#013: Musings on the Measurement of Display Media

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.”

[00:00:24] Welcome to the digital analytics power hour. This is Episode 13.

[00:00:32] Well thanks everyone for joining us. We’re very excited. Our 13th show just means we’ve been doing this podcast for exactly one half a year. Thanks for listening and being a part of it. This show we’re talking about an exciting topic measuring display media. Everybody’s got it. Nobody’s good at measuring it. What are we talking about we’re talking about display. You know there’s a famous quote I know half of my privatizing budget is wasted. I just don’t know which half. Well you’re about to find out. Joining me tonight are my cohosts Tim Wilson from Columbus Ohio. He’s a partner at web analytics demystified and Jim Kane double CEO of napkin and Babbage systems. And Michael hobbling. I lead the analytics practice at search discovery in Atlanta Georgia. Gentlemen welcome my call. And Jim and you well guys display advertising tons of money behind it and frankly we probably all think measurement is done poorly most of the time what are we going to do about it.

[00:01:40] Well I was the one who lobbied hard for this topic because they have such a love hate and largely hate relationship with display. Maybe I can I can start. And I know exactly where to start. Except maybe part of it is framing that to me there really it comes down to sort of two fundamentally different ways that display can be used. And one of those is kind of in a direct response you know I’ve got a call to action come in buyer come and give me your name and get a coupon or you know become a read so I can sell you a car and then the other use of display is more akin to a historical TV advertising.

[00:02:24] And it’s oh it’s it’s awareness.

[00:02:26] We have a very we might have a call to action but maybe we’re just trying to beat you over the head with our brand so that we’re working our way into your awareness and consideration mindset and I feel like display it’s sometimes sort of muddled between which of those two in a specific execution is trying to be and it’s out there with some strong call to action.

[00:02:53] And it’s initially thought it was some direct response. But then when you look at the clip Trues and the conversions on the last click and we could talk about retribution and that looks terrible compared to every other channel. And then all of a sudden there’s this pivot to oh no no we’re just trying to get impressions you know we’re just trying to you know drive awareness of the brand or the or the messaging and that just drives me utterly berserk because you actually have such a misalignment between measures and what you claim you’re trying to do. So that’s kind of an opening salvo you guys know you’re firmly on the record as being anti display.

[00:03:31] Well I take a softer view now I’ll say this I think display typically gets either too much credit or not enough. And that might sound contradictory but that’s usually where it lands it’s usually in a place where someone is giving it some whack a doo 90 day view through Crazy Eights kind of metrics to pump up its numbers and give it credit for everything that ever happened under the sun or it’s that one to one direct response model where it never performs because displayed by its very nature is just not a transactional type of media. It’s a good display builds consideration. And if you have great advertising if you have great creative you’re going to build consideration with that advertising.

[00:04:19] I think you are if you are really going heavy on display like I feel like what will happen as you run some puts some in some heavy into TV and you do a little bit of display and all of a sudden display it looks awesome because people are theoretically getting exposed to your ads on digital but they’re really getting exposed on TV and they come to your site and say well you know they they’ve got a picture or you know my name has been viewable impression when it comes to the impact of the impression we were talking a little before.

[00:04:50] Michael I see that you said you weren’t really familiar with this. Jim you ever heard of bazoo they’re owned by. Now well you know Tim Elena Tuath well so to me this is such a simple concept and it’s been around for years and I had the fortune of a client and their media agency kind to surprise me and actually ran a visit study.

[00:05:14] So you’ve got studies where you say oh we’re doing dynamical logic or something that takes months after the campaigns run you get these very crude measures. Visu is a fairly simple simple concept and that is a one question in banner survey.

[00:05:32] So they carve off a portion of the media by and it’s used in a lot of times it can be just like value add by the you know as part of the buy.

[00:05:41] And they carve off a portion of that by and because they’ve got people pixel they know who has been exposed and who has not been exposed to the ad and how many times you know through their network. So one of the times instead of serving up the ad they’ll serve up this awareness or consideration or purchase content question in banner one survey and you’re like whoa. Not everybody is going to click the survey or you know that’s fine. So there are challenges with surveys. But but the last don’t ask a question you know which of these you know have you heard which of these brands are you familiar with or you know this message. No have an unbranded survey to ask a question and because they’ll be getting responses they have to have I think like 100 responses is what they normally put it that something like that and they will then have effectively conducted a testing control environment where they’re asking people who they know have been exposed to they have had the ads served up and they will also ask people that is you know from their records the ad has not been served up I don’t know at the left between those measures and say yes to people who’ve been exposed to 10 percent after they’ve been exposed three times and they can get that down to the creative level and they get it in a very timely you know just like we’re used to getting with web analytics they get that data very timely reasonably granularly and it kind of drives me Berzerk if somebody is doing a saying oh this is about branding we’re driving awareness we got 10 million impressions.

[00:07:11] Michael if you get 10 million impressions I want to know what was the lift and brand recall that you got or the lifted purchase intent that you got between the people who were exposed to that ad versus the ones who weren’t. And what frustrates me is that that is kind of on the media agency to do that. And they can optimize their creative they can optimize their placements with that data. But a lot of times they seem sort of reluctant to do it and it was great this week actually saw one where it was fantastic like they’ve done it and they actually got great results. And I was like This is the same frustration when people not want to measure their websites are being revoked. But keep your eyes down to say look this is what you are measuring. There is a platform out there. You know I’m getting nothing from Nielsen I have no relationship with them but the handful of times that I’ve managed to get through and we’ve used visit it’s been very very eye opening and kind of you know super useful.

[00:08:10] So I mean my initial two senses you know you get a bunch of digital analysts in a room and there will start bitching about display and about vendors and agencies that sell display. We did it right before we started recording the show. And you know if if you look at the International Interactive Advertising Bureau or the IAB it’s huge and it’s extremely well funded and it has a lobby group in Washington. If you look at the DA or the digital analytics association it’s about 5000 of us and we periodically have a lobby bar in San Francisco. You know there’s the level of influence and the level of budget because the measurement is significantly smaller than what goes to advertising. The IAB now has a digital data certification. So through the IAB they will teach you how to measure and show data that helps someone understand their advertising and their display budget.

[00:09:09] If you try and explain that to a Web analyst or a digital analyst whose job it is to understand how everything is connected to everything else it comes off as a little bit bigger. You know we’re looking at the holistic health of the body and they’re kind of doing podiatry because we only sell shoes.

[00:09:26] But that’s another good point that if you look at where that budget because if you look at the scale which is a great point. But that’s coming from like display is more of an offshoot of TV. And if you take the dollars you can spend on TV advertising and then what it takes to get a ton of impressions way more visits than you’re going to get to your website.

[00:09:48] So there is that kind of going from TV all the way to the Web. And I think when you get to the site it’s kind of seen as this sort of free thing. And so we kind of scrapped to say we need to invest this site but still when it comes to the marketing budgets what they’re actually spending on advertising. If you’re spending on TV that sort of dwarfs everything if you take if you take in absolute terms what’s going to digital advertising.

[00:10:15] It may be that paid search and display you know that those of us sitting just on the Web site are looking at that saying well that’s still a massive number but it’s kind of small compared to the TV budget. So I think you’re onto something in that if you’re coming at it from a TV perspective and how horribly unmeasurable is you know Nielsen and how messy and noisy that data is there’s this belief within but it came from that mentality that your display is so much more measured. Metalworks you could count the specific number of impressions whether they’re viewable or not. It’s a little harder to account and people are trying. Trying to do that whether they’re being cooked bottled to the the end of the earth. So they’re not really clicks. That’s a little bit harder too. But there is a you know we look at it on the Web site and say it is certainly more reliable than a click though academy is saying a click is certainly more reliable than an extrapolated Nielsen box. It’s trying to estimate in the world of time shifted viewing. So I think that that may be kind of two different perspectives of how good is this measurement.

[00:11:21] Yeah and the other part is that as digital analysts and people who are digital marketers we have it just like you said Blake trying to measure TV forget it. We’ve had it relatively easy in terms of being able to measure digital marketing campaigns in general because most of them would be the equivalent of direct marketing. Edwards show a message.

[00:11:41] Did they click it. That’s all that counts. E-mail sent a message. Do they open it. Did they click it.

[00:11:47] And I think display needs to be considered as an aspect of a much more significant multi touch campaign and not a campaign on its own.

[00:11:56] And sometimes that’s hard to wrap your head around. Secondarily I think that sometimes digital marketers have gone. I got an extra 20 million dollars kicking around because we’re a huge company. And I’ll buy a whole bunch of display because it’s like paid search rate and this is interesting with paid search.

[00:12:14] You’re partially buying on keywords and you can have some context around what were they looking for and there are some rules of controls that you can’t just march completely off the reservation that you’re trying to push them towards whereas display you’re buying audiences that have affinities based on behavior or the content of the site that they’re amik is just inherently to me it’s a loser connection. You’re still trying to get your target to display but then that’s why people jumped to retargeting and they say other retargeting. I know they’ve done something they’ve been to my site. They’ve done something where they’ve expressed some interest and look retargeting all the sudden magically does get higher click through rates it costs more but higher conversion and that logically makes sense.

[00:13:06] I’m not anti retargeting although I. All right metrics.

[00:13:09] Those are the ads that I get that I might get that is Retargeting or just visit search discovery dot com some time we retarget like a mo fo. I’ll save a couple bucks and not go there. There you go. Yeah I tend to like retargeting over just sort of hey let’s throw some display advertising out there and because it’s not analogous to television. Right. With television we kind of have a sense for who’s watching even a DVR and fast forward commercials and that sort of thing. It’s people still see the advertisements. You know it’s well proven that there’s a lot of banter blindness out there which is why media creative HESTA really stand out. If you’re going to do run of the mill or brand based advertising or retargeting feels more relevant. And what I’ve seen you’ve seen a lot better results from using retargeting yet and also using DMP and things like that to build audiences to do targeted display too. So not retargeting but targeted display is another area where I think that’s why you see DMP just being now blowing up like they’re like some of the fastest growing digital companies right.

[00:14:25] Sorry just to clarify what you’re talking about other people are programmatic display right Michael.

[00:14:30] Well programmatic I’m bad at defining things programmatic is more like I give like a system a set of rules and it’ll automatically optimize within that set of rules. Targeted display to me is just sort of I find look alikes across the Internet. So I have you know hey these all these people visited my Web site and I’m going to use a DMP to tie other attributes like 30 percent of all my purchaser’s happened to drive Volkswagens. Now I want to target every Volkswagen driver out there and see how many more ACTUP get to my Web. That kind of stuff. Doing segmentation and look alike target me.

[00:15:06] But it’s one of those where you say programmatic that implies you start saying machine learning and an AI and it’s an easy sell and it’s an easy play for venture capital for acquisitions.

[00:15:19] All the big players have bought something that’s a DMP and it’s I’ve got a CPG client where the head of analytics is kind of on a tear to say OK you told me that you’re doing this but still how are they paying the agency they’re paying them a cut of the media spend. So the way they won the business was we are going to spend your dollars wisely and we were going to do programmatic we’re going to optimize throughout the year.

[00:15:48] But the fact is once you set a budget they’re just getting a cut of that and then they have to be able to tell some sort of story about how it got better and the more that he digs in because it’s actually complicate right people have this vision that oh I have all this data I’ve get on it DMP I can look for all the Volkswagen drivers. So I think it’s a good story but the complexity that comes with that when you start making those buys to then say was told me to figure out is it working or not and how am I measuring whether it’s working or not and do I have that closed programmatically or am I having to do it manually. So I think that’s a whole other issue that I think there’s the story the hype around programmatic the hype around DMP is I think is I mean honestly the hype around Web analytics all of those are exceeding the reality.

[00:16:40] So I don’t think I don’t think site analytics is necessarily any better but if you’re not actually closing the loop and saying how am I measuring if this is successful which is where attribution comes along and says oh the only way you know if it’s successful is if you collect massive amounts more data and hit the magic button on our system which is going to take this wildly messy data and telling you whether it’s more successful or not. But at the end of the day you’re like based on what like what is your end of the day conversion. You know what if you’re selling shoes online sure you know that is a reasonably low consideration purchase. I can see multiple touches on one device leading to some direct response stuff to a purchase but there’s a ton of display out there that’s not doing that. And there are companies buying the technology to do attribution and they don’t have a conversion that they can really feed into that system.

[00:17:39] On the point that you made Tim that was a good one. You said the agency that’s doing the analytics and by the way this is not a statement about the quality of their analysts or the work that they do with the contractual relationship they had with the customer was as a percentage of spend. So their mandate as an organization is to increase spend as much as possible and not get fired. It’s very hard to be a good analyst. We know that’s your mandate. And I find that as analysts who literally just do data our mandate is pure offense like how do we come up with clean actionable insights to grow the business and display analytics tend to be owned by the seller of the display and they tend to be defense. How do we protect the customer. How do we grow the budget. And I think that’s another kind of D.A versus the West Side Story thing.

[00:18:32] I don’t disagree sites side analysts have a bias towards the web analytics platform and I’ve said more times than I care to count that you know what if Google or Adobe didn’t see the visit if they weren’t on the page long enough for the page to load and for the javascript execute then I don’t really think they visited. You know if they bounced before it even got red whereas you’re counting a click which was the point that you passed through some redirect and liked it on your server.

[00:19:02] So I expect some fall off in that situation.

[00:19:06] But it does seem like there are cases more often than not where the media side analyst their system their universe what they can control is counting to the point of the click and then maybe reading pixels that got deployed on the site.

[00:19:22] But you know that’s where the friction occurs and I can understand why it’s happening and I can understand where my bias lies but it doesn’t mean that it doesn’t wind up with a lot of potential for fingerpointing and a kind of a fractious relationship between parties.

[00:19:42] Well you know again I’ve never had the opportunity to really take even digital a really big multifaceted marketing campaign like it’s our spring launch and we’re going to do this kind of display in these 10 e-mails and these locally targeted ackwards ads but they’re all fanatically similar and measure them all together which is a much more traditional big marketing campaign measurement program like we tend to focus more at the channel level because people don’t want to roll up. If we were in a situation where we could do that traditional marketing campaign level rolled up. I have a feeling I’d be going we need to get a lot smarter display who can I talk to. So again I’m not necessarily just shitting on agencies. I think that we need to start is analysts thinking about campaigns is not. Yeah I sent an e-mail but this is a significant organizational marketing push and that’s complicated stuff.

[00:20:37] You know I feel like to that point of kind of viewing as a multichannel ad I’ve done this a play this role a little bit of it as an analyst not on massive massive cross-channel campaigns but playing the role of somebody sitting down and saying are we even thinking through how we expect these different things to work together and this will be maybe the analyst being the heretic but that just putting conscious thought into that and saying we’re not going to be able to measure exactly which one of these things that worked or not we’re going to to measure the overall success of the campaign I think would be higher value more often than not saying we’re just going to assume that’s kind of happening. And we’re going to try to measure something that is largely measurable which is the incremental contribution of each display relative to the TV relative to the e-mail relative to the page search.

[00:21:34] I’m going to say something that I’ve only said to one person out loud about media or display and sort of my thoughts on it at least as it pertains to like attribution and that kind of stuff. You know we’re recording this right. Yeah I know the other person I told about it was June Dershowitz just threw a pastor one time and she looked at me like House of crazy and probably I am but I kind of think that there’s also sort of a sunk cost to advertising like a sunk cost to servers to run my Internet Web site. You know I mean like I have to do some advertising otherwise I will have zero customers. Right. Or I can depend on word of mouth only. And that is one way to grow a business. But if I want to grow my business beyond that there’s an amount of advertising that I have to do you know. So in some senses like running a display campaign is something that just kind of has to happen. And I don’t want to be somebody who says hey let’s not measure this but let’s understand that if we look at sort of customer life cycle how like my new customers coming in and my old customers going out with measuring turn a new customer acquisition and those kinds of things. When I look at my world like the way that display fits into that is very much around how I get new customers into the final. That’s a very high top of the funnel activity that’s about gaining attention and driving awareness and consideration towards the new customer.

[00:23:03] Or I guess the read retained customer but they’re separate except for retargeting right retargeting kind of by definition are beyond awareness by the time you’re retargeting them right.

[00:23:14] That’s fair. Unless I’m doing lookalike targeting which is not retargeting but if they’ve done something to interact with me before but it’s still kind of moving down the funnel right. It’s sort of like if they’ve come to my site and shown intent. And then while I’m browsing the web while I’m listening to Tim talk I’m seeing Zappa show me a shoe that I looked at a couple days ago. It’s a retargeting campaign. The other thing I was going to say is so there’s my crazy notion that out there someday history will prove me right. I believe that.

[00:23:45] I don’t I don’t disagree with you. I mean I will Quickflix I don’t disagree with you. I think aiming for more precision on measurement as opposed to saying we have to spend it. And what are the things we can reasonably do. Can we use a tool to try to estimate viewable impressions because of the impressions not viewable. Ben it can’t possibly be helping.

[00:24:05] You absolutely do that. If all of the publishers would use a really high quality tag management tool it would be no problem.

[00:24:15] When I we use that to throw away and just because anybody who has hasn’t been seen the ad age is kind of a big splash. I mean at age Kerry others did when craft maybe like a year ago less than a year where Kraft said they came out and said they reject like 75 to 85 percent of their digital Adam Prestons due to quality concerns and I think that was a few of the ality and goes clicks. And I think saying hey we think this is happening what can we do now to try to figure out how we pull up the sites that we’re advertising on. You don’t get that from say I know it’s the network you’re pushing on the network. But show me where the impressions the bad and a look at them and say This is ridiculous let’s start blacklisting these let’s sit down once a month and blacklist sites that just make no sense for us to be advertiser on without trying to get to the point of saying no we’re going to wait for the data to magically tell us exactly what the contribution was we’re going to do it magically through the data.

[00:25:07] So I actually think I think I agree with what you’re saying or I just completely misinterpreted it to do my own ends.

[00:25:14] No let’s keep going down that path. I like it now.

[00:25:19] You know this sunk marketing cost. You’re talking about Michael is really just how do we tie display back into Costas sale like. I mean he’s still thinking like an analyst you’re not saying some things are just truth and we should hope that advertising just works like you’re saying we need to measure it. It’s not last past the post attribution model. But if you turn up all your display you’re a huge brand. It’s not going to be a good idea. It ties back to something you guys are normally really good at educating the on these podcasts by referencing Chris wrote a blog post. This person wrote a book I read the interesting thing here has a Web analyst ever approached this issue in a really really detailed way before and you would like loved it. And if not who would you want to write the article.

[00:26:10] I’m differing views of that somebody I’ve read a lot. It does quite a bit of work in talking about this is probably Kevin Hellstrom and talking about multichannel. Surprisingly he’s kind of down on the omni channel idea but only down on the hype surrounding it as opposed to you know whether or not it’s meaningful Saad’s. Yeah mind that data Dotcom is a good site for anybody to read.

[00:26:32] I don’t know about you Tim but seems very very rare to me for web analyst to jump over to the media side or vice versa. So part of the reason I wanted to talk about it I feel like nobody really wants to face it head on or can’t figure out how to face it head on. And the IAP versus DKA is a pretty crystalised way to frame why that may be.

[00:26:57] Do you think we it put some peer pressure on Mr. Malvo to take this on. He seems like the right guy.

[00:27:03] Actually you know you bring up. See yeah that’s what I would have said.

[00:27:09] Yeah definitely made it up to about that. I don’t think I don’t think we can have peer pressure on them.

[00:27:14] He’s a fairly independent thinker but I just like his breakdown over the plate for him to go. OK. Here’s what it means. Here’s where it. Here’s how you measure it.

[00:27:24] Yeah. Think of it. He’s actually the one who the first time actually started to believe that retribution could work was actually a presentation he gave and he was I think more talking explain kind of the you have to. You have to have a hold out. Right. That’s the only way you can really do retribution is you have to have some level hollowed out and it’s tough to do with that display. But yes this is our digital analytics power our requests that Jim Nuovo tackle that head on.

[00:27:55] Yeah all seven of our listeners please contact Novotel and Twitter.

[00:27:59] Right. Time for you to come out of retirement. Jim Dustoff of the our allies spreadsheets that I’ve found on the Internet. When I first started analytics are basically solving this problem. The other one that I thought of and I think I tend toward more adversarial view to display but there’s a good blog out there called Ad contrarian who I find refreshing. Check that out too. All right so let’s wrap up here’s how we’re going to do it. Each of us tomorrow we’re starting a new and exciting job as the director of Internet advertising and marketing for a large agency. First thing we’re going to do is sort out our display capabilities was the first couple of things on your agenda as you start into that go.

[00:28:43] So now I’ll run first I think that I am first going to get crystal clear on where I am going are going to draw ideas elemental white board and I want to get the people in the room who have opinions on it as to where display actually fits in are. Well it’s going to be the selling cycle although I may talk about it as the buying cycle things are going to say where do we think display is and then I’m going to go straight to that and say What can we reasonably measure and what are we comfortable with kind of taking on faith and try to set expectations really really clearly that there’s a lot of technology out there but there’s a lot that’s really expensive that may or may not actually pay off when it comes to measurement. I’m going to find what I can measure in the near term but I’m going to be super rigorous going to I’m not going to measure somebody on CPM when they’re bidding on CPM and I give them a fixed budget because that’s that’s just ludicrous and happens but I’m going to really map to if they say we’re driving an awareness and say well how are we going to be measuring the awareness lift and what you’re contributing. But I’m probably going to keep it still fairly channel siloed and not get caught up in the multichannel Hi Jim. What are you going to do.

[00:30:01] The very first thing I’m going to do is really take a strong emphasis on starting to think about the concept of the campaign and you know it’s funny because up until today I didn’t realize it you had such a hate on for attribution. It’s another topic for another day but attribution can be like big data as it is a concept but you know kind of thinking about the waiting for the types of marketing campaigns or marketing mediums and actions we do as part of an overall campaign and how they fit together. But I would immediately start. Thinking about the weighting of every type of digital marketing that we do and bring it together into a theme or a campaign. So for example you guys have holidays in the year that are big sales holidays. If I’m a big brand I’m doing email display retargeted display ad words online marketing all around come to my website and buy something. So rather than thinking of display as a thing I would start to really figure out how to rationalize it as a successful part of an overall marketing initiative. And just like Michael said earlier who is trying to hold out testing or what I call controlled testing. Yeah I mean those two things together gives me the ability to look at a holistic digital marketing campaign against cost per action and then I think I would really be able to do some cool stuff with display that as a standalone it would be a big expensive Mass just to quickly defend myself.

[00:31:31] I’m not a total ad I am bearish on attribution as it can be applied to any business at any time. I think there I think it really is contingent on the business and the business model and the industry. But I think there are I think that reason is totally doable in some scenarios but that is a topic for another podcast. I don’t think they haven’t had them take a stand.

[00:31:53] I mean taking a stand that there are lots of places that are thinking about airport in our next episode called Tim Wilson last click apologist here is I actually had a set of steps and then I’m going to change them because of what one of the things Tim said which is probably the first thing to do is figure out how the organization perceives display because you could walk into an organization that is pretty focused on direct response and display just doesn’t get any attention or love because you can’t measure it. Last click or you walk into an organization where you take the temperature of the organization. I think that’s a really great first step and then educate from there about getting more sophisticated I think that’s that’s what I do. And then I look at how do we incorporate strategies that are going to give us left but build a model that shows us here’s how display fits into our customer model. Here’s how we’re going to understand display as distinctly different in its impact versus response channels like email or search or things like that. And then you go to town spending the money. It’s probably why no one is jumping over themselves to hire me to do this job. This has been interesting. I think you know as digital analysts through and through. This is not our sweet spot as a topic and so I think there’s a ton that we could learn from our listeners. So if you’re out there you’re listening to this and you’re shaking your head like all you guys have got it all wrong.

[00:33:30] It’s a great time for you to jump on our Facebook page or on Twitter let us know your thoughts join us on the measure slack group. If you’re not a member of that already. Love to hear from you there and give us some of your thoughts on this topic.

[00:33:44] We’d love to hear from you. Thanks again for listening. For myself and for our co host Tim Wilson. Jim Kane we thank you and thanks for.

[00:33:57] Thanks for listening. And don’t get to join the conversation on Facebook Twitter. We welcome your comments and questions. Facebook dot com slash and Onizuka now all on Twitter.

[00:34:11] Smart guys want to made up a term called analytic links don’t work.

Podcast: Download | Embed

Subscribe: RSS