#036: Attribution Revisited with Jim Novo

May 10, 2016

Back by popular demand: attribution! This time, we brought in an
adult on the subject: Jim Novo of The Drilling Down Project. A lot
of questions get tackled in this episode: Should “gut feel”
ever trump “the data?” Which is a better analogy for attribution:
PV=nRT or the distillation of bourbon? Will this podcast *ever*
have flawless audio quality? These questions and more definitively
answered. All in under 52 minutes.

Show Transcript

The following is a straight-up machine translation. It has not been human-reviewed or human-corrected. 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.”

[00:00:27] Hello everyone.

[00:00:28] Welcome to the digital analytics power hour. This is episode 36. You know after we did Episode 28 attribution we got letters lots and lots of letters and by letters I mean comments on our Facebook page and Twitter and the measure slack to the effect of Hey you need to go deeper on attribution and don’t go looking for all just trust me. This is what people want. But what could we possibly add to the tour de force that was Episode 28. Well we did say that we wanted to get an adult in on this conversation. Unfortunately we could not find an adult but we did find someone whose attribution roots run deep. Our guest this episode is none other than Jim Novo. You might know him as the owner of a thriving marketing analytics consultancy but prior to that he was the V.P. of Marketing at home shopping network leading the TV catalog and online marketing efforts. At a time when most people had never even heard of online he’s the author of drilling down turning customer data into profits with a spreadsheet. And the managing director of education for the digital analytics association serving as the lead instructor on development of the DA’s four core series teaching students how to create knowledge drive action using Intel digital data. He has a 25 year track record of helping marketers use data more effectively. And today he’s our guest. Welcome Jim.

[00:02:00] Novo that was quite a canned speech. Now you’re going to just read it like that.

[00:02:08] I mean I tried to put a little flair in that well here let’s you know I don’t get points for originality but other 2 cohosts certainly do.

[00:02:19] And they are none other than the fourth from the north north. The Ottawa in the senator the Maple Leaf are waiting there for a minute.

[00:02:29] KANE Well Norway Canada editor and then Maple Leaf Oh yeah. We don’t do that. That’s not done. Otherwise.

[00:02:41] So story knowledge of course our other host. He is the lightning bug. The blue jacket. The Crow.

[00:02:51] By God. Well. Oh yes. Professional sports mecca that is Columbus Ohio. All

[00:03:02] right let’s get into this topic. So Jim no. Thank you so much for joining us we’re excited to have you. We did an episode of an anti-abortion but we feel like we just scratched the surface. So we’re excited to go a little deeper on this episode and have you along ride. So let’s get into it guys. Where did we leave off last episode and Jim Nova.

[00:03:22] Maybe you can weigh in on where do people go wrong with attributes the most or where should they start. Because maybe that’s a good pickup from last time or start wherever you want to. I don’t care.

[00:03:36] You’re tired you’re going to take it all.

[00:03:40] I think I’d have to say what happens mostly is there isn’t really enough of a robust discussion around what people mean by attribution. You know how is it important. What decisions are we going to make after we figure out attribution. You know and there’s sort of like a software to do it right where you just kind of looking you know the tool interface and it says oh we’ve got you know this share and that chair and these campaigns contribute and so forth and and that’s fine. Right. But that’s that’s a pretty high level and you might be able to argue not so accurate view of what’s going on. So then do you want to dig deeper or you don’t want to dig deeper. I mean you know I just think generally there’s not enough upfront discussion sort of the way actually that I think Tim I think it did at the metrics you know you were talking about getting people to come up with a hypothesis right and say you know OK so if we get 10000 visitors from this campaign is this good.

[00:04:44] Oh yeah that’s really good. You know if we get 10 visitors from this campaign is that bad. Oh yeah that’s really bad. OK so right. And so you kind of go back and forth with them and you try to figure out what the metal is. And I think you know often in attribution it would be great have the same conversation you know. So you say to whoever is driving this thing. So let’s take you know a Chano or a campaign and you know what do you think that that campaign contributes and they come up with some number. And and you say to them well OK if it was 10 percent of that number would that be bad if it was 90 percent of that number would it be good. You know and you try to find someplace where you’re actually getting some kind of traction where a person wants to make a decision and then you know you talk about how we’re going to get there.

[00:05:35] I feel like that with attributes in a lot of times people think attributes in it’s one word. I understand what it means. Let’s go and do retribution then well it’s going to give us so many insights it’s going to find our our media mix and it’s going to be awesome. But they haven’t really put the clarity around what sorts of changes might they actually make.

[00:05:59] That’s what I was driving towards. Yeah okay. And so you know I don’t want to step up to you know. I mean it depends on what level you’re talking about this right obviously. But you know if you’re not really going to make any decisions based on this if you have essentially a marketing plan or a way the company operates or you know silos that don’t talk to each other whatever it is and nothing’s going to change based on the result of an attribution model test or just even what it says in the tool. Then why are you talking about this. Shouldn’t you spend your time and money doing something else right. I just come at it from sort of a business place right. You know everybody’s really busy doing all this different stuff and a lot of it has very high value and you know it has very high value right. You can prove it has very high value. So the question is if we’re going to go through this attribution thing are we just kind of kind of look at the tool and go Well that’s interesting you know there’s these touches in sequence and so forth and that’s pretty interesting and that’s good to know. And we should take that into account when we buy media and that’s kind of where it stops or are you actually going to go in and say what happens if we’re way over spending on paper click. And what are you going to do. You know could we put that budget somewhere else. What if we find out it’s only 50 percent as effective as we thought it was.

[00:07:29] That’s probably not going to happen in a lot of cases but it would be nice to find out upfront if people are going to actually make decisions about this stuff when they find out what the attribution actually looks like.

[00:07:42] I agree with a lot of that. But you know I think that any organization that’s not doing even some entry level not last touch attribution planning and they understand what it is and they understand how it applies to other things is doing themselves a disservice. You know when you when you start looking at the kinds of you know ad hoc questions or hypotheses that people are turning answering Google or Adobe or what have you. A lot of times you know you’ve got that that short list of things you go through or was it a landing page you know is there’s something wrong with the content was there something wrong with the last touch. Was there something wrong with the performance of the website and you inevitably in almost any piece of analysis you do now end up going What was the traffic mix that led to the visit that didn’t do the thing that I wanted and even a basic even a basic weighting model that gives you the ability to not think about things last touch gives you the ability to maybe find that aha moment. So I don’t I don’t necessarily think that you need to go into attribution full bore with the business improvement focus in mind. I think he can be in a position where you can get your toes far enough into the water that you can just evolve your overall quality of your overall measurement practice. I don’t think that’s the best way to do it but I think anybody that’s still using last click for everything you know they’re doing themselves a disservice.

[00:09:04] I don’t disagree with them. Maybe I came out too hard.

[00:09:10] I’m looking for a fight. This is the. Most Canadian fight for them.

[00:09:15] That was the part where I was saying look if you want to look at a tool and make those kind of decisions I guess what I would call soft decisions right. It’s really good information to have and you allocate proportionally. But let me try to see on the way. So when someone looks at a tool let’s say and they see you know this contribution that contribution right and they start to formulate these ideas and then they come out and they represent that specifically and exactly what’s going on. So we should therefore do this. That to me begins to sort of push the edge of that right because you’re kind of making a call it’s says you believe without any kind of really critical you know walk through of it.

[00:10:07] That’s exactly what’s happening. But is their value if you start there and say let’s just accept what’s just agreed for right now. We’re going to say that’s the way the world works. And then as an analyst or as a marketer then saying OK let’s now figure out how we can test the things about that sort of absolute that we most think is a little squishy you mean yeah. Is that the only way to come at it though and say let’s just let’s just lock in this is what it is. And now let’s be very opposed to saying we’ve locked in now let’s throw all the data at a very expensive tool that’s very expensive to implement and try to get all the answers.

[00:10:50] Is there an argument to say let’s just let’s agree what we think the world is and we know there are parts of it that are wrong. And now let’s find though the incremental ways and this we brought up on the last episode Episode 128 your email example from looking looking out 90 and 120 days and say Now let’s just do a little bit of a test in the control in just one area let’s see if we can make that one area better and make it really clear to everyone is that is that a acceptable strategy I guess.

[00:11:20] Absolutely. And what I’m trying to. I think what I’m trying to do is encourage people to start going in that direction. Right. You know if attribution is going to make a difference to the practice that they’re engaged in and they really want to spend a lot of time back to the whole resources thing are you really going to spend a lot of time and money and make a lot of decisions on this stuff or are you just going to kind of go with what everybody thinks is the appropriate mix and that’s fine. Right. So yeah so it’ll be great to say let’s cordon off this one little place. You know whatever the companies you know businesses in the mix and all that will determine what that might be. Let’s cordon off this little place and let’s do it right. Let’s either increase the spend or decrease the spend in a particular campaign or channel and see if it changes right and do it long enough so that you can actually see those changes. So that’s one way to approach it right. I would prefer like in the example you just cited the idea of well let’s find out exactly what we mean by attributing value or attributing some kind of action to a campaign right. So in the case of customer value for example the value of the attribution tends to change over time as you move out further and further in some customers or some types of visitors it will start to accelerate and really ramp up as they become a best customer let’s say. Right.

[00:12:53] Who’s visiting all the time or buying all the time or whatever it is that the desired outcome is on others. It’ll just be a one time occurrence and then nothing happens again. They don’t come back. They don’t buy again and so forth. So wouldn’t you want to actually sort of figure out the value side of it before you spend a whole bunch of time on trying to figure out the Chinle Macs and campaign mix and all that kind of stuff right. So that’s a different a different way you can go at it you know and so that requires a discussion about something like this just a short term. Are we OK with short term decisions or long term decisions right. Like if we if we have the ideal app attribution and it’s completely focused on very short term values is that what we want. Have you ever had that discussion type of thing you know. Or would we want to be more focused on sort of generating longer term value. We want to create customers who come back and visit us or buy again and again and again. Right. That’s really what we want to do and that’s probably a different attribution that’s a different media mix in all likelihood. So now you sort of have another level to talk about. Right. That really should be talked about somewhere or you kind of just go off on a path right.

[00:14:15] Well so does that have you seen it happen like when describing it that way I could see a discussion starting with we’re going to do retribution and as the white boarding is happening on that what comes up is we really should be looking at customer value customer lifetime value or a longer horizon for customer whether it’s the total value of a customer or whatever it is and what comes out of this we’re going to do attribution is holy crap we need to get smarter about how we’re measuring the value of customers and conversions. And you wind up kind of kicking the attribution can down the road because you actually are in the main to shift the organization which I don’t think can happen overnight or inside of a month to have more of a customer value as opposed to a you know what conversions happened today. You are those are those hand in hand that you actually wind up uncovering hey you’re trying to go after retribution but you haven’t. You’re not really thinking about your business the right way. And we need to think about that and get our get our systems instrumented and get our data integrated to the point that we can actually really measure that.

[00:15:30] Is that a possible outcome. I haven’t personally seen that but I could now see that playing out that way.

[00:15:35] Yeah. And that might be a good thing. Right or if you look at what they’re doing with cohort analysis right a lot of the SAS guys you know they’re saying OK so you know the people who signed up in January by the time we got to march this met this person you know dropped out and we’re kind of associating that with this promotional effort that we did. But then if I look at the cohort of people who joined in March we had a different kind of promotional effort going on and we had much better conversion from so or from freemium to paid or whatever it was right in that model. So if I’m going to attribute right if I’m going to start making selections it really makes a difference that these people and they don’t have to be you know buying products or you know they could be just visiting over time or interacting and posting even right just posting. When was the last time this person posted how many times did they post and blah blah blah. Right. That kind of stuff. So you kind of have to figure out what is it that you. What is it that you really want to measure in terms of value. And then you go back and start to attribute because that attribution mix is going to be different. If what you’re shooting for is long term value versus short term. So

[00:17:04] that’s interesting because I think last episode we we focused only on digital attributes so attribution of digital channels. I heard you say a couple things that I think I want to explore a little more. One is I think sometimes it’s the wrong way to go if measuring the customer is the right way to measure value and the other thing was attribution based on tactic as opposed to or or campaign type. Maybe I’m not saying that right but instead of channel in other words hey I’m going to run this type of campaign and I want to look at the attribution of that as opposed to looking at search versus e-mail versus display. So I don’t know. I’m not saying I’ve got an answer there. I just heard two things I thought were interesting. I wanted to see a little more.

[00:17:55] You weren’t saying that you thought measuring customer value wasn’t the right way to go.

[00:17:59] I think that I tend to think that because I most of my experience that this was in an environment where that was how we proved the value was by customers. And like to Jim’s point when those customers actually increased their sales velocity that was how we knew we were doing better.

[00:18:19] OK so if you’re saying you think customer customer value is the right way to go and that’s where know some context I don’t know.

[00:18:29] Jim you said you could use visit’s if that was the way that you are you know finding value that way. And I just you know I’m interested in that and I think probably a lot of listeners are. I think when you focus only on digital you cut out the customer concept because you weren’t tying you back to our customers being measured unless you’re doing a little bit more than what you get out of the analytics tools that would most of us have today.

[00:18:58] Does anybody even you know maybe maybe this is a question that’s like completely outside digital for now right. People just don’t look at valuation. They’re so focused on conversion that it doesn’t really matter. Do you get that sense when you win. I mean I certainly have worked with people that are basically in that camp as long as I can get you know this is the one sale whatever it is it’s over for me. I’m not interested in tracking that anymore and it’s pretty much over. And so if that’s the case then if that’s the way the company is structured that’s the way people think then you’ve got a certain kind of approach to attribution.

[00:19:42] So just to be clear value to you means lifetime value of a particular customer or does it mean applying appropriate valuation to all the key activities on your site. Like I would pay ten bucks for an email address or I would pay 50 bucks for someone to create a wish list. Or is it somewhere in the middle.

[00:20:04] No it’s you know whatever activity there is there is downstream value from that. Right. Because if you’re willing to pay someone ten dollars to give you an e-mail address then isn’t there residual value somewhere in the background from that activity. Otherwise you wouldn’t pay them ten dollars for an email address but totally.

[00:20:24] It’s just you know just like you’re kind of touching on earlier value is one of those nebulous terms in our industry right. And a lot of the stakeholders that I work with that have big omni channel especially strong kind of print catalog brands specifically means lifetime value right. So in two years a new customer will be worth 380 bucks to us.

[00:20:47] Yeah. I hesitated to use the phrase lifetime value because I think that that really trips people up and gets them into you know just like what we were talking about a few minutes ago. Now you’ve sort of mis directed everybody into like OK now we’ve got to figure out how to do a lifetime value equation. I mean I don’t really think that’s necessary. I I like to talk. I mean in certain businesses would be completely appropriate. But I mean for most people I would love them to measure the value of some activity that they’re paying out marketing dollars for longer than a week or two weeks. Right. Can’t we kick this out to like 90 days or something like that. Right. Were you just trying to again you know it’s a good example is a lot of the stuff that people are doing with cohorts if that’s the way that it should be thought of you know inside the tools and so forth then I think that’s a great demonstration. If you look at some of the types of models and tracking and a lot of the blogging that’s going on from these guys. You see the exact same kinds of curves and the same sort of value definitions. Right. I mean you can certainly define value however it fits into your business model right. It doesn’t necessarily have to create revenue.

[00:22:12] This is maybe a little bit of a tangent but when you’re looking at a 90 day horizon there’s this constant tension between what data. You can’t look at even though you have it. You know if you’re looking at 90 days and we’re in we’re in May right now.

[00:22:28] So that means that the stuff that happened in March and April I have data because I have the people click through I have the people converted. But when I’m doing my analysis of I’m doing a cohort analysis and I’m picking a 90 day window then I have to say I’m going to cut off my last action 90 days ago. I think I mean I struggle with this code analysis I’m like super excited about it and yet I have not managed to really dive in and make it work.

[00:22:59] And part of where I struggle if you start talking customer lifetime value and you say well you’re discounting stuff you don’t have to go out 10 years because you know a dollar from them. Yeah. But are you in this in this challenge. If you say we’ve got a say it’s considered purchase and we have a two month typical purchase lifecycle and then I want to put some more time on top of that to go a little farther than just the conversion. Like how do you manage that tension between if you got to 120 days you’re like holy crap you know that’s like four months. So you know we’re cutting off the data in January and that other stuff we can’t look at yet. How do you how do you kind of approach that or manage that. Or am I completely missing that there.

[00:23:44] That’s not the challenge.

[00:23:48] I mean it is a challenge. And so you kind of go back to the same question about how serious are we about attribution and how much are we willing to really care about. I think a lot of people you know spend probably too much time and money on it because they’re not organizationally set up to actually go through the process right. What’s going to happen in the end when you find out. I mean are you really going to change the mix in the media or if you come out with some kind of result that’s contrary to what people believe in the company. Are you going to change your behavior or not.

[00:24:28] So does that mean there’s a there’s a starting point where you actually have to get realistic about what sort of time horizon is right to look at. If we’re serious then we need to agree that it’s not a last click today conversion. It’s maybe not even a week it’s maybe not in a month. Let me show you just some historical data. This isn’t attributes and just let me show you that these things happen after initial activity and things don’t really taper off. Customer Value Perspective for 90 days or 100 days or 120 days and then you have to say if we agree to that then that means when we get serious about this attribution that means the window we’re looking at is I’m not going to have an answer for you for six months because when we try something we’re going to have to try it let it run long enough for enough data and then sit and wait for that window that kind of retribution window. Is that where things get tripped up and get messy that we slow the speed of digital is once a year.

[00:25:34] You No I understand you know there’s a huge cultural issue inside. I think attribution I would just rather have it out on the table. Right. And if that’s what people want then that’s what they want. And if they if they want to stick with the shorter term thing then that’s fine too. Either way it depends on what the business believes. But I do think that people should at least have this discussion and say OK this is kind of how it works. And so are we really going to get into attribution to the point where we’re actually going to start shifting spend based on the financial outcome of these things or are we just pretty much interested in just generating visits. And we’re not that concerned about what happens downstream. And I wonder if a lot of people have even had that discussion. And if it’s not it doesn’t matter to your boss or his boss or her boss and so forth and then it doesn’t matter. But then that really changes the importance of attribution as a topic and really you discussion because what you’re saying is that what we want is a lot of visits. So whatever you guys want to do you know in terms of the spending allocation doesn’t really matter as long as you drive more visits. We’re not really interested in how it works. We’re not even measuring it. What we’re interested in is the outcome. So you guys figure that out. And you know that’s one step beyond business.

[00:27:09] It’s even if it’s the last touch conversion rate I mean I think that’s probably more you know as from last click perspective I mean how do you measure customer experience for example.

[00:27:21] Right. You hear everybody talking about customer experience and you know oh we’ve got this experience that experience and how does that ever get measured and translated into money. This is one way to do that by doing real attribution if you really want to know what the value was of the new interface you created are those kinds of things you can get there through doing some serious attribution work. But mostly that doesn’t happen yet. Right. Mostly people will pop surveys and say Do you like it or you know and those kind of things which aren’t necessarily stuff that translates into bottom line money for the company. It’s great to know. But you know am I too focused on trying to generate money for the companies that cut that out because that is sort of it. You know I don’t know sometimes that doesn’t matter and I understand that too.

[00:28:12] But it’s true that discussion right our creative team feels that they’re being inhibited by your desire to generate money.

[00:28:20] Right. He’s not going to create a new creative was to generate money as much as they are wonderful wonderful people and awesome at generating experiences that make money for the company.

[00:28:35] One of the things that we’ve gotten good at in getting customers to care about attribution is to again it’s just kind of fun with language right. So the point I made earlier about what’s an email address worth that’s worth 20 bucks like there are there are some metrics that have an offline equivalent that aren’t just I bought something and most of our examples again will be retail. We do a lot of retail so there’s that very clear. I have a cash machine on my Web site. I get this much cash in it every day and people have a tendency to not look at anything else and that tends to make them more likely to not think about a more sophisticated way of looking at their marketing mix because I spent the nickel I got a dollar all the numbers are nice and clean. But if he can start to get people to say things like we used to serve as a couple of years ago where we’re paying 20 bucks an e-mail address therefore if we generate our own I guess it’s worth 20 bucks and we know that when we have someone’s e-mail address we are likely to sell them something in the first year like most of the pieces tend to be there to bring together the narrative to start to get them to think about attribution as an on purpose. Okay that makes sense when you put it that way initiative. And another one that I found to be very effective is like we haven’t shit on display yet and I’m looking forward to that day my way. Dr. Tunis ready to hit it.

[00:30:01] Hit display with but you might call that the standard temperature pressure assumption. If this then that provided you know we’ve maintained the system at the same standard temperature and pressure.

[00:30:14] How’s that work out with your marketing clients. The ideal ideal gas law and you’re right I think it’s a high school chemistry because in our team just like that I’m a data scientist.

[00:30:31] All right.

[00:30:32] So let’s do talk about awareness and digital display because I do think that’s something that companies run aground on all the time because they try to do things like measure display like direct response. They try to run display only on awareness only. So what have you guys seen out there. What really take your fancy on this topic or is display just completely worthless.

[00:30:59] The point I was just going to make a minute ago is that as far as display is concerned and then you got into like gas large center temperature pressure Yeah it’s my favorite retargeted display or programmatic display. I have found in the data to not be very good at generating a first sale with someone but it’s very very good at generating that second third and fourth sale. It is a great driver of what you would call lifetime value and if you start looking at the attribution mix or the campaign mix that generates a first time buyer and you compare it to the marketing mix and the cost for generating a second time buyer. You start to very clearly see that certain types of campaigns play much better for certain sales versus repeat.

[00:31:46] That just brings that I have got a client where the media agency says basically ignore everything we’re doing to get people to the site. Initially they have an insanely high bounce rate for their display but we’re just building their retargeting pool. Yeah I guess how the math works on the retargeting there only factoring in the what they paid for the retargeting click and I’m like what the math is pretty simple that if you spent a gazillion dollars just you could drop a fucking pixel and they bailed from the site and then you bombarded them with retargeting. But that is trying to get to first say oh it’s interesting that if you really thought about that retargeting post putting a retargeting pixel on the order confirmation page. It’s kind of an intriguing when it does. Don’t keep trying to sell me the same kayakers shoes or shorts that I already bought but just started keeping me top of mind because you know I purchased. I actually haven’t seen clients do that because they’re trying to get the first sale. That’s kind of an interesting idea just extrapolated erroneously. But that’s where I went.

[00:32:53] Things that tend to be high funnel for a first time purchase tend to be much more efficient at generating revenue for subsequent purchases and that carry through to social media as well. So normally when you look at an attribute model there are things that are kind of like top of funnel or very high influencer campaign types. So social media display and again when I think about display I try and focus on clicks and not impressions and you know there are things that are more likely to lead to sales later and you need an attribute model to frankly validate why you pay for them in the first place. But if you were to isolate people that are buying not for the first time second third fourth fifth time display tends to be closer to the purchase than at the top of the funnel and certain types of social media tend to be more likely to close or help close than to influence 90 days in advance. There are very different mixes at least I you know sometimes it’s like you got to make it plain as the nose on the face to get someone to go Oh shit there is money in there. It’s not fun with math you know. Go forth and do this and showing you know again it costs us five dollars and 80 cents to generate a first sale. And here is kind of the the waiting model for our campaigns. Here’s a totally different weighting model for campaigns and by the way it costs three dollars and eighty five cents for the second sale.

[00:34:15] Now we’re starting to build that kind of narrative that a CMO gives you a mandate against you. And we’re starting to touch a little bit now. And I was just curious because it was one of the things Michael said in the preamble and Jim it’s weird to say my own name names dealer. Jim

[00:34:31] are you referring to asking yourself a question are you just just taking over the show. I’m

[00:34:36] just looking into a mirror adoringly but when you listen to the other attribution episode we did talk a lot about impressions and best practices and initial. Did you face palm a bunch of times. Did you have any. Did you have a few moments for your like children.

[00:34:51] No I mean I think that that discussion happens a lot you know. And so really the question now is how with display is will you know if you think that it I mean if we’re talking about your vision if you think that it is so valuable in terms of helping other types of campaigns why don’t you just kill it for a week or two weeks and see what happens. Right. And then added back and can’t you do that kind of testing because you know is it worth it to you to invest that money elsewhere. I mean how serious are you about figuring out the value of display. Right. And so in some cases people just say we believe in awareness and so we need to you know be out there and just generally create awareness. And you know from a marketing perspective after all I am the marketing guy supposedly. So I’ll say Yeah well that’s great if that’s what you believe in and you don’t want any proof for you don’t need any proof for whatever it is then that’s what you believe. I mean maybe you know this would be sacrilege to say in an analytical context but I do believe that I’ve seen enough times where a gut feel was actually the right way to go. And that was based on that person’s very in-depth experience with that particular product or that particular kind of customer or that particular kind of media or whatever. And they were right. And actually when we ran the test the test came back you know he’ll let you kill it.

[00:36:31] But I’m telling you you know this media works for us and I’m just telling you that we’ve never done a controlled test. I know all that you know I know the whole lying. We’ve never done it before. Let’s go ahead and do it. And he was absolutely right that when we killed the display everything else the performance of everything else dropped and we proved it in a controlled tops. So you know it’s one one way or the other. I just I’m just more in favor of people actually talking about it rather than just kind of going on autopilot. Right. If we’re going to have a serious political discussion let’s prove it one way or another if we want to get right to it though my issue with this play isn’t actually display it’s display salespeople.

[00:37:20] I thought display analysts media analysts that we get to throw them into now who do they work for. You know this helps people.

[00:37:28] See at least I know that creative people don’t listen to this podcast so I can like say bad things about them and get away probably I’m driving listeners away one at a time. You can attribute every minute of the episode.

[00:37:43] So I do think so awareness separate from display. Just saying we’re going to drive awareness in general. I think there are and maybe this is to the gut point that if you are a small company or a small brand and you are trying to grow in a market that’s got established players. There are cases where you say you know what we’ve got to spend a year just getting our name out there so people feel like it is a known entity a known brand and they feel comfortable. And I think that gets back to this really long cycle that in those cases you could say we have got to build awareness. I think I use Carbonite on the last example or there’s I think Bluebox or some sort of I get a lot of the podcast examples are things that I’ve never heard of and then I hear them repeated repeated repeated repeated and I may still not buy. I may never buy. I may not buy for a year. I’ve never actually done anything on SquareSpace. That’s a good example. I’ve never created a website with Squarespace but I can. I can totally see this is where we’re working on our first promo. We’re going we’re kicking it old school TV secretly Squarespace is paying 7 percent from Waisman.

[00:38:57] But the fact is that’s when it gets really Squarespace is a good example.

[00:39:01] It’s really tricky when a random person is saying hey you’re on the Internet. Can you help me build a Web site. And I’m like fuck off go to Squarespace. I’m sending him there. There is no way in hell that the umpteen times that I’ve heard Squarespace promo code on podcasts is going to be directly attributed to real me as an individual but I could see somebody Squarespace saying we have got the kind of carpet bomb the right audience and that will we’re going to kind of get that put back into their lizard brain so that when the people come to them somebody drew out a a kind of causal model of saying we need to go tell the people who are never going to use our service but who are going to have people asking them to do the work for us. If we can get ourselves planted in there then ultimately those people will buy. That’s a incredibly challenging attribution problem and I’ve been here about Squarespace for three years and I might not recommend anybody until two months ago. And I guess you could still argue was that a good good outlay of money but that to me seems like the one of the really challenging attributes in scenarios dots. Yes.

[00:40:16] Well the fact that you know Squarespace more than you know Wick’s or Weebly or Jim which would have been next on that list and have created a wiki site. But the thing is obviously that awareness building that they did put Squarespace is the first one that you think about.

[00:40:36] But then I built a site with a Wick’s ironically for a trip to the Bourbon Trail which I’m going to tie this back when the bourbon comes into needing to bourbon is one of those things that has to sit in a barrel for a number of years before you try it.

[00:40:50] So there are certain distilleries that are trying new things and so it gets back to that whole time frame thing is that when you’re when you’re making bourbon or when you’re taking a longer window for attribution there does seem like a case to say you have to very deliberately documented very clearly say we are going to try things out now and then we’re going to wait for them to to age because we need to have cohorts to work with. We need a decent attribution window. And that I just don’t run into that with companies at all today. Now maybe I’m not. Maybe I need to instead use them the you know gas based formula I can use bourbon analogies and say let’s do this now.

[00:41:33] But we’re not going to get results for six months. We have to let it ripen.

[00:41:38] I like that. You know instead of being like standard temperature and pressure we can just be like you know I’m the brewmaster your campaign brewmaster.

[00:41:48] We’re going to concoct a media mix that’s going to work for your company might get some Pappy Van Winkle might get some rotgut furniture cleaner that fits our culture pretty well.

[00:42:03] OK. Well as always the time is flying by.

[00:42:07] We should probably start to wrap up but would let’s go around the horn here and like what did you hear was interesting today thing prompted you to think about doing more research on a topic and if you’re Jim Novo just you know you could just say something nice about people out there trying to do retribution or give them a good tip that maybe people named Jim people named Jim Kane.

[00:42:30] STERN This is kind of getting me thinking I’ve got to go make another run at cohort analysis and play around with longer windows. Just looking at results and the kind of pivoting back to the customer stuff that you know should be every analyst knows they should be doing anyway but nobody listens as far to the episode.

[00:42:49] So hopefully it won’t be revealed as an utter fraud. You.

[00:42:58] Know I’m right there with you like I tried to avoid this topic for as long as possible. Mostly because I just didn’t want to get. I didn’t want to end up focused on media analytics. For the longest time of my career probably because of media salespeople.

[00:43:16] I think that’s the other. The having the Have we really agreed to what changes we would make and how we’re defining value is everybody on the same page. We’re not having debate over that as a we want to do attribution so hey we just got to knock out these other couple little things first which are pretty meaty things that will deliver value in and of themselves.

[00:43:39] Yeah. And so like that’s kind of one of the things Jim Novo that I really liked that you kind of started out with is like OK put aside the minutia of how we’re going to do this and just let’s talk for a minute. In frank terms about what this really will mean for us and what hypotheses we’re going to test and what we’re going to do different if we find out that you know 90 percent of our revenue or value comes from this channel directly or that channel directly or if we cut that out it doesn’t do anything to us or it only limits us by 10 percent. So I think that is probably something that’s really worth doing. It’s just you know be the frank conversation. Mr. Kane anything from you.

[00:44:23] Yeah I mean I could do attribution every week. That’s one of my more favorite topics. I mean I get right into it all the time right. Seems to be the place where senior executives want to go. Very few people have a path. And then once you start down the path to trying to do attribution they realize they need to rebuild the whole damn house like it’s not one thing. Right. So it ends up leading the data quality campaign tracking in what is a campaign and how do we tell it like people don’t get to I want attribution to I have attribution relatively quickly. But again especially in e-commerce that their conversations we’re having all the time and I was just thinking as we were talking about core reporting and again you know I tend to think more and Google Analytics personally but there’s a section in the customer segmentation building functionality in GA that are called advanced sequences and sequences give you the ability to build behavioral segments over multiple visits. And if you’re thinking about trying to follow through on some of the best practices from Mr. Novo at all today asking questions with advance sequences would give you the ability to formulate some pretty cool hypotheses and it is a feature of GA that people do not use enough and Ben Gaines has paid me to tell you that you can also look at cohorts. Now an analysis works space.

[00:45:47] You get out there.

[00:45:51] Great stuff. Any closing thoughts. Jim No.

[00:45:55] Yeah I think it was great to you know sort of walk through this. I always you know more data is usually a really good idea and I thought that when you know attribution first became easy and accessible and so forth it actually got people thinking in a new direction that hadn’t been taken before and I always think that that is a good idea. I guess what I’m trying to evangelize a bit is going are we gonna go to the next step because as you guys pointed out this kind of covers a whole bunch of other things that maybe we should think about at the same time. So you know I would rather see us start to really dig into some of the rest of all of this right instead of just kind of passing it over and moving on to whatever the next thing is right let’s you know let’s dig into it some and let’s see where it goes. Because I think there’s a lot of value to be had in it. And so once those discussions happen then maybe some changes and some other kinds of thinking happen and maybe it will all evolve into into a better group of people in terms of driving real business value at the bottom line.

[00:47:18] You’re a dreamer. Jim No yeah I guess.

[00:47:24] Well actually no.

[00:47:25] I think that vision is awesome and honestly it’s what makes me love this industry. You know as you’ve been listening out there to us I’m sure other questions probably popped into your head or other things that come up and we would love to hear from you. You can reach out to us on our Facebook or Twitter or on the measure slack and if you liked what Jim Noble was laying down there’s a couple of ways to interact with him some more including next month. The metrics Chicago he has a full day workshop called applying digital analytics and I think you can expect more of the same awesomeness at that. So I think you’d be well worth your time to check that out. So we’re very delighted to have you thank you so much for taking the time. This is a topic that doesn’t doesn’t get enough smart thinking and we’re glad to have some of your smart thinking on our show.

[00:48:20] And of course let’s take you now to the rest of the world then for my other two cohost Tim Wilson. Jim Kane penalising care.

[00:48:34] Thanks for listening. And don’t forget to join the conversation on Facebook Twitter. We welcome your comments and questions to the dot com Fullwood slash and on and on. Now on Twitter. Smart guys want to play it off a little.

[00:48:52] Later. All. Right. Perfectly. Let’s roll. Try saying something. I’m saying something. Oh yes that’s it. Got it. OK.

[00:49:11] This could be attribution. Part 2 for.

[00:49:22] Making it personal said 2050. No not really. Is that a thing. Good reporting. Yes.

[00:49:38] Yeah right it’s definitely one that if you can if you miss like like a bee you missed like three jokes. So the writing is very compact delightful kind of like our podcast. Oh yeah.

[00:49:51] Our audience likes Tim’s tone of voice.

[00:49:54] So it’s it’s all right a statement statement made with no data to back it up whatsoever. It tests well with our Web analyst demographic. Or.

[00:50:07] We’ll just keep that part that involved all the attributes and then delete everything else will be a very short episode. Start with it. Yeah. Yeah. Jim Dovo says attributions socks.

[00:50:22] Yeah it was literally we were sitting in a bar and it’s like we really ought to come up with some kind of thing for the best student in the UBC classes and so people were just talking about different ideas and then somebody said let’s call it the Jim noble word so we can talk about something else. And. They bring it in and you’re doing like ten thousand dollars and. Then that drops down to. 2000. And. Was. Like wow you find some what they can do and.

[00:50:59] I’ll be glad of what you do five bucks a minute. What. Would you be kidding. Me. Oh you see that is what I’m holding up my microphone. I look all poised and excited. Yeah I’d probably like when we have gas there like. I am you. Know the short story. I don’t have it. Red.

[00:51:34] Flag. No no. I think that’ll.

 

Leave a Reply

Your email address will not be published. Required fields are marked *

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

One comment on “#036: Attribution Revisited with Jim Novo

The Digital Analytics Power Hour © 2019