shape, or here's what this shape, or here's what that's, and and or it may be a unique model to your business, and and and then it's about. We would then apply, give you almost like four or five different scenarios of here's how you should measure the impact on your numbers, from which channel yes, actually I was going to get into that in terms of medium it's modeling, I mean.
Speaker 1:I mean the way attribution was described to me by a friend. Tom vice used to be the chief data scientist at an old business. He said it's a bit like a sort of a football team. He said what last click is is you give all the credit to the striker.
Speaker 2:Yeah, that's a good guy. I like that.
Speaker 1:Yeah, and yeah, probably not a good example with Leeds or someone like that, because you know sorry.
Speaker 2:I'm not talking about football at the moment.
Speaker 1:No, I thought, but yeah, but basically he said like so you try to work out who gets the credit for the goal and in a very simple way, you could go okay, there's an assist, was the final pass before the goal was scored. But actually there may be much bigger patterns behind that in terms of why the goal scored. So is that a fair analogy for media mix modeling? If I take this one in terms of media mixed modeling isn't so much about the immediate transaction or the goal. It's trying to work out your patterns over time and who on the team contributed to your overall success or failure rate.
Speaker 2:I also think a lot of the media mixed modeling is done with a retrospective view to prove that we did spend the money or we didn't spend the money in the right place. So it's very much taking into account everything, whether it be TV, out of home print, and it's trying to work out how can we stick all this together and how can we spot the patterns in all of this. But also, I think there's an element of validation to prove that either what we did was right or wrong, and then hopefully it gives people some sort of insight into the future, but a lot of it is. I I think where we'll end up is a sort of a future where there's and Google are trying to build one of these at the moment and Meta's trying to do one, and they're saying that their platforms are agnostic. But I can't believe there's a platform from Google that doesn't give Google a little bit of a couple of extra percent points or bitemers of a Meta. But I think where we're going to get to because all the measurements getting harder, even and I suppose what's been easier, henry, digital measurements been easy for the last 10 years, all right, whereas now it's getting harder because of a lot of the privacy issues, uh, which absolutely is correct. But what we need to sort of look at is where do we, what do we then do with that data, and what is it then telling us, and how are we then going to do things differently?
Speaker 2:Because I still think a lot of the MMM modeling stuff is loads of clever. People go away, sit in a room, crunch loads of numbers for a few months and then come back with a PowerPoint. Everybody slaps themselves on the back, going great, we spent our money in the right place. Well done everybody. But then they're not necessarily answering the question of well, what do we do next? Or where do we spend the money next? Or is this and I think the world's moving pretty quickly now? Um, so what google and meta are trying to do is very interesting, but I think he's he's still quite flawed, because would you trust google to tell you to spend more money on facebook or vice versa?
Speaker 1:I don't know and and at the moment they do do that in principle, the new systems do do that. It's just you've inevitably going to have a little bit of a suspicion as to are they telling you to spend enough on competitive platforms?
Speaker 2:yeah, I I think I think a lot of the people I talk to. So I've been talking recently an imrg event and Justin and I were talking about and he goes and there's some quite interesting platforms coming up. There's almost and there's a lot of brands now saying do we even need Google Analytics and should we look at a completely separate analytic fund? We're not an analytics platform where we're reporting and forecasting platform, and that's going to be interesting to see what happens, because there is genuinely a a trust issue. Um, I think with some of the big spenders.
Speaker 2:There's a big article, uh, yesterday and on the bbc about how, um, people are saying, even like the seo results are are favored in term in the in the direction of the bigger brand spenders on PPC, and that's always been a bit of an urban myth. But there's a few people saying now they've been tracking it for so long they can prove it. And if that's the case, then the whole point of the Google organic search engine was to give everybody a fair chance if they've got some good content or a good product to answer a search query. Chance. If they've got some good content or a good product to answer the search query With all this measurement and everything else, the one number that I think people often overlook and think that they can't do enough about everybody's got addicted to where do I spend my money and how can I reduce what I spend. But actually, some things that's hugely overlooked is improving and focusing on improving your conversion rate.
Speaker 1:Okay, and how do you go about doing that?
Speaker 2:Well, there's a load of measurement and testing involved. That's with all these things, but it's ultimately a lot of people. So, fundamentally, what we're trying to do is then test, so testing different layouts of baskets or different images or different products. So you're already getting X amount of traffic and if the number of traffic stays the same and you've got a conversion rate of, say, 2%, if we could increase that to 2.5%, that's like a 25% increase without having to pay for any more traffic. So actually, the biggest lever in any online business whether you're collecting leads for b2b or whether you're selling things the biggest lever you could pull will be to improve your conversion rate. Uh, assuming everything else stays the same.
Speaker 1:But then you get. But then you get into the tricky thing going back the board question, don't you? Because they their inclination, we go. Let's about the board question, don't you? Because they their inclination, we go. Let's place as much focus the very end of the funnel, trying to improve the conversion rate and then potentially neglect the upper funnel and the brand building and the demand generation yeah, I'd say I'd, I'd hope they'd take it out of a different budget.
Speaker 2:But to focus on that, I'm more around, but often people go all right, the website's finished, uh, and it's a bit like, um, well, and it's like I think the website is never finished and there should always be a plan of testing and of of whether that be conversion rate or UX, like just split testing what's working, what's working Then, even when that test wins, you should then set up another test, because you can always improve it yeah, and and how does the?
Speaker 1:is that how you would evaluate the different creative, because I mean there's lots and lots of, there are lots of studies that show the importance of creative within the the media mix as being kind of a determining factor. So you basically you tend to look at it in terms of split testing.
Speaker 2:On the UX side, we'd definitely be analysing, creative and testing and we did some work with a big shoe retailer and they've got loads of nice pretty pictures of shoes on boxes or shoes with a white background, but actually people with shoes all sell more shoes and it's a fact.
Speaker 1:And that's interesting, and do you think that that I know it's a bit of speculation is that because actually there's something around the aspiration and the image that people hope is projected by wearing a certain shoe, as opposed to the actual shoe itself. So you put it on, you know whatever it is a great looking model of a certain type, and actually that's why they buy the shoe, rather than what the shoe looks like.
Speaker 2:It might be because of the aspiration of well, if I put those shoes on, I suddenly become more handsome, maybe, but I think what we do is, yes, we help people test creatives, and we do this with creative ads and other things as well, and creative, I think, what's the right way to say it as well, and creative, I think, has what's the right way to say.
Speaker 2:I think, over time, the investment in brand and creative over the last few years has probably taken a hit, because everybody's just become and I, as someone who runs a data analytics business, I probably shouldn't say this, but everybody spent too much time focusing on the data and analytics rather than necessarily focusing on the branding, and I suppose it's like how can you stop that scroll? All right, the other thing I think is massively under invested in and undervalued now is copywriting right, and then everybody's saying chat gpt is going to replace all of this and I'm like, I'm not sure it's like it has a place in speeding up ideation and automation of some things, but you still can't be an amazing copywriter for a headline. That's a skill that chat DPTs are never going to have.
Speaker 1:Yeah, it's interesting. I mean, one of the businesses I'm involved with we're looking at that a lot. Yes, of course, you can speed things up, but then the market changes very quickly as well, and so the AI models will tend to predict based on what they already know and what's worked in the past. And then you run the risk in any competitive market that you just start to look very, very samey well, you end up in this echo chamber where it's you're just and actually um.
Speaker 2:Should you just do something completely different and I'll happily share. The best performing marketing campaign we do for a marketing AI platform is direct mail because it stands out yeah, we send really provocative postcards to people in this, so we send cfos of brands. Uh, a bright yellow postcard in the post say is how much money marketing is spending on google keeping you up at night?
Speaker 2:um, when then on the back there's a chamomile tea bag and it's a lot of day, drink the tea, uh, or should be, find out how much money you should be spending on google, and some of that's about building a brand and getting our brand, our right yellow brand, in front of people, and then some of it.
Speaker 2:So we do get direct response from that um, and again, same to sort of agency owners is um, we send them. Everyone, put your feet up and have a brew on us whilst we produce your report, so your team don't have to, because there's a huge amount of time and effort spent crunching numbers when actually the technology exists now. But yeah, and actually I was telling this story, henry, to someone else and then someone said well, do you know who, over the last 10 years, has been one of the biggest B2B direct mailers, who spent the most on Google? Right, right, there you go. So the minute someone registers a business, they get sent in the post a voucher for Google ads and they, they send lots of these companies stuff in the post because you know it's probably going to get there, it's probably going to get open, whereas if you send an email now, how many emails have we all deleted today without even reading them, unconscious of time.
Speaker 1:We should probably start to wrap up a little bit, but I did want to bring it back into this kind of question about the extent to which you work with a lot of the listeners to this podcast, the kind of consumer insight departments and companies who tend to be doing a lot of work around brand perceptions, brand trackers, maybe testing creative, all that type of thing. Do you connect with them much or is that a gap in the market?
Speaker 2:I think it's a bit of a gap. What's interesting, henry, though, and sort of taking into account, I think, the brands who do this well, get all the right people in the same room and they understand right, what data have you got, what data and how can we combine all of this to end up with a really good picture. I think it's especially in a lot of the sort of D to C type brands that are online only. I don't know if it's a gap or they're just not doing enough of this stuff, I think, because it's so easy to set up a Google account and a Facebook account and just load in your credit card and start spending money and something starts happening, and they're maybe not getting enough data to understand.
Speaker 2:Well, who is our actual consumer? Where are they hanging out? What are they doing? How should we be talking to them? How's this going to help shape our product or our overall strategy? And, yeah, I don't know if enough of that is going on. We did a big one the other day, and it was around someone trying to validate the actual attribution model, and I said, well, the easiest way to do is actually just ask the customers, right, where did they first hear about us. All right, well, just let's start asking them.
Speaker 1:What did?
Speaker 2:they first find out about. Well, we can't do that and I'm like, well, why not Right we asking them? Well, we can't do that and I'm like, well, why not Right, we could have like an exit. We could have a thank you exit survey. Maybe give them in the in the thank you email, we could give them an incentive to complete the survey and we can go. Where do you first find out about us? Do you follow us on this? And actually we got quite a significant amount of data.
Speaker 2:But I just just I think a lot of people are too sales and revenue focused and are maybe missing the opportunity to understand their customers more, and I also think people don't spend enough time thinking about customer lifetime value. Um, and understanding, right, because a really interesting, well, hopefully you'll find it interesting. But, um, and this isn't necessarily considered purchase but, say, beauty products. We had a big beauty retailer. We were trying to help grow new customers, also the most valuable first new customers. So we looked at all their transactions.
Speaker 2:We started going, well, who are the most valuable customers you've got? And is there a common first basket? Is there a group of products that they put in their basket? Because if you're buying beauty stuff. You normally buy multiple bits and are there two or three products that everybody buys in that first basket, in the most valuable customers. And there was. There was two or three products, certain types of things, and then we could start saying to the brand well, actually, if we know people that buy those and they're a new customer and they're going to be worth more over X period of time, we can potentially afford to lose money on that first transaction Because we know they're going to be more valuable.
Speaker 1:But, john, going back to what we talked about earlier, I mean, in some ways it shows what a gap in the market there is, because actually a lot of these consumer insight departments are doing a lot of that work brands, in terms of understanding consumer segments, identifying that 7% of a brand's target audience account for whatever the percentage is of the profit. They know vast amounts about them, but somehow it seems to get lost in the flow to your point about not putting everybody together in the same room in terms of actually, then in terms of then the marketing targeting.
Speaker 2:Yeah, I think there's two things here. One is, if you can because Facebook and Google say they do all this audience stuff but actually if you've got genuine, proper audience research and customer research, you should definitely be feeding that into the digital teams to help them optimize better. That is a complete no-brainer. But also, I think some of it comes down to how marketing teams are structured and who owns what. And years ago I think companies like Habitat would have a head of marketing but then they'd have a head of digital or e-commerce. I think companies like Habitat would have a head of marketing but then they'd have a head of digital or e-commerce and they both had one had the budget, but one had the KPIs and that's like how is that going to work?
Speaker 2:And I think sometimes the way businesses are structured haven't evolved as the way people actually attract their customers and sign up their businesses evolved. So, yeah, I think there is definitely an opportunity for the insights teams to work closer with the digital teams, taking those more sort of survey-based and insight-based research and feeding it into the optimization. And I suppose if people have got that on scale, arguably you could automate a lot of that and you could, you could, you could really, um, yeah, that could be sort of yeah, give it some sort of superpower.
Speaker 1:I do think there's an opportunity there conscious of time, might jump onto a quick fire round. So have you had any mentors and if so, what do they teach you?
Speaker 2:I've had lots of uh, lots of different mentors, um, I suppose the I guess this is a bit of an easy one, but I suppose the first one, the first and most impactful one, will probably be my father. He, he used to phone me, uh, at my desk and my because this is back in the day when you had a desk phone in my office in london in my first job, when I'd failed not to get into the air force, um, and he'd phone me at quarter to eight every morning and if I wasn't at my desk there'd be a message going how on earth do you expect to achieve anything if you're not the first person in the office?
Speaker 2:I, I love that as a yeah and if I wasn't my desk right, what are you going to do today? Right, and every morning he'd phone me at quarter to eight, so and that just sort of set a sort of. My father was a before he was an airline pilot. The family business was butchers and he just used to always get up early and work really hard. So, yeah, so I suppose that was more about the sort of work ethic and I often think sometimes and I know it's meant to be quick fire but a lot of this stuff you see on the internet now is how you can do this and that bit, and everybody forgets the bit where you've actually got to get up early and work hard. Well, a lot of people miss that bit out fair enough.
Speaker 1:Um now, what would your partner say? Are your best and your worst characteristics?
Speaker 2:this is probably the same one. It's going often that's the case, isn't it? Um over optimism. So I'm the sort of person if the house was on fire I'll be like, oh well, we were looking for a new house anyway. Or people's warm, uh. So yeah, I'm. I am very, very energetic and optimistic, but also sometimes too much got it.
Speaker 1:If you could be the ce of any business, which one and why?
Speaker 2:I'd love to, I'd like, I like the sort of classic answers would be the sort of apples and of the world. So I would have said Tesla, but I've heard a lot of mixed things. I am a Tesla product fan, but I don't know if I am now an elon musk fan.
Speaker 1:So, yeah, let's go with apple gotcha. And uh, have you got any favorites or most impactful books or recent books? Or it doesn't have to be, but could just be a piece of media uh, yeah.
Speaker 2:So I suppose this is very timely, and this was I would have said this even if whatrows haven't died this week, uh. But there's a book, um, called the last, uh, the last mile, I think, by kevin sinfield. That's all about his fundraising over the last few years. That's a real interesting book about friendship and about determination. Um, even if you're not into rugby league or or anything, it's really worth a read, just about human determination. So, yeah, I'd rate that um. And then there's another one called the multipliers, which I recently read, which I now encourage all my leadership team to read. Uh, and that's all about how good leaders create other good leaders. Uh, so, from a sort of business point of view, the multipliers- is a great book.
Speaker 1:Thank you. A couple of very good recommendations, but for listeners in america, rugby league is a sport primarily played up one motorway.
Speaker 2:In the uk and australia yeah, yeah, it's like, it's like, it's like american football without helmets pretty much.
Speaker 1:So, uh, joel, thank you so much. Um, really, really. I know we rambled a bit, but actually you helped clarify quite a few areas that I think can be quite complicated for people. So thank you very much and a pleasure doing it.
Speaker 2:No, thank you for having me on, henry. Really really enjoyed our time.
Speaker 1:Well, I hope you enjoyed the chat with John. It certainly helped my understanding of a few different areas around our funnel marketing, attribution, people trying to cycle to Australia, that type of thing. No, really it was great. And next time round we're switching on to another viewpoint and lens on the insights, marketing and data sector with Vera Chin, who runs global corporate strategy and insights for Warner Brothers Discovery. So that's the home of some of the world's best-known brands, such as DC Discovery. So that's the home of some of the world's best known brands, such as DC, eurosport, discovery Channel, hbo. The list goes on. Vera has a fascinating job. It's a great conversation, so please tune in. Thanks again to John, to Insight Platforms for their support and to you for listening. See you next time.
