#028: Attribution!

Attribution is like a box of chocolates. It can be really expensive, or it can be really cheap. It requires making a lot of decisions as to how you actually want to consume it. It may leave you feeling ill! Join the guys for a 45-minute walk across the attribution landscape. And back. And back again. Because mama always said you shouldn’t stop at the last click.

Episode Transcript

The following is a straight-up machine translation. It has not been human-reviewed or human-corrected. However, we did replace the original transcription, produced in 2017, with an updated one produced using OpenAI’s WhisperX in 2025, which, trust us, is much, much better than the original. Still, we apologize on behalf of the machines for any text that winds up being incorrect, nonsensical, or offensive. We have asked the machine to do better, but it simply responds with, “I’m sorry, Dave. I’m afraid I can’t do that.”

00:00:03.71 [Announcer]: Welcome to the Digital Analytics Power Hour. Three analytics pros and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com forward slash analytics hour. And now, the Digital Analytics Power Hour.

00:00:26.00 [Michael Helbling]: Hi everyone, welcome to the Digital Analytics Power Hour. This is episode 28. To what do you attribute your stance on attribution? Are you 50% influenced by last click, but closed by linear? And how many contact points have you had within attribution? And were they all the same channel? Well, we’re testing response rates from a new channel on attribution. It is this episode of the Digital Analytics Power Hour. And if you’re playing at home, anytime someone says the word incremental, or incrementality, you take a shot. And who better to join me on this customer journey, but my two expert shot taking pals. It’s Jim Cain, double CEO of napkin and Babbage Systems. Does that mean he takes two shots every time? I think so, yeah.

00:01:07.97 [Jim Cain]: Okay. Unless you’re driving on the way to work, I’m listening to this.

00:01:11.14 [Michael Helbling]: on the way home.

00:01:13.44 [Tim Wilson]: Unless you have a fancy electric car, is that going to be allowed once an electric car, or not electric, but self-driving cars?

00:01:21.65 [Michael Helbling]: Self-driving cars. Drink and be driven.

00:01:24.67 [Tim Wilson]: I think I’ve been ambivalent on self-driving cars until this very moment.

00:01:29.84 [Michael Helbling]: And that, of course, is our good friend and senior partner at Analytics Demystified, Tim Wilson.

00:01:37.21 [Tim Wilson]: If you’re going to listen to podcasts, listen responsibly.

00:01:40.72 [Michael Helbling]: OK, that’s probably going to get us some negative phone calls from some concerned citizens.

00:01:46.55 [Tim Wilson]: All right, gentlemen. The temperance sub-movement of the analytics community is going to have a problem.

00:01:51.33 [Michael Helbling]: That’s right. They’re a powerful lobby.

00:01:55.54 [Jim Cain]: I’ve never seen them in the lobby. They must be in the room.

00:01:58.20 [Tim Wilson]: That’s right. I think there’s, I’ve saw some lady with an axe coming through the lobby bar. Hotel, maybe.

00:02:05.95 [Michael Helbling]: Guys, attribution, where do we want to start? I think I will start with my story because it’s the most important one. No, I’m just kidding. What do companies mean when they say they’re going to do attribution and what could they possibly hope to gain? I think that’s a great place to start and then we’ll go into how the terminology is confusing and everyone’s screwing it up.

00:02:25.78 [Tim Wilson]: Well, so I mean, I think what they want to gain is the classic, I know I’m wasting 50% of my spend, but I don’t know which 50%. So I think it is a little bit, there’s definitely a degree of a unicorn of, I just know it’s this thing. Attributions like a 360 degree view of the customer, that there is this lofty aspirational, And maybe it’s based on who is saying they want to do multi-channel, multi-touch attribution. I think generally the higher up you go in the organization, the more aspirational unicorn seeking it is and it is this, oh, I’m going to understand the specific combination of touch points. And I am going to be able to ratchet back my paid media and get step function improved results, which I think is a not where any journey, you know, starting down the path looking for doing attribution is actually going to wind up, but I’m a little bit of a skeptic on the, on the topic. So maybe, maybe Jim can weigh in.

00:03:30.78 [Jim Cain]: And I shall. So there’s a real concern with this particular podcast that we could end up where we were probably about a year ago when we did big data, which is us going, what the shit is big data for like 45 minutes and not coming up with the clear definition.

00:03:45.61 [Michael Helbling]: I feel like it was a reasoned and intelligent discussion where we’re asking, what is big data?

00:03:51.14 [Jim Cain]: Thank you, editing.

00:03:54.04 [Michael Helbling]: I think this is much more definitely down to like two things.

00:03:56.76 [Jim Cain]: I think for this one, we can very clearly bound at least what attribution means to us because I do think that there is some boardroom speak around attribution and omnichannel and customer journey. And the emphasis that I want to make mostly today is around attribution as a digital only endeavor.

00:04:14.16 [Tim Wilson]: Digital only on visit, click through visits or is it digital including impressions?

00:04:20.34 [Jim Cain]: I would include impressions, but not for certain kinds of attribution. But this is the conversation I want to have. So if we want to talk about the likelihood that someone saw a commercial on CNN and then went to the website and clicked the branded ad word and filled out your email and six months later, buy your shit.

00:04:33.98 [Michael Helbling]: I mean, if I sent them four catalogs this year and their purchase happened four days after they got the last one and then they used paid search, it’s hard to do digital only there.

00:04:45.10 [Jim Cain]: I’m not saying it’s bullshit, but I’m saying that when you look at it, I mean, if we surveyed 100 companies right now and asked them how many of them had a legitimate, actively using an attribution practice, like not answer one question or get one insight, but we actively use this to do better, to kind of go back to what we were talking about with Gary in the last podcast about, you know, how do we have data be the spoonful of sugar? Like it’s not a separate silo, it’s part of everything we do. With attribution as an on purpose thing that’s part of everything we do, trying to do omni-channel is a couple of years away from now, in my opinion. But if we bound it exclusively to things with zeros and ones in them that live on the web or that can be measured in the way that we do them, we’ve got a very, very good story to tell today. And I think if we were to talk about all of the ways that people can come up with attribution plans and what it could mean to different people, we’re going to have a big data chat, which isn’t bad. But I think that there’s all three of us have been involved in some pretty significant wins around tying pan-session, multi-campaign data together for digital.

00:05:44.80 [Tim Wilson]: Well, I mean, I think I think to me, definitionally, fundamentally, there are two ways attribution get used and they are very, very different and they get used. It kind of gets thrown out. But to me, that the same thing happens for reports and analysis. And there is attribution when somebody says, oh, you’re just using last click attribution. Both of these approaches say agree on what last click attribution is. But then the next step is one way is call it kind of the Google free available to everyone who’s got Google attribution where you pick how you want to attribute and it’s limited to last click and it is you pick a model and you pick if you want it to be linear you could do first click you could make your custom models and models is a tricky tricky tricky word there because you’re choosing how you want to assign value you want to choose how you’re attributing value and And that’s kind of one way that attribution gets used. The other way is what more the tier one enterprise attribution management platform sell, which is more, we are going to try to figure out how much value the first click and the second click and the impression that happened after that. And God, they tell a story that is incredible. incredibly vague and hard to understand how it works of how they’re going to tie in TV and print. But they can tie in, you know, they’re kind of of the anything you can pixel plus wave their hands and we’ll do this other stuff. But we’re going to kind of try to model it. And that to me is really, really hard to conceptually I get it in kind of a hermetically sealed universe where you only have maybe a million different unique paths to purchase.

00:07:27.57 [Michael Helbling]: I don’t mean to like totally disagree with you, Jim, because in a lot of respects digital attribution is about as much as could theoretically be accomplished most of the time, unless you really simplify contact points and unify them in a central data warehouse or something. And then you could in theory see sequence of points of contact up to a certain level of granularity across every touch point, both online and offline for a known customer. But at the same time, like to what you were saying, Tim. I think honestly where I struggle the most with this is that I feel like so much importance is placed on attribution where it is seen as something really core and strategic, but at best could probably be used for tactical purposes to align spend to better channels and sort of see where you can get more bang for your buck in an individual digital marketing channel.

00:08:23.76 [Tim Wilson]: Let me try this out. Jim, Jim’s made a point, you know, prior, not, not today or not, not since we’ve been talking, but made a point about kind of the starting point is, are you really nailing last click and Jim, I may be putting words in your mouth, but I think it’s actually a hell of a place to start because all your direct traffic. Now, if you’re CNN.com, sure, you’re getting a crap ton of direct traffic. But for your direct and organic search and stuff that doesn’t click through on a campaign tag channel, do you believe that there was or was not influence that you provided through marketing stuff that you did in the past? So if you’ve got a bunch of referral traffic and that’s because it’s not tagged with campaign tracking, you know, that’s a problem. So there is kind of a, you know, is your last click attribution really robust? Do you really, really believe in that? Because as soon as you go to any other step beyond the last click, you’ve just gone to an order of magnitude more complexity. If you haven’t, don’t have your last click nailed, right? I mean, I don’t know, Jim, I may have just extrapolated from comments you’ve made in the past. and unfairly put words in your mouth.

00:09:34.94 [Jim Cain]: No, you’re dead right, Tim. Part of the reason I’m so bullish on this topic and attribution in general is that I really firmly believe it should be a first-class item for anybody who’s spending on the web, along with understanding your e-commerce data, you should be understanding your attribution data at the same level of detail. So I’m huge on the topic. And I think that you can’t get started unless you’ve done your blocking and tackling and you’ve got your fundamentals. And we’ve never talked to a company that really wants to get into attribution before they could answer the point that Tim made. What’s your level of confidence that your current campaigns are being tracked properly? So if we want to do just pass the post or last click attribution, will I look at it and go, yes, there it is. And now I want to do something more sophisticated with it. And the answer is always no. But the interesting thing is the conversations around figuring out what someone’s campaign naming conventions should be. Like, again, to use Google as the example, but you know, medium source campaign and now at the top level channel, figuring out what business rules apply that are as business relevant as possible, tend to help create some of the initial parameters for doing interesting attribution. But absolutely, you need to have your house in order before you start doing the interesting stuff. And most people don’t think to do that.

00:10:45.46 [Michael Helbling]: I like that point, and I agree with it. Here’s a question for you, Jim. Can people effectively use attribution tools that come inside of products like Google Analytics, or do they need to buy a standalone solution?

00:10:58.50 [Jim Cain]: So I think the attribution functionality in Google Analytics is to attribution what Google Analytics was to measurement three or four years ago. It’s very, very good. But once you hit a certain level of sophistication, you can graduate past it. And to be clear, I wouldn’t say that about the Google Analytics product in general. But their attribution functionality is very good for an in-tool suite. You will hit a point where you need to graduate once your practice gets to a certain level. But the thing that I actually like about it, per Tim’s statement earlier, about being able to pick your own model is that most people who buy an attribution suite go from, I don’t trust, last click, to, I spend a shitload of money on algorithmically driven attribution that’s giving me a report, and I guess we’ll just go with it. But I don’t really trust that either. I just paid more for it. It’s a black box. Yeah.

00:11:42.60 [Michael Helbling]: Yeah, that’s always the fundamental problem is yeah, you either go from a very simple model that you don’t believe is the real world because it’s too simplistic or an algorithmic black box that you also don’t trust because you don’t fully understand it.

00:11:56.27 [Tim Wilson]: Well, it’s tricky and Google’s is very I mean the the one where you where you pick the. pick the assignment of value. I think that’s tremendously useful. I don’t think, I don’t want to call it attribution because I think it’s more very, very healthy visibility understanding into for the visits who click through for the ones who came to the site multiple times through multiple, one or multiple channels before they converted. So we’ve kind of narrowed it down to a subset of visitors, a subset of people, or a subset of devices, a subset of cookies, and we rely on having a meaningful conversion event on the site, right? So that’s kind of narrowed us down to a limited set of what we’re looking at, but it’s still super, super useful because I’ve had clients where they’re getting almost no traffic, they’re getting 85, 90% of the traffic is first time visit and they want to dive into those reports and I’m like, knock yourself out. You know what? Pick every single flippin’ model and you’re playing at the margins, right? You have bigger issues than trying to figure out if it was paid search, then email, then display. Oh, you’re not running any display. So I think as an exercise of saying, let’s just get a handle of how many visitors are actually coming to our site multiple times before they convert. Oh, look, it’s a tiny, tiny, tiny percent. And that’s not the case for many, many sites. There are some that are set up to have recurring business, and that business may be coming from multiple channels. So I think it gives a ton of visibility, but I don’t know that it’s necessarily attribution. Or if you say, pick two or three different models, compare the models and see, is it a material difference? Moere often than not, it’s not a material difference because you’re playing at the margins.

00:13:40.10 [Jim Cain]: Yeah. Sorry, I should say it’s a very good point that if you don’t have enough traffic or you don’t spend enough money on digital, looking at attribution is about as smart as doing multivariate testing. You must be this tall to ride criteria for some of this advanced data stuff.

00:13:56.13 [Michael Helbling]: Tim, you said something which I want to disagree with, which is it is absolutely an attribution model, even if you go in and pick it. I mean, that was the old clear sailing many years ago. Just go in and be like, I think 15% is how much

00:14:10.12 [Tim Wilson]: Well, you’re, you’re choosing how to attribute. You’re choosing how you want to.

00:14:15.35 [Michael Helbling]: Yeah. So you’re making an attribution model. Last click is an attribution model. It’s just, you know, a well understood poor model. You know, and if I tend, if I say, okay, I’m going to give 35% to search and. 10% to add you know display and 15% to email and I’m just making arbitrary choices that is also an attribution model it’s just not necessarily one.

00:14:37.08 [Tim Wilson]: That’s the distinction I’m trying to make between visual IQ or atometry or clear sailing or pick your poison.

00:14:44.36 [Michael Helbling]: Well, Google has an algorithmic capability now. So the point being is that that’s that’s the tipping point is like whether or not you’re going to let into your point, Jim, you’re going to have enough data to pour through that to create an algorithm that’s smart enough to actually build you an attribution model that works.

00:15:01.05 [Tim Wilson]: But when you’re doing that, when you are picking how you want to assign value, definitional. So if you are attributing value on last click, and maybe this is where I just get a little hung up on this semantics, that maybe it is attribution is me picking how much value I want to give to each touch point. that I’m measuring. By the way, we’re ignoring in many cases, we’re ignoring impressions, we’re ignoring TV, we’re ignoring the legacy of the brand and the fact that, you know, gee, I grew up with my parents use Tide and I’m using Tide, like I, which in many, many cases I would argue for many, many brands is there’s not shit you’re going to do with marketing other than making, you know, maybe putting some tiny little incremental offer, but you’re going to have a hard time overcoming whatever they grew up with, Colgate or Crest.

00:15:53.16 [Michael Helbling]: Yeah, I mean, obviously, but now we’re straying outside of digital and therefore moving out of the arena that Jim has set for us.

00:15:59.99 [Tim Wilson]: But I think it’s part of the challenge with attribution.

00:16:02.89 [Jim Cain]: I’m fine to go out of it and I’m fine to talk about brand loyalty and I’m fine to talk about Omni Channel and print and what have you, but there’s a reason that people tend to do AB testing or any kind of testing initiative at the landing page level. It’s because focusing on the landing page and the visitor intent gives you the most possible control and the highest likelihood of getting a win. The nice thing about taking digital, and I actually think using impression data for attribution is bullshit too, and that’s a little controversial, but I like using click-driven multi-visit multi-channel attribution campaigns because it gives you the most possible control, the most accurate data, and the ability to do what you talk about a lot, Tim, which is sit back and say, in looking at this linear model, I believe that if I reweight branded paid search to have a higher percentage at the third touch, I will see something that will allow me to optimize my paid search spend. So I really look at it at the same way that you would look at a testing tool and less like a customer journey tool, which some people do.

00:17:02.30 [Tim Wilson]: I think was it was a Gary who was making the comment. I’ve definitely made the have evolved from the five years ago saying everyone’s in digital and TV is therefore going away. And now I’ve kind of shifted to say, wow, throw a bunch of money at TV. And even if you’re getting time shifting and all these other things, it’s happening. I was thinking actually today before we were recording of some of the brands that, you know, Carbonite, which I’m no longer a customer, or Squarespace, right? That’s kind of a joke amongst podcasters. Everybody knows about Squarespace. I’ve recommended Squarespace to people, although I’ve never been a customer and that has been through podcast impressions. And I get that there’s some level of I want to do attribution, eliminate to the channels that I can control, but I do look at, and I’ve got clients that are spending a huge chunk of what they’re spending is on TV. And that can be frustrating as the digital guy when I have zero visibility and they probably don’t even have a flippant call to action to point to the site, you know, at the end of the ad. But I kind of think, yeah, TV’s working because in, in to a sense, in a sense they have, logically, it’s working. but also they may be doing mix modeling, which is this big kind of macro level attribution, which they can’t control as much. So I get very conflicted of saying, well, we’re going to be the three blind guys with an elephant, and we’re only going to worry about the rough surface that we’re feeling, and that’s what we’re going to operate in. When there are a ton of other factors in play that are all part of what marketing’s doing and part of what the brand’s doing, it just feels disingenuous to call it, pretend that we are accurately attributing when there’s a bunch of factors we’re just ignoring.

00:18:53.79 [Michael Helbling]: It’s almost like 50% of our spend, we just don’t even know where it goes.

00:18:58.80 [Jim Cain]: You should write that down.

00:19:00.90 [Michael Helbling]: I’m going to make a quote out of that. Tim, when Jim, both of you, and maybe this is more just to the world, my rant is one of the things I see is wrong with the pursuit of attribution out there is that it’s sort of like this pursuit that will magically create a Frankenstein of perfect marketing and perfect advertising because the computer told us what to do when it’s The advertisers responsibility to have a plan for how to get customers and attribution tells you how successful that plan is and how to tweak that plan. And I think that’s the tipping point where I see people go off the rails. They’re like attribution will answer every question and fill this lifeless body with. movement and create a marketing juggernaut for our company where in reality you have to sit down and answer some strategic questions like is it more important for us to go get new customers in this phase of our business or do we need to look for ways to retain loyal customers and how does that impact how we go into the marketplace and spend our advertising dollars.

00:20:03.40 [Tim Wilson]: But that’s an issue with digital analytics overall, right? Saying I want the data just to answer the question. But I think to that point, and this is, let me tie everybody together here, one of Jim’s clients, who I happen to be friends with, who happens to be a listener and he knows who he is, has said, hey, paid search. You know what? The best way for us to figure out this attribution thing is to actually sit down and do an experiment with our paid search and to control for the DMAs where we’re running this, And I remember hearing Jim Novo actually speak at any metrics where all of a sudden I became a little bit more of a believer in attribution because he was kind of pointing out, and I think it’s aligned with your point, that you have to have that plan. You have to say, this is what I think works. Now, what can I do to actually test if that works or not? Digital is actually trickier with TV and print. You’re kind of built in, and certainly with email and print mail, direct mail, you can control and say, I’m going to do an experiment. I am going to do this heavy spin here, and I’m not going to do it in this demographically similar DMA. It’s a little tougher with just my website, but I can do it with display. I can do it with paid search. But it does argue for saying, how are we setting this up? Where can we hold out a group that from exposure to our advertising and then compare that group and see what happens? And that particular customers a CPG company that, you know, did that with sat with Google and basically said, yeah, you know what, this, this works. But oh, by the way, we’re also, we have diminishing returns. Like we shouldn’t spend necessarily more than we’re spending there, which is another challenge with attribution. So what if you understand the perfect world of where things work, it doesn’t mean that you turn around and shift all your spend from one channel to another, and no longer necessarily works. So that’s kind of another challenge with attribution. So what if you have a perfect understanding of exactly how things are working today? As soon as you say, oh, I’m going to pull out a radio, or I’m going to pull out of paid search, then the model is kind of by definition going to change. That was two or three rants in one.

00:22:17.30 [Jim Cain]: That was a response. It was a valid point. And I don’t want you guys to think that I’m trying to spike anything that’s not digital attribution or omni-channel or what have you. And we’ve done some neat work in the past. You guys have seen the routine I did before on looking at the impact of losing a catalog list on on-site conversion. It was part of one of my presentations last year. And we’ve also done some work before at the metro level on the impact of A focused radio and tv on a metro and comparing it to a dma for a metro that is of the same size and i’m sure you guys have done similar stuff before like there’s there’s neat stuff you can do there it’s just my real concern is that for anybody listening to this they’re going to get off the phone and go shit that’s really hard or that’s really enterprise i need to buy really expensive software and there are. so many wins that you can generate with a well-thought-out initial approach to getting into digital-only attribution and then realizing that as the technology gets a little bit better and some of the other marketing tools get a little bit stronger, the business is going to come to you with the rest of those channels. I don’t think we’re far from seeing TV given the amount of net-connected TVs. We’re not far from seeing TV as part of a measurable, non-BS digital marketing mix. And if you have a solid foundation and you can add that in, as opposed to kind of wait for a shootless Joe to show up, you know, you can get wins now. And I think that’s my, my really big, big point for today.

00:23:35.81 [Tim Wilson]: Which are, so name rattle off what those, and I’m not, I’m not challenging them literally saying rattle off what some of those wins are. I mean, give the hell the catalog examples, not a bad one.

00:23:47.64 [Jim Cain]: Uh, give a couple of the examples of kind of the specific channel once or just did anybody can get them digital only wins.

00:23:53.49 [Tim Wilson]: Anybody can get themselves. I’m running GA-free and have meaningful traffic to my site, and I’m spending on paid search, I’m sending emails, and maybe I’m doing some display.

00:24:08.18 [Jim Cain]: A couple of examples. The second that you start, and again, the criteria here is that all of your campaigns are being properly tracked at the ideally medium source campaign level. So you’ve got some detail. And in a perfect world, you’ve built channels for your marketing that are business relevant and ideally tied to how finance pays for shit. If you had those in place and you were arbitrarily a retailer because we work with lots of them, you would quickly realize that you have two kinds of affiliates. And one of them are what we call coupon stuffers and the other ones are what we call content marketers coupon stuffers and you can do this analysis and i’ve probably talked about it before you will find that certain affiliates one of them rhymes with schmitt tell me not uh… you’ll find that someone makes it all the way through to the shopping cart which is you know much higher likelihood of buying than your average visitor they see the coupon code box they leave the site they come back to retell me not so not only did you lose or did your abandonment numbers get kind of artificially inflated but you had to give up x percent of the sale to someone when you had a visitor who’s going to buy anyway just to make that tangible but you would you would have you would update your and we’ll go a little google analytics specific here so you would update your oh no this will work for adobe too you’d have a channel that you’ve defined as affiliate and you’re saying

00:25:24.48 [Tim Wilson]: In that case, you would see a bunch of different channels, direct, paid search, organic, whatever. And then right before the conversion, well, in Google’s case, you’d be resetting a new visit. You’d see them coming in through affiliate. Is that what you’re seeing when you look at those models and say, wow, affiliate, if I do a linear model, 50% is going to affiliate. But when I look at it, it’s always the second channel, or am I misstating how you’d see that?

00:25:49.03 [Jim Cain]: You see enough at the channel level to start to again start asking some detailed questions. So even in free Google, if things are set up properly, you can say, I just want to see the attribution at the affiliate ID level. So you can go down to a level of detail that lets you see which of your affiliates are highly weighted towards being the closer. and which affiliates are highly weighted towards not just being a closer but very, very crappy influencers in terms of bringing people in earlier, then you can build them into custom segments and isolate them and kind of do some hypothesis validation. But being able to find a subset of your affiliates that are not making you anywhere near as much money as you thought is a pretty straightforward win with any free set up properly attribution practice. Another thing is starting to see the impact of email and really optimizing when you send and how you send your email.

00:26:36.65 [Michael Helbling]: No, I was just going to mention, especially around affiliates, depending on the look back window, you can even be paying multiple affiliates at that point. It’s like you almost need some sort of system to manage the tags so that that doesn’t happen.

00:26:52.20 [Jim Cain]: Why called? You guys remember Convertro? Another like called AOL attribution or something? Convertro was my first exposure to attribution because the founder of Convertro was the founder of an online retailer I worked with a long time ago. I had a customer who was getting all the bills from and this particular retailer was doing way too much cost per acquisition marketing. He was just getting hammered. with his bills every month. And it was more like, you know, I know that 80% of my marketing dollars wasted. I reached out to Convertro and I said, I’d like you to build parameters into your tool that are tied to the cookie window and the payout percentage for each of the marketing partners that this particular company has. Because then their return on ad spend models would actually be tied to the bill and not to the opinion of Convertro. And they’re like, that’s a really cool idea. That’s impossible. Like no one does that. But again, that’s part of why my approach is always around, how do I understand what happens when I spend a nickel? How do I optimize that nickel? It’s all about revenue generation. If you’d replaced me with someone who is a brand marketer, they would say everything I just said was bullshit because you’re thinking about customer journey, you’re thinking about customer loyalty, you’re thinking about long-term engagement and branding. But my response would be, the model that I’ve just described, which you can do again in free tools, you could ask the hypothesis How much does it cost to generate a first-time purchaser? And what is the optimal marketing mix for a first-time purchaser? And how is that different than the cost per acquisition and marketing mix for a returning purchaser? You could do that right now. And that’s big kid stuff.

00:28:31.34 [Tim Wilson]: But how are you doing that? I mean, take this medium right here, right? I’m doing podcast advertising. I can attribute sales if I provide an offer code, right? Every podcast says go to www.mysite.com and enter the offer code power hour to attribute from that advertising.

00:28:50.47 [Michael Helbling]: That’s trying to make actually Tim would be more brand corrected, make the offer code analytics hour. Just keep going.

00:28:57.53 [Tim Wilson]: Yeah, well, that’s one of these. If you look at our style guide, I’ll check the style guide as soon as we have a sponsor, because I think that’s still a challenge, right? Because if you’re, if you’re, you’re wind up definitionally saying we’re only, we’re going to skew to the channels that we can measure the click through. And I, I mean, the funnel is dead. Essentially, I get that. But at the same time, there is the concept of I got to hear about a brand 17 times. Take Squarespace, take Carbonite, take, oh my God, audible.com. If you’re listening to podcasts, I don’t know about display. I mean, I’m trying to think what who is just completely carpet bombed me with display ads. And it’s nothing’s coming to mind. Only retargeting is. So when you’re not in a direct response world, when you are saying we have a new product or we are a new brand and we need to get that out there and we are going to have multiple exposures before people say, huh. You know, I’ve gotten enough, I’ve seen pre-roll video, I’ve seen homepage takeovers, and it’s just kind of worked its way into my psyche a little bit that now, six months, a year down the road when I’m looking to build a website or to find a way to listen to Audible audio books or find a cloud-based backup solution for my company. or looking for, you know, Citrix has done a lot in that area for whatever the, what, go to meeting when plugging all these non, they’re not sponsors. It is, it is hard for me to wrap my head around where the biggest impact where I now have them top of mind, if and when I have a need that I’m going to go and check them out as an option. And I think that’s a huge challenge with trying to do attribution. If you’re just looking at it, digital clicks.

00:30:44.65 [Jim Cain]: No, the only reason that you would look at impressions as part of an attribution initiative is if you’re paying for impression. And that’s part of why display companies make so much damn money. And they provide, frankly, not as much value as they generate for themselves, in my opinion.

00:30:58.99 [Tim Wilson]: But that’s the purpose of it. There are TV impressions. There are podcast impressions. There are radio impressions. One of them is way more than display.

00:31:08.43 [Michael Helbling]: This is one of the fundamental problems people also have, which is we set all channels equal to each other, and they all have to be measured on a direct response. I can’t believe I’m the one defending display advertising here. But channels don’t mean the same things. They don’t actually, you don’t start out trying to accomplish the same goals. And this is, I think, where people just sort of get dumb around analytics and attribution. And they just have to think about the fact that, like, Fundamentally, I want customers to do something different when they come face to face with this method of advertising that I’m putting out there. When I’m advertising on paid search, I’m looking to intercept intent because people are there looking to do some action or find something. And so I’m going to advertise in front of that in a way that’s going to get me to where I need to go.

00:31:59.12 [Jim Cain]: Whereas display is much more the old model of I want the world to know who I am and I’m gonna pay money to try to build that awareness But because people are still dumb enough to pay for impression Google doesn’t charge for impressions of adwords if you buy a billboard on the side of the highway You don’t pay for the number of fucking cars that drove by it. It’s it’s don’t disagree with you an impression is to generate a click and And the purpose of a click is to generate a conversion.

00:32:29.21 [Tim Wilson]: No, no, no. I disagree with that a lot.

00:32:34.32 [Michael Helbling]: Yeah. Well, on display, like you shouldn’t be trying to get a click because it doesn’t work. Everybody can see that. We’ve tested it a lot. It doesn’t, you know, like response rates from display are like ridiculous. Yeah. Yeah, they’re like less than a half a percent or worse, probably maybe less than a tenth of a percent, like banner blindness, blah, blah, blah. There’s a whole other podcast about display and now ad blocking. But that’s why impressions do matter because it’s the build of those impressions and what propensity it creates to put you into a channel that you can then convert in, that you’re actually trying to capture to understand, at least in some tactical fashion, And you can actually measure this, right? You can do blackout tests with display, just the same as you do with other marketing channels. And this is another whole topic, which is measuring that stuff in digital takes you away from what you’re really trying to figure out, which is, did you have any impact at all or incremental impact as it pertains to customers or the value of those customers? And that’s where a lot of people are missing the boat.

00:33:43.41 [Tim Wilson]: And let me throw beyond just display ads even if they’re rich rich media or not I mean I think there’s a there’s a reason that video is getting a lot of hype take video pre-roll You know there is a level of needing to educate people that this is a legitimate concern I remember when carbonite was kind of pushing hard and saying, your stuff should be in the cloud, right? You’ve got Dropbox and Google Drive and lots of other things. And hey, my 14 year old said, yeah, sure, wipe my whole laptop. Everything is in the cloud now, right? That’s not the world we were living in three or four years ago. And there is a value of kind of impression based stuff, be it video, I think video more so than display of saying we have to A lot of times, companies are saying we have a product. We’re different. This is new. You need to understand that you have a problem. There’s a four P’s, you know, marketing 101 aspect of where, yeah, you’re just trying to get people to understand there’s this other thing they could buy. They could, they could be excited about a pumpkin flavored drink at the holidays. And it’s really, really hard to tie to say, we’re just going to measure that based on the click. So I guess I’m echoing what you’re saying, help me.

00:34:53.94 [Jim Cain]: So two to one. I have to change my mind then.

00:34:56.23 [Tim Wilson]: Yeah, that’s right.

00:34:58.39 [Jim Cain]: The jury has it. Although I’ll tell you one thing.

00:35:01.34 [Tim Wilson]: Oh, go ahead, Jim.

00:35:02.04 [Jim Cain]: I was going to say, I would love to. And we’ve said this before, I would really like to have someone on the show who could give a strong defense of display and fresh data. I shit on it all the time. I have a bad experience with people who are trying to justify their paycheck, selling these kinds of things. I know it’s important. It just seems like the majority of the measurement is BS. I would love to have someone come in and tell me what time it is.

00:35:26.18 [Tim Wilson]: I’m actually kind of with you on that as well. I mean, just the ad blindness, spanner ads do sort of drive me a little berserk. But I will also argue that ad words, ads are, there’s no way in hell those are going to catch your attention unless you’re looking at them. You’re not going to get messaging around. Those are kind of keyword driven and okay, that’s a whole, whole other topic to to go down.

00:35:49.73 [Michael Helbling]: Yeah. So I third that, although a lot of times in talking to those folks, I find they’re on a train to Houston and I’m on a train to Chicago.

00:36:00.34 [Tim Wilson]: We don’t want to, we don’t want to be, we don’t want to shit all over again.

00:36:03.64 [Michael Helbling]: No, no, no. It’s, it’s not that I’m going to the right place or they’re not going to the right place. It’s that, you know, they have a very, never mind. Let’s get a guest. Let’s, we’ll find one for our listeners.

00:36:16.05 [Jim Cain]: That was 80% of a world-class metaphor. I was following along.

00:36:19.67 [Michael Helbling]: The other thing that I’m really curious to see what happens that allows a lot of people to start to do something that’s a little more sophisticated than what they can do today with the tools is that both Adobe Analytics and Google Analytics are starting to introduce the concept of cohort analysis more into their tool sets in a more fundamental way. And the timing aspect of cohort analysis from a channel perspective is going to be very helpful in answering some very simple this-then-that type questions. And I’m really excited to see how that emerges as a way for people to get sort of up that first step of attribution or understanding how channels interact with each other. Because again, if you think about it from a customer journey perspective, it’s not about creating a magical path, right? Which is sort of sometimes the way that people talk about it. Alright, let’s get to takeaways. Did we even say anything smart enough to take away? What do you guys think?

00:37:17.21 [Jim Cain]: Nope, Jim. How come the one topic I seem to know something about is the one you say nope for? What the hell, man? This is a fun one. And again, the big thing for me, and I narrow cast us from like minute two, and I know it’s kind of the holy hand grenade, but this is just such a cool topic, and it seems to be one of those ones like big data. You know some of the other things we’ve talked about that are broad where people feel that they can’t get started like I almost felt like we were getting into the shades of web analytics is easy hard debates of 2005 No, there’s no can to open Tim Wilson, which is I’ll take a couple a couple of episodes Google search and you’ll get back to us and football and Tim Wilson So you guys are like, it’s super sophisticated. It’s all these channels and you need to take these things into consideration. And I couldn’t agree with you more. I just feel that if you were to treat it more as a maturity process, which is not a phrase I’m a huge fan of, but something where you can walk before you can run, there are some very, very sophisticated things you can do quickly with Purely Digital that will grease the skids for you to really win with some of the omni-channel, especially when the data starts to get unlocked at the clickstream.

00:38:38.70 [Tim Wilson]: I’ll do the cheating where maybe throw in something that’s a little bit of a new concept on my rap. I don’t disagree. I think the ideal of attribution is absolutely fine to chase. If that is in the service of saying, we’re going to understand how our marketing works, which to a point Michael made maybe halfway through means that we’re going to sit down and think, how do we think it works? The more we can start with if we have the assumption or the belief, which I think is just never going to work out that we’re just going to feed all this into a machine and it’s going to come back and tell us what the right formula is, it’s just doomed to fail. And that would be kind of my one. one piece of advice for anyone who has an executive saying, we’re going to chase attribution or we’ve got a vendor in here telling us this is what it’s going to do. That is just not going to work. It’s got to start with saying, hey, do we have a really, really good sense on our last click? Now, how do we think when we’re spending money on these other things, and I would say digital or not, how do we think those are playing in? And how confident are we? And if we’re super confident, you know what? Let’s just roll with it. And if we’re not confident, then let’s figure out a way to validate whether it’s working that way or not. And that is going to lead us down an attribution path that will give us incremental value along the way. I just think it can’t be looked at as a panacea and I think that’s, it’s interesting. We spent many, went down many rabbit holes of saying, what is the scope? Is it only digital? Is it only click? Is it, oh my God, it can’t be that because there’s all this other stuff impacting it. And then we swung all the way back to saying, yeah, but if we just look at the stuff that we can measure now today with confidence, we’ll still get some value. So it’s a messy, messy space and maybe that’s my big takeaway is and maybe it’s because I’ve ranted about it now for a while. Anyone who thinks that that’s attribution is like the question and the answer is just insanely naive and don’t go ahead, don’t head down that path.

00:40:42.58 [Michael Helbling]: Yeah, no, I mean, I think I really like that too, which is, well, the first question to answer is, how confident are you in your last click? And is that being captured today really as effectively as it could be? I certainly think we are, you know, we’re stumbling around the issue, but have some pretty good spidey sense about it. And I think it’ll be really interesting to get an adult in the room with us and do another show maybe about this topic again. because I think that’ll help our listeners out quite a bit. But I really like what you guys had to say about it. I think my favorite book on advertising and marketing I’ve ever read was Scientific Advertising by Claude Hopkins. And what I wouldn’t give to go back to such a simpler time where attribution was pretty straightforward. We live in a crazy multifaceted world, but the principles still remain the same. We just have to disassembled the sweater of the internet that we’ve created. How is that for a metaphor?

00:41:40.16 [Tim Wilson]: It’s not, it’s not Jim Cain-like. Let me just, anyway.

00:41:47.45 [Jim Cain]: You nailed it. And it reminded me of actually my favorite marketing book, which is the original Moensters Compendium by Gary Gygax, 1984. Nice.

00:41:59.11 [Michael Helbling]: No, but guys, this is good and it’s a good start on a topic that confuses many and frankly gets a lot of people in a lot of hot water because you can take you down some weird trails and it can cost a bunch of money to do and may not give you results you want. So if you’re listening and you’ve got comments or you just want to jump in and tell us, wow, you guys really tried to bunt at this one or you hit it over the wall, it’s a home run or I can’t believe you made me drink in my Tesla. Post us a message on Facebook or Twitter or on the measure slack or hell.

00:42:34.65 [Tim Wilson]: Leave us leave us a review on iTunes, you know, we you know every once in a while for that.

00:42:39.77 [Michael Helbling]: It is 2016. We got to get that going again. Okay. Thank you for listening. We’re going to wrap up and we’d love to hear from you for my two co-hosts, Jim Cain and Tim Wilson. Keep analyzing.

00:42:58.36 [Announcer]: Thanks for listening and don’t forget to join the conversation on Facebook or Twitter. We welcome your comments and questions. Facebook.com forward slash analytics hour or at analytics hour on Twitter. We’re not recording right now are we?

00:43:20.70 [Jim Cain]: Yes we are.

00:43:22.16 [Tim Wilson]: I just made Texas State Fair an additive.

00:43:28.06 [Tim Wilson]: You’re super compelling, Jim Cain.

00:43:31.10 [Jim Cain]: What happened to your 10-minute walk there, buddy?

00:43:33.94 [Tim Wilson]: And Jim is going to be like, ew. Do it. But drinking were open, too.

00:43:42.92 [Jim Cain]: Napkyn isn’t hiring anyone and doesn’t want customers.

00:43:47.32 [Tim Wilson]: And models is a tricky, tricky, tricky word there.

00:43:51.01 [Jim Cain]: Well, you set me up to say something vulgar that I won’t say.

00:43:54.33 [Tim Wilson]: Oh look, it’s a tiny, tiny, tiny percent. Everybody’s getting a chance to talk. Wow, I lasted literally less than a minute.

00:44:05.12 [Jim Cain]: We have Uber for sled dogs.

00:44:07.82 [Tim Wilson]: Rock, flag, and attribution.

Leave a Reply



This site uses Akismet to reduce spam. Learn how your comment data is processed.

Have an Idea for an Upcoming Episode?

Recent Episodes

#274: Real Talk About Synthetic Data with Winston Li

#274: Real Talk About Synthetic Data with Winston Li

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_274_-_Real_Talk_About_Synthetic_Data_with_Winston_Li.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares