Why your B2B attribution model is probably flawed

Haydn Woods-Williams: 0:26

Welcome to the latest episode of the Marketers of the Universe podcast. I'm Hayden Woods Williams, a digital marketing team lead here at Blue Digital. With me is, as always, Tom Innis, our content marketing manager, and also Michele Raffaeli, who is one of our paid media managers. Today we are delving into the dark, murky forest that is marketing attribution. I know I've had plenty of conversations with different people about whether or not this is a good thing or not. I think the only thing we can be sure is that there are ghouls and monsters in that forest, and we're going to try and dig them out and work out what on earth to do with them. So without any further ado, let's get on with the podcast.

Tom Inniss: 1:32

Great. Thank you, Hayden. Lovely to be back. So to get us all on the same page, have working definitions to start with. How would you define marketing attribution in a simple, uh, easy to follow and short way?

Haydn Woods-Williams: 1:50

It is basically just assigning responsibility for driving revenue to a certain channel or activity down to its simplest level. That is what it is. We can go into more detail if we want to and look at the different models that are available. Um, but that may be kind of defeating your question. I don't know if Michelle's got a different answer for that, but that is how I'd kind of approach it.

Michele Raffaelli: 2:15

No, he asked for a simple description or for a simple um definition, and that's it.

Tom Inniss: 2:23

There we go. This is how you got the customer. Yeah. Yeah. Okay. Pretty much. So that's a very simple definition and it sounds pretty easy. So why is it considered so difficult?

Haydn Woods-Williams: 2:37

Because it's flawed as hell. Um, I think I think if you look at the different models, you you have single-touch attribution, like first touch and last touch, multi-touch attribution, um, like linear or time decay, and then you have data-driven attribution as well. They're all really complex and really important and really useful in their own ways, and they're all good at different things. But the problem is people use them without the wider context and without um taking the time to really understand the limitations of those models. I think one way that I can't remember who it was, but the one way I was explained attribution to and why it's so challenging is to compare it to raising a child, which is quite relevant to myself at the moment. If that child goes on to excel in dancing or video gaming or cycling or whatever it is, if you try and pinpoint the key moments in time that have driven the child to that expertise, you can't then make sweeping statements on them. So if there's a really important moment in that child's life where they're at a dance class and they get awarded top of the class, if that's in marketing, we're going, cool. So to become a professional dancer, we need the child to, or we need the prospect to become top of the class in their dance class. That is what turns into a sale or a dancer in this situation. Um I think you have to kind of change your consideration and look at it as in there is a journey for the prospect, there's a journey for the child of discovery, curiosity, self-investment, self-discovery that leads that child and every child or prospect. This is getting weird now, uh, is different. And when we kind of look at it more holistically and we understand that um we can't attribute responsibility to single moments when something is purchased, you can start considering the wider context, you can start understanding there's hidden things that are telling people or that are influencing people. You look at it from B to B, you've got not just one customer, you've got five, ten, sometimes even twenty customers. Um, so that is why attribution is such a crazy thing, because we're trying to assign revenue and responsibility to like glasses of water in an ocean. I apologize for that.

Tom Inniss: 5:32

I accept it. Uh, I also admire your commitment to the metaphor. You really, really try to keep that going as long as possible.

Michele Raffaelli: 5:42

If you want, I can bring it to the next level. Go on, go ahead. Billy Elliott. If you think the journey of Billy Elliott, why become a dancer instead of a miner or a boxer? That's exactly the attribution model that's playing through the the child attribution, as Hayden was trying to do the example. So was it the first if you look at the single touch point, the first interaction would have been boxing or mining, because that was the first thing that uh he was exposed to, but then the last click, sorry for spoiling if you didn't watch the film, it became a professional dancer. So if you analyze this journey with the last click, you could make some assumptions. If you look it at the first click, you could make others, but if you use a more evolved instead of a single touch point, a multi-touch point um attribution model, a data-driven one, for example, or um a time decay one. So, what event during Billy Elliott's um journey made him convert and become a dancer? And that's actually how it works in absolutely completely.

Haydn Woods-Williams: 6:57

This I I wanted to be at Billy Elliott in the first five minutes of this podcast, but I completely agree with Michael because it it makes sense in that context. And if you're a marketer, what you need to be doing is not looking for that magical moment where Billy puts his shoes on. That's not the moment that you need to capture and recreate for every single one of your prospects. The thing that you need to recreate is the thing that's consistent throughout those things that um are maybe less specific, but still really important. Billy attended a dance class and he did that every Tuesday for a year. I don't know exactly what day Billy Elliott went to his dance classes, but we're gonna say Tuesday. Um, and it's exactly the same in in marketing. If you want to reach your customer and influence your customer, especially in B2B where sales tackles are long, you have to find those places that they are and be in those places with them, with valuable content over a long period of time. You build trust, you are then kind of top of mind when it comes to purchase. Um, and that is what's wrong with attribution, because we're trying to pinpoint those magical light switch moments, and most of the time they don't exist. And even more often than that, we can't measure them.

Tom Inniss: 8:20

So why do we do them then? And why do we try to go to such lengths to do them? Like those sort of complex attribution models like multi-touch. Do we need them? Or are they just vanity metrics to make marketers feel like they are having an impact?

Michele Raffaelli: 8:37

Don't call them vanity metrics. Are they though? No, they are not. No, no, no. Um we the I don't the only problem that we have is that we are limited by our technology at the moment. There will be a time when we have enough infrastructure to properly track those things. Now, the way the model works now is with big data. They need a lot of uh data from customer journeys, and then they create a model based on that. The reality is there are so many inputs during this journey that we can't track for legal reasons, for privacy, for um lacking technology, cross-device, IP, ad blockers, that they don't tell us exactly what happened. I did experience 10 years ago a company that was the first one was doing it. They made this model collecting all the possible data, and I saw it in action. Uh, it could work only for a brief time because then it was breaking all the and it could it consider so many inputs that it would break, or to analyze a proper model, it would take days. So that's why the technology is the limit we have now. But if you have enough data, you can understand what exposure in your journey actually made you take the decision to convert. I was looking for some case studies where a company managed to track the 47 steps in a journey, and based on that, you will apply the model that actually tells you what moment in that full journey convinced you to purchase something. I break down this example with the fridge. If today your fridge breaks down, how long will you stay without a fridge? 24 hours, 48? So breaking down the fridge, that's what ignites your journey in becoming a fridge customer. My fridge is broken down, I need to act immediately. How much time do I have to go out physically, offline, go in a store, get to look uh at fridges, or have the interaction with my neighbor that bought a fridge and it's so good brand you should try it. How much time do I have to go online watching influencer and packing a fridge and telling me about these differences? How much time do I have to go read blogs, see articles, get my information before I then go and click, buy online, buy in the store, but you make the conversion. So this journey as an attribution window, attribution window and attribution model are two different concepts, but they work together because the attribution window, so the time that I use to analyze your journey is what then I apply to my attribution model. So the fridge example, it's it's something that everyone can understand because it can happen to everyone. My journey is 24 maximum 48 hours. Now, if you apply the same journey into buying a car or a house, this journey can be 12 months long. And you can imagine how many interactions, how many touch points you have in the 12 months journey. When we go to B2B, is something very similar. The difference is instead of having one person or two deciding for the purchase, you need to assemble a team that make the decision. So this is what we need to understand in these uh models that we need to touch, interact with different types of people in different moments in the journey in order to then get the conversion that it's what we are looking for at the end.

Haydn Woods-Williams: 12:40

Completely. Yeah, I think I think you made some really valid points there, Mikelly. I think uh from from my perspective, and you you touched point this, is that most attribution models tend to be flawed in some way or the other, whether that's the model being flawed or whether that's the data that it's that's being run through it being flawed. And I think that's where people who are you or marketers who are using those models need to understand those nuances, understand those flaws when they're making the decision making. Um and instead of kind of picking out those exact moments, use larger groups of data or smaller case studies if if you want to as well, but more than just one, to notice what the the signposts are that are pushing people towards a purchase. What's consistent over your last 50 customers? What touch points are consistent? And when you look at it that way and you work with the wider team to understand it, it's it's very different. Uh, and I think Mikelli made such an important point about B2B as well, is you're not influencing one person, you're influencing five, 10, 15 people who all need to agree. If you think about the fridge, all those 15 people agree need to agree to buy that fridge. Hopefully, we're not selling fridge. Uh some people sell fridges at B2B.

Tom Inniss: 14:12

Um, somebody's buying fridges, at any rate. So he spoke, like we've spoken quite a bit about the flawed nature of these attribution models. Um, and I want to sort of touch on another topic, a favorite of Marketers of the Universe, which is that relationship between marketing and sales. How do you get the sales team to trust and use this data if we ourselves are saying it's flawed? I don't think it's flawed.

Haydn Woods-Williams: 14:39

Of course it's flawed.

Michele Raffaelli: 14:40

Of course it's flawed. You have limited data to analyze, but that's not a flaw. That's just the how things are in this moment.

Haydn Woods-Williams: 14:48

Okay. I think I think we're maybe talking about the same thing, but using different strengths of language. Because I think those limitations are the flaws in my in in that I'm referring to. And I think the problems that marketers gen tend to have, or actually sales teams, and I I was going to talk about this in a second, and C C level teams, is they don't understand the limitations or the flaws. So it's this model is gospel. Everything that we every decision we make has to be driven by this attribution model that we follow. And I think that is that is where it's flawed.

Michele Raffaelli: 15:26

And I agree with you. And to go back to um Tom's question, how do we use this information with a sales team? So every single team should be involved in this discussion. The creative team, the web, the web, um, the website team, uh, the developers, the marketing team. Because knowing what makes you take a decision is what gives you the power to improve. If from our um journey we see that a specific landing page is what triggers a lead, then that's what we want to improve and what we can use to test to sacrifice maybe other points that they didn't bring enough um volume back to generate that so we can try and improve that specific point. So the sales team should be aware of this because they should know what changed, let's say, the customer mind in order to buy. Was it the piece that the content writer wrote about the solution that we provide that made them realize oh, actually, this is something that is useful for me, and then go and buy? Is it the the piece talking about the prices? They saw that we are the cheapest, that's what's converting. So that's what the sale team should focus on on the sale. We are the cheapest. Um or instead of we provide solutions for these specific things, so that's how you share this data and you you you use them, you make them as leverage for your final objective.

Haydn Woods-Williams: 17:05

Completely, completely. And you you can't you said it um in the beginning, it really relies on everyone having an understanding of the model and the data and the limitations, um, and also kind of working in a feedback loop as well. When everyone understands the data you're looking at and what they should be taking from it, you can start having things like sales telling um marketing that actually 50% of the customers that they've been talking to have been or saw an ad on Reddit as an example. Um and the other way around, marketers can tell the teams, uh the sales teams, actually, we're seeing that I don't know, like BT uh shoving loads of interest in our marketing at the moment, it might be worth reaching out. But for those things to work and and for what Michele discussed as well to work, it really relies on collaboration. And we have to move away from that sales versus marketing. And that comes from the C level. Like if you're putting pressure on your sales and marketing teams with separate lead targets, you're effectively setting them up, or sorry, inquiry sales targets, whatever it is, you're setting them up to face off against each other. That's a sales lead, that's a marketing lead, that's a sales lead. There's no such thing as a fully sales lead, and there's no such thing as a fully marketing lead because marketing uh leads need some kind of sales conversation or sales touch point or customer service touch point to convert and have a good experience. And for sales, whether marketing has directly talked to them, they will have seen some kind of branding or been on a website or seen a LinkedIn page. Um, and having that understanding is really, really important. So you need to make sure that your marketing and your sales teams are fully understanding of everything that's happening in that attribution model. They need to be working collaboratively, not against each other. And lastly, they all need to be doing their bit. Um, and when I say their bit, it's those horrible little admin bits. And I talk to sales admin here because I know it's a hot topic with some of the sales teams I've talked about. Those little conversations you have that needs to be logged into the CRM are so important to understand touch points. And those are the ones that are missed. And if a marketing team hasn't connected up LinkedIn ads or Google Ads or Meta or whatever it is you're using, put all of that information in the same place and allow the attribution to see everything that it possibly can. And you'll be working well.

Tom Inniss: 19:53

So you touched on the CRM there, and I was gonna ask you about where should all of this data be? Be stored. Is it in a CRM or are there specific platforms that are better for this sort of collaborative work?

Haydn Woods-Williams: 20:06

Yeah, this is this is such a hard question, especially in this age of AI, because there are going or there are tools after tools after tools after tools that say they can do this. But ultimately, the at the most basic level, what you need is one place where all of your sales touch points, all of your marketing touch points, and all your data lives. And ultimately, that is a CRM like HubSpot or like Salesforce. And a lot of those platforms and tools have tools within them that can help you do this kind of thing. There are other tools that will help and can help, but I think if you're starting from a very basic place, which I think most companies are, you need to make sure your CRM is in is in a healthy state. You have solid processes, your sales and marketing teams are following those processes. And then you can start looking to invest in other tools.

Tom Inniss: 21:09

So this might touch upon the flawed element that at least you, Hayden, feel is appropriate to describe attribution. But how do you account for what's known as like the dark funnel elements of attribution? So all of those touch points where you can't easily track it, like word of mouth or like just discussions, or even like B2B events, for instance. How do you track all of those secret microphones?

Haydn Woods-Williams: 21:36

Like phone hacking. These things are really no, I'm j obviously I'm joking. I'm sorry. Who knows? Maybe, maybe Google are really doing that with your your echo, and that's that's feeding into everything that you uh you're doing. Conspiracy theories are like going in my head now, but I'll have to leave that. I'll I'll head to to Reddit forward splash. Well, rather podcast, yeah. Exactly, exactly. I mean, I mean if you are. Um stop getting distracted, Hayden. For me, uh, and I think for McKelley, I I reckon it will be something like similar. I think the big thing is is trust. Um, and and this is again where it's really challenging when C-level or sales puts a lot of pressure on marketing to deliver a certain thing, because brand is often hidden and that kind of growth is hidden. So you kind of have to trust that this is happening. You have to trust your team to be creating valuable, interesting, shareable moments that people take into their private areas, be that WhatsApp or real life or wherever it is. Um you then kind of need to make sure you're consistent with what you're doing because you know you you wouldn't expect it when everyone talks about going viral, but actually, consistency is what builds the dark funnel. If you're kind of always there in the conversation with timely relevant content, you are going to be discussed in those kind of areas and then find data that does suggest kind of causation. So if you're running or you've been running an organic social and a brand ad, and you've been creasing blogs recently over the last three months, and you've seen your brand searches go up, but actually your organic engagement is a bit rubbish and your paid engagement isn't great, and not that many people have come through those channels to the blog as an example. But you've seen a huge spike in brand searches, you kind of have to look at that and go, this is one of the only things we've been doing. There must be some kind of link there so that you're not then changing your strategy to go, you know, we've had an increase in brand searches, let's run some brand ads and just invest money into putting paid ads on on Google search and then stopping your organic social and your your other activity. Um, so that's kind of how I would see it. I I feel like I can't finish that part of the the conversation though without a an anecdote, not an anecdote, an analogy or a metaphor of some kind, because that's what this conversation is about. Like it is a you go a lot further, easier running a kilometer day for 90 days than you do if you try to run 90 kilometers in one day.

Tom Inniss: 24:35

I have nowhere to go with that, so thanks. Um we're just gonna move on. We've mentioned the C-suite and how they might be applying pressure to sales and marketing, as is their want and definitely their job. How do you use attribution data to report to those C-suite individuals and move a conversation beyond cost per lead and towards this sort of marketing influence revenue?

Michele Raffaelli: 24:58

Educated guesses. So we do have clear data that shows how people that have seen your brand at least once are 80% more likely to validate a proposal with your brand in it. So if they've never heard about you, they are unlikely to choose you. So the two separate steps of this journey that we need to do are um reach the final user that will benefit from our service or our product, and at the same time reach people in this decision-making place that will match our brand with the request they're going to receive.

Haydn Woods-Williams: 25:46

That's a really valid point, especially when we consider kind of like those hidden buyers. I think if we're looking at how to communicate upwards to the CEOs and the CTOs and the CFOs, there is a big fear, and there is a lot of kind of misinformation and fear-mongering chucked around on LinkedIn. Um, I think the biggest thing you can do is just make sure that there is a consistent conversation going on. It's a lot harder to report any numbers to a CEO or a CFO that you speak to once every three months. Whereas if you're speaking to them every week, they'll have a better understanding of what you're running, what you are trying to achieve, how you are trying to achieve those things. Um, so then when you're reporting to them, they understand the whole context of everything more. I think you still need to talk in their language, so you still need to talk um marketing fluence revenue cost per lead, where you can, and you have to use those attribution data models to try and feed into that. But ultimately, you by having that kind of regular conversation, you're building trust. Um, and especially important is building the understanding of the model of what you're doing marketing-wise, and of how it's kind of delivering results that they want that matches the business strategy. Great.

Tom Inniss: 27:17

Thank you. So, with time very tight, I'm gonna move us on to our final question. Given the shift away from third-party cookies, what should marketers be using right now to build a reliable attribution model? Or as reliable as it can be given the flaws we continually refer to? And Michaela, I'm gonna start with you.

Michele Raffaelli: 27:42

They should keep things simple, they should um allocate the right time for the channel to develop and deploy their potentials, um to keep changing strategy, it doesn't help understand what's working or not. So if you do your homework at the beginning and you have everything prepared, all your audience analyses, your creatives, your ad copies, the landing pages, feedback from the sales team, everything is prepared. You should trust your strategy and you should just go for it within the time that in average takes you to convert someone. Keep changing it every day or every week because you don't have the results you want, it won't help you to understand what's actually working.

Haydn Woods-Williams: 28:41

Yeah, I think the key symbol point is is really, really poignant. If you are not like before you even consider tooling, if you're not doing the basics right, that's the sales admin, the connecting channels app, the having a strategy, all of those things, then you've probably got a flawed or limited attribution model, regardless. Um, I think what I would say is use common sense as one thing, use collaboration, and as a starting point, ensure that you have a CRM that the teams have processes around that they are using um effectively, that all sales conversations are linked to, that as much of your marketing is linked to. Um, because that just allows you to ensure that when the time comes that you are mature enough to start implementing more complex tools or more complex attribution models, you have a catalogue of data that is consistent and reliable.

Tom Inniss: 29:47

Thank you very much, both of you. Hopefully, you have found this podcast useful. It's always nice to get Hayden and McKelly together, and I love it when they disagree. I love it more when they agree. But that is all we have time for today. Thank you so much for listening, and we hope you found some useful snippets in this podcast. Uh, we love that you've made it this far through your listen, and we love making this content for you. Uh, so if you could recommend this show to one friend that you think would enjoy listening, then we would be exceptionally grateful. If you didn't enjoy it, then give it to somebody you don't like because there's no point in both of us being disappointed. Um, we have been the Marketers of the Universe. My name is Tom Innis. It's been Hayden and McKelly talking about market attribution. And until the next one, we hope you have a wonderful day.