Statements of Intent

This week on Statements of Intent, we dive into how personalisation is really about supporting customers' desired intent. Our guest Tom Bailey, VP Product Analytics at Made With Intent, shares his perspective on moving beyond surface-level targeting to truly understand each individual's needs. Tom is an expert when it comes to data and isn’t afraid to ask the tough questions to find practical solutions.

Topics Covered:
  • The evolution of personalisation in eCommerce - from Boots loyalty cards to real-time intent signals
  • Why conversion rate alone paints an incomplete picture
  • Using natural language processing to interpret customer intent through on-site language
  • Creating cross-functional teams focused on the customer experience
Key Quotes:
  • "Personalisation is not necessarily always about giving something away monetarily. You just need to be suiting people based on what their needs are."
  • "A website is fundamentally made up of pieces of text. Underlying the website is text we can interpret to understand intent."
  • "We have to start listening beyond the noise of sales data and start tuning into the real people we're trying to serve."
  • "True loyalty comes from making your customers feel seen, heard, and understood as individuals."
Episode Chapters:
00:00 Introduction 01:10 Guest Introduction 06:20 Discussion on Personalisation in eCommerce 08:18 Statement of Intent: Personalisation Isn't Surfacing Content, It's Supporting Intent Effectively 14:09 Technicality of Inferring User Intent 22:17 Outro

Social Media:

Creators & Guests

David Mannheim
David is a big kid, a big Disney fan and a big geek. He founded User Conversion which was acquired by Brainlabs, is the author of The Person in Personalisation, and now the host of Statements of Intent. His mission is to help retailers care more for their customers by listening, being appropriate, being familiar and creating a relationship. He is doing that through his new start up, Made With Intent, a platform that helps retailers do just this by diligently understanding customer intent.
Tom Bailey
Tom Bailey, VP Product Analytics at Made With Intent, is an expert when it comes to data and isn’t afraid to ask the tough questions to find practical solutions.

What is Statements of Intent?

eCommerce has lost sight of the people at its heart—the customer. After working with the world’s biggest retailers for 15 years, David Mannheim (author of The Person in Personalisation, founder of Made With Intent) wants to change this.

Through solo shows and guest conversations, you’ll hear eCommerce leaders share how they intend to break the status quo gripping online retail. The lack of care. The short-sightedness. The sea of sameness.

Each 20-minute episode centres on the sharing of a statement of intent—A personal promise or guiding principle. It’s a vulnerable yet fun source of solidarity for weary commerce marketers who want to change things for the better.

~ This transcript is automatically generated so may contain some errors ~

So once you start to break it down into the fundamental of this page is. This page's purpose is to get you to the next page. This page's purpose is to get you to add to cart or to get you to understand more about the product or even when you start to break it down into parts of the page. So someone's looking at reviews, you need to support them in the reviews.

It's not all about using the single conversion rate as the, as the sort of, um, the measure of success of everything on the website.

Welcome to Statements of Intent. In this 20 minute episode, we're addressing how eCommerce has lost sight of the people at its very heart. You, the customer. It's a chat that's optimistic, it's casual, it's probably slightly ranty in places, but that's okay. But it's a place where I talk to senior eCommerce marketers.

And share their statement of how they're looking to change the status quo of eCommerce, adding more care, being more considerate to those very people that they're selling to - the customer. I'm your host, David Mannheim, the founder of Made With Intent. And we're going to jump right into it. Have fun

Hello, hello, hello, and welcome to Statements of Intent.

I'm here joined with guest speaker, Mr. Tom Bailey. Hey, Tom.

Hello, guest

speaker. I don't know what to introduce you as. You work, you know, you work here, so it feels like, this just feels like a normal Thursday, Google meet, right?

Well, I think because people know you as a speaker.

So, um,

yeah. Well, I'm the host and you're, well, yeah, you're the guest, right? So,

Someone brought up the dragon, your dragon's den award earlier. So it made me think of that.

I have it just here next to me. If you want me to, if you want to just grab it and, and show it off. It's very heavy. Very heavy. It's pretty good.

Uh, anyway. Get distracted. This is clearly going to happen throughout this chat. So, uh, Tom, I mean, rather than me introduce your accolades, credentials, and experience, why don't you tell everybody your accolades, credentials, and experience?

Mm hmm. Okay. So, I mean, my My experience is a bit less traditional, I suppose.

Like I went to uni, um, dropped out of that about halfway through. That was a waste of time, waste of money. Um, still semi paying some of it back now. So I didn't go down like a route where perhaps a lot of data people go down, but although So, um, working in data as I do, um, a lot of stuff is new all the time and it's just one of those industries where you have to learn things as you go.

So me having that kind of mindset towards just continually learning, developing, um, kind of ended up being able to get around not having like a formal education in science or whatever it might, whatever it might be that you're working across. So, so yeah, I, I just jumped around different jobs and I ended up working at Boots, uh, initially in the, um, In the online trading team.

So working on, um, aspects of the homepage originally, and that was my first exposure to, um, putting stuff in front of people that they're going to be interested in, and then. You quickly start to learn, to learn what people want to see. You start to see patterns and stuff, even in that really isolated sense of just putting something on the homepage.

Um, and I talk about this a lot, but like, I forget how big Boots is in terms of traffic, like the amount of eyeballs on that homepage is quite unbelievable. And when you compare it to like a lot of other things that you see, uh, and even, even stuff to today, um, nothing's quite as big as what Boots was and it's much bigger these days, but back then, um, you're putting stuff on.

On the website and within a few hours, you'll get tens of thousands or hundreds of thousands or millions of impressions on something. You get to learn stuff really, really quickly. And part of that means that we're always trying to change stuff up all the time. Every day there's a new thing at the top of the page at the bottom of the page and all that kind of stuff.

And you can kind of get away with it because you can learn from it quite quickly with good significance. But that was my first exposure to, um, putting some in front of people and then obviously tracking that information that. sort of logically moved me across into working in data more fundamentally and working more broadly across, um, uh, trading, marketing, um, and just all digital, really, um, interestingly, that was part of the personalization team.

This was back when, um, I don't know how much analytics was part of e commerce itself. I think e commerce was your marketing and your trading. That's even if marketing is incorporated, but. e commerce was just, well, you change images on a website, you merchandise, you, you do the inventory and the promotions.

And then those are the disciplines that have been reserved for, um, for boots. It'd be in store activity or it'd be, uh, email campaigns or something like that. So it's quite a novel idea to have like a dedicated analytics team just for digital. And, um, whilst I was there, um, the guys who worked with, they grew that team from, uh, one person at the time when I joined Scott Kroll, and then.

When I left, it was probably about eight people. Um, but in that time, um, that was, again, that was like a really big eye opener into how personalization works on a massive scale, um, somewhere like Boots having the advantage card in particular, um, and we'll probably get onto it anyway as to why I think that may have caused some conflation as to how personalization should be done.

Digitally now, because traditionally you send out things in the post and people get it every quarter and that's the sort of personalization that would have done and then translating that into a very different landscape of what we've got today, um, is part of the challenge. I think a lot of big companies are facing despite the maturity in market.

Um, and it leaves it up to like, in a lot of cases, smaller companies to be able to do more in that space, even though they don't have the maturity or the budget to even get there. So Yeah. Obviously we'll probably talk about that some more. So, um, yeah, and then that's, that's the bulk of my experience. So, um, everything I've done really is related to digital, um, moving into personalization, but pretty much all of that has been across the data analytics, um, areas.

And then obviously brought me to where I am today. I'm sure we'll talk about that more as well. We could

do so. Yeah. Currently our VP of analytics, for those of you who don't know. So Tom spent seven years at Boots, one year at Neomorganics, experience ranges, predominantly in e com. Like you started in trading, right?

Which is, well, I think we're the best e commerce people start because it's got that commercial angle.

It's really interesting that you talk about Boots and personalization. It almost feels as though they're semi pioneers within personalization. Their Boots loyalty card is known. You know, nationwide has been one of the best examples for personalization.

And I wonder if it got a stigma, if the term personalization got a stigma from that association with boots, because it was expressed as like a form of loyalty, almost. And I don't know where that's gone, but it certainly doesn't have that connotation. I don't feel like it has that connotation now. What do you think?

Yeah, no, it's a good question, actually. Because, yeah, I mean, you literally call it the loyalty card. And therefore, the expectation is that for that loyalty, you give us data, and then we give you some form of something back. Yeah,

same with Tesco, right?

Yeah, precisely. And there was the, um, what was the, the old, was it Sainsbury's one as well?

I don't think I found it. Nectar card, there's nectar card as well. The nectar card, yeah. Yeah, yeah. So it's interesting that they never really merged into anything like more of a global kind of loyalty card. But then I suppose it shows the value of the data is fundamentally what people want and companies don't want to share that.

And that is the trade off. But, um, but yeah, the advantage card, I think even now, like you're seeing the bits that pull back, even the amount of points that you get. Um, for a purchase, they used to get four points for a pound. I think it's three points for a pound. Obviously, everyone's in up in arms about that, but it's showing the direction that personalization needs to go into.

It's not about necessarily always, it doesn't always have to be giving something away monetarily wise. You just need to be suiting people, um, as to what they're. What their needs are in terms of personalization rather than literally just giving them cash back all the time. And I think being a company is starting to recognize that.

And if you start to pivot away from that, then the expectation will be, well, you need to do something better with your data, not just who do we give the most money away to and when do we give it to them? Um, it's doing something a bit smarter with it. So yeah, I think even at that really, really high level.

They're probably looking to do that as well. Well, that's

lovely. A little segue into your statement of intent. You know, we invite people on, you know this, right?

But we invite people on and talk about a guiding principle, like a personal promise, something that goes against the status quo almost of e commerce and yours is you have this belief that personalization is supporting a desired level of intent, a desired intent.

I mean, obviously. That's what we're trying to build. That's what we're trying to do as a business, right? But what do you mean by that? Why do you see personalization as the, the equation to intent or vice versa?

Well, I think it goes beyond that to be honest, because I think one of the things I always used to say was Um, whenever whenever you're looking at analyzing online because you get all you get to see all the touch points is not like when you go into into a store or in a lot of other places where you might gather data, you don't get to see literally every single touch point into the point where you can determine every touch point you want to see.

If you want to see what someone's doing every second, you can do that. But what that allows you to do is to enable you to break up every interaction that someone does Um, on site and anyone who has looked at digital data will know is that the aim of every partition that you're trying to optimize for is to get them to the next action and always trying to focus on, um, the conversion rate that you see at the end, um, is just incredibly watered down.

And you take a scenario of, um, a marketing team who are trying to optimize for conversion by where they land someone on a website. And then there's the, there's the, you know, the 10 hours, maybe not 10 hours, but say a couple of hours of browsing and behaviors that are happening between that in order for the landing page all the way to conversion that completely waters down the whole concept of I'm going to use conversion as my metric to be able to determine what marketing channel or what landing page it's going to be.

So once you start to break it down into the fundamental of this page is. This page's purpose is to get you to the next page. This page's purpose is to get you to add to cart or to get you to understand more about the product or even when you start to break it down into parts of the page. So someone's looking at reviews, you need to support them in the reviews.

It's not all about using the single conversion rate as the, as the sort of, um, the measure of success of everything on the website. And it's easy to do that. And obviously we can all do that. Um, but then what that, what that basically means is that all you're trying to do when you're analyzing a website is to try and figure out what the average thing works for each part of a website, whether it's down the page or across the pages or whatever it might be.

So the template that we set along for a PDP is because we think the average person kind of roughly needs that type of page in order to be able to eventually convert rather than this page has to be built specifically to dedicate itself to. The support of whatever's on the page, whether it's a PDP or a PLP, um, and then, and then, and then moving one step further from that, um, you then take that down to personalization.

So rather than dedicating the page to the average person, is it possible to break up the page and then test? Who it might work for based on what they're trying to do from that page. So I know we talked about it going from coming away from page templates and moving more towards perhaps stages or the mindset of the person.

So maybe that's kind of a long winded answer. But, um, but yeah, it goes all the way down to how you analyze it, even beyond personalization. So if you actually do apply to personalization as well. Um, then yeah, you come to where we are today, I suppose. You had a second part to that question as well.

Well, it almost, it almost feels inevitable, doesn't it?

Like one of the findings in, uh, you know, when I did all the research and writing for my book, just up there, available now from Amazon 1199, is, uh, is, uh, that people don't really know where to start within the realm of personalization. They do, like, new versus returning is a very popular one. Device space is a very popular one.

Traffic source is a very popular one. And I love your, your philosophical idea that those are good ways to start, but they're basically just proxies for what you believe is intent, right?

Yeah. Yeah. That's probably touching on things we've talked about in the past, but yeah, that's, that's a really good way of putting it.

Cause everything you do when you're trying to analyze, you're trying to, you're trying to in some way figure out. What someone's trying to do in some way. So, like you say, you split up mobile desktop and tablet. So why are we asking that in the first place? Other than that's what we always see. So let's look at it again is well, I see I can infer that if someone's on a mobile device, they're probably on the move.

And they're probably less likely to buy right now and they might come back later. So you've been inferring that and then you infer from desktop that it's a bit more consider the sitting down somewhere. They've got the tabs open, maybe doing some comparisons, whatever it might be, the more likely to convert at this stage.

And you see that obviously in the numbers. So as you say, we use them as proxies for the intent that we think someone might have. So we go, well, I want to know that on mobile, because I want to cater to the fact that they'll be on the move. And I want to know that on desktop because I want to cater to the fact that they're probably more likely to purchase.

Same with new and repeat. New, they don't know a lot. Repeat, they know a bit more. So we're going to be able to like skip them along a bit faster. But why, why boggle the mind with all those different iterations of how you can, let me think of all the dimensions. You can split that. Then you have the dimensions for marketing channel and then device type and then new and repeat.

Don't even get on to repeat multiple times or break it down to campaigns. And suddenly you go, well, Now you've got like a million iterations of how to personalize to all those people, but all the, all you were trying to do in the first place was trying to figure out. Some level of intent for each person at each point in time, so you can cater that because then you'll find all sorts of patterns across all the devices and across all the marketing channels.

You're ready to buy, you're ready to buy. Does it really matter for a mobile desktop at that point? Because the key point is that you're ready to buy, take into account all those different things. And obviously that's what. We focus on is to try and make sure that all that's distilled down into something you can use, um, rather than having to focus on all those, not that they aren't important for different reasons, but, but then we don't have to use it as a proxy for intent.


The attributes to determine something else, something more human, right? Like, like you say, a mobile device, when you, if you're personalizing for a mobile device, which is a silly thing to personalize for, but okay, let's go with it. If you're personalizing for a mobile device, the chances are that they're more browsing.

Uh, because you know that conversion rate is generally less because there's still a trust issue when it comes to mobile, unless of course there's Apple Pay and all these different inferences that you can assume. But perhaps a greater indicator of one's level of intent is their real time signals that they provide to you.

So one question that I have is, well I feel like I already know the answer so I'm like setting you up here, is why now? Why are we able? So why has nobody done this before? It's probably question A, and question B is why now?

There's a few, there's a few answers to it. One, I think, smartest answer. Smartest answer. There's a few answers. One, one comes down to it. One's obviously technology. I mean, at the end of the day, if you want to, if you want to be modeling all users interactions in real time, in milliseconds. Um, to a high degree of accuracy and be able to do something off the back of it.

You need, um, good cloud computing. You need it to be scalable. It needs to be accessible for, you know, people like us to be able to do that. We don't have to go out and hire a warehouse full of machines in order to be able to do this. I mean, like I think eBay had like two 50 grand machines when they started.

You haven't got to do that kind of thing. So we can scale up in that way. So that's. That's an obvious answer to that kind of thing. But I think it's more fundamental to perhaps the way businesses are moving one, because people become more expected to be able to drive through on site activity rather than just pulling the Google level or pulling the Facebook level.

So it just becomes more of an appetite for it because it's not simple. It's like, it's not an easy thing to do. So, well, if I can just put a little bit more money through Google or Facebook to drive more revenue, uh, and that's discounting the fact that. Um, e com generally has grown maybe up until around the late 2010s where it started to sort of like flatten off a bit.

Uh, and then obviously you had, uh, the, the COVID period where people perform better than as well, but everyone's feeling like, uh, everyone's seeing the writing on the wall a little bit when it comes to this. So everyone's kind of getting that appetite for, I'm going to have to figure out ways to get more out of my traffic.

So I think there's that as well. But then I think internally in the big companies that you would say a bit more mature. Um, the movement away from isolated. Um, departments of your e commerce, your marketing, your trading, um, you're the buying team, you're this, and then even breaking that down into even sub areas.

What I've seen and, um, and I know other people have seen that as well is those teams are already starting to merge across disciplines. So you start to see. Personalization teams, you start to see teams that are focused more on a conversion rate, for example, and that they will have influence in different areas of the business, utilizing those resources.

Now they need tools in order to be able to do that. So, you know, in order to do it, you need a tool, but you also need the people to be able to spread that across. You can't have, here's a great tool that works across all your platforms. But everybody works in silos, so no one's going to end up using something like that.

And maybe companies have tried it and it hasn't worked, but generally speaking, um, I think the teams are starting to formulate in such a way that they can utilize this kind of thing as well.

Yeah. People like organization moving more towards product functions that have. An outcome, you know, like acquisition, retention, conversions, a very popular set of, um, product teams are in an e commerce setup as opposed to, like you say, just marketing, trading, e com.

I think that that move has happened, is happening, is immaturity of the, the whole hierarchy, um, thing. Let's double down on the technical thing, if that's all right, because you, let's be honest, Tom, you have a technical brain. So how do you infer somebody's level of intent?

So how do you, how, how do we,

how do we infer someone's level of intent?

How do

we, how do I, yeah, how do we, without giving it all away, I suppose. Um, so to infer intent, um, you don't need as much as you think. Um, you perhaps would. Um, it's, um, already pretty obvious to everyone that if you were to take all your website data now, and you were to say, I want to trend conversion rate.

probability of conversion by a metric. If you just went page one, page two, people who reach page three, page four, page five, you will probably see your conversion rate go up with that number. So you can already get an idea of. Inferring, whether someone's like to convert is, um, is, is not impossible. Um, the difference obviously comes into all the different variabilities to that, because again, you're looking at an average.

So everybody gets to the fifth page. There's a wide spectrum of where people are at that point. It doesn't tell you what direction they're moving in. It doesn't tell you, it didn't tell you anything really, but it just gives you an average. And we've all done that kind of thing where we just go. Oh, let's use anybody above page five and we'll just assume they're ready to buy.

For example, and you go, okay, that's fine. And what about the person who's not ready to buy? Oh, we don't matter about them. Cause they're not, they're not the people we're interested in. So that's, that's, that's gives you a little eye opener into the kind of thing that. Um, will feed into, um, a model that will infer intent.

The other thing that we did, um, that we were mainly interested in was, um, with, with the emergence of language models now and how you can ingest, um, obviously, uh, language data to be able to, um, support in this kind of thing. A website is fundamentally made up of, um, pieces of text. Like we could all read, whether it's the URL, whether it's the thing that you clicked on, whether it's the code that sits behind the website, because we built it, we have to physically understand it underlying the whole website or in front of your eyes is text to some degree.

So, uh, we utilize that, um, really effectively and it's not, it's not the only thing we use. Obviously we use all the different metrics that you can probably imagine anyway, mobile devices, the types of pages you've been on, how long you've been on the website, all that kind of stuff. Um. But the magic really does come into the ability to interpret the language that's coming through.

Um, so the difference between whether, um, you followed this path exactly the same as someone else, but they've clicked green and you've clicked blue, that can then start to split you out into green as green means something different to blue, whether that's just because blue, we know blue is top selling, so therefore we know that's probably look better in the images and it will go down a different pathway rather to green or maybe even greens out of stock.

So those are the kind of things that you'll be able to pick up on that where. In any, anyone who's a data scientist will know that a very, very, um, granular form of, um, feature engineering. So if you were to say we're going to track absolutely everything you do with every single color involved, you'd need the widest table in the world and it would have to go into a model that we have to handle.

Almost an infinite number of features because you can't handle every color or every iteration of this every variation of that. So, um, language models now can just suck that up basically and start to work with it much better than we ever have been able to in the past. So, um, yeah, without going into too much more detail on that, I suppose that would be where.

Um, the difference has been for us, um, to be able to get to that level of detail to really do something special really.

Yeah, I love that. I remember hearing a story a while ago of, uh, I don't think I've ever told you, but the late great made. com. I remember when I was younger, 10 years younger than why I am now.

So, uh, 19, uh, that I always wanted to buy some furniture off made. com because I don't know, I always thought it was like super cool. You used to, I don't know, Swedish chic stuff. Um, and I remember hearing a story that they used to always track intent or You know, the last few years that they did, maybe it's the reason why they went under.

Um, but they based it on, uh, on page views. As like a proxy for intent. I never know how successful that was. I don't know any more than that, but it's just interesting how, you know, they went under what, how many years ago, three, maybe three years ago, pre COVID and now we're able to do what we're able to.

It's, he biasly says very impressive.

Um, Tom, that's what we've got time for. That's 20 minutes of you telling us basically your statement of intent, which is that personalization is supporting a desired. Intent. Intent is inevitable. Any other final words to sign us off on? Um, combine. Combine. You mean like a Fabrizio Romano, here we go, kind of ending


I don't know. Yeah, probably not. Since you just dropped that on me at the last second, I'm going to say, I won't say anything else. Dropped

all of this on you at the last second. Uh, all right. In which case, should we just say bye?

Yep. Good stuff. Bye. Bye everyone.

There we have it. Thank you so much for listening. Please do like, subscribe and share on whatever platform it is that you're listening to on today. This show comes from the team behind Made With Intent, the customer intent platform for retailers. If you are of course, interested in being more profitable, whilst being more personal.

And please feel free to check us out at madewithintent. ai. Thanks again for listening and joining us on our mission to change how eCommerce sees, measures, and treats their customers. I've been your host, David Mannheim. Have a great day.