Governing Digital Marketing Spend Practically and Reliably
By Colin Temple of Napkyn Analytics
Thanks Brian. Hi everyone and welcome. So I’m Colin, Director of Analytics Solutions at Napkyn. I have been working in digital marketing for about fifteen years. What I’ve been doing there is really all over the place. I’ve written copy, I’ve written code, I’ve designed campaigns, I’ve done analysis. Usually I’m the guy on the fence between the technology and business sides of this practice. I’m often the guy communicating between them. In my role here, as Brian said, I’m really focused on keeping us and out customers up to speed on what’s going on. As I’m sure you all know, there’s a lot going on.
We’ve been hearing quite a bit about governance today and there’s been some amazing content so far, so I hope you’re really enjoying the summit so far. What I would like to do with our time together is think about how marketers and analysts can really keep up with the pace of marketing today but still govern the data and the expenses that are involved.
I want to approach that in basically three steps. Steps one and two here are setup. I want to identify the problem and what we’re going to need to solve it. A lot of that is going to have more to do initially with changing our attitude and perspective rather than implementing new technology or jumping on anything like that. Then what I’m going to spend most of the time on is describing an approach to managing marketing activity that I’ve seen success with personally and among our customers. I basically want to make the case for a very product focused approach to your marketing stack. So, let’s do that.
I’ll start with an obvious statement. Online behavior is constantly changing. I’m going to go through a few examples next of how those trends are changing. This is the part where you roll your eyes as I go through a few numbers and some market research as I try to set up my point. I’ll try to Tarantino this a bit so we can really get to the point, but here we go.
So when we look at mobile, we see a lot of growth and traffic, but customers still convert way more readily and spend more when they do, on desktop environments. If you ask them why, they’ll tell you it’s really because that’s a better experience. We’ve really, overtime, trained people to expect mediocrity on mobile sites. That experience usually leaves much to be desired. That’s gradually changing but not yet.
As more and more organizations do move on to mobile first user experience and beyond, that kind of buzzword trend, is this the trend that’s going to continue or are we going to start to see some parody here? That’s something worth considering.
Common ongoing concern here also attribution. Shopify looked over the year to year growth of customer’s transactions recently and they segmented based on the number of channels that users interacted with before they checked out. The growth jumps here with each new channel that you add. So you saw a 25% growth for people just not really interacting with marketing channels beforehand, and then when they’re interacting with three channels before their main visit nearly 200% growth in the amount of revenue that lives in that segment.
Now this graph is a little bit sensational. If you look at it with an analysts eye you might notice that you could skew this pretty easily. You could have like a 1000% growth if you went from one visitor being in a segment to 11, but there is volume to back this up. This is something that we’re seeing consistently with our customers. I’m sure that you are all as well. Those customer journeys are getting more and more convoluted, and they are often, there’s just more of them. There are lots of different paths you can take to end at a conversion and customers are taking a wide range of them.
Globally, we’re seeing a lot of change as well. It’s really different from country to country. Some researchers at Oxford, Michigan State, and here in Ottawa actually, worked with Google to survey thousands of web users in the US and Europe. They looked at their search data directly as well and automated aggregated search data of course. They were focused on the concern of political eco chambers really.
Their findings were more general with that. They started out by just kind of outlining what they were seeing in terms of people in different countries using search primarily as well as social media. There are a lot of profiles that really started to emerge between different countries. They had some countries skewing towards traditional media and a trust and interest in broadcast media. You had other countries where people were frequently using the web to check facts and to get their primary source of information. Then we had Americans that in the study, they called media omnivores. They’re really all over the place in terms of the kinds of media they consume. Not just in terms of lots of people in different silos, that happens sometimes, but overall, individual consumers were using large amounts of sources of information and in shopping, lots of different ways to research products and find reviews and lots of different social platforms as well.
So, we’re really seeing a lot of diversity here.
More obvious, advertising platforms are constantly changing.
That’s pretty clear if you look at this big bunch of logos here. Making this image, I actually had to update some of the logos because this stuff is in constant flux. Again, you know that, but what happens when you start to put these trends together as well as the state of technology together?
More importantly, identifying that customers are all over the place, there’s a lot of places you could reach them at. So reach everywhere, right? I would love to have that budget, I really would, but diversity does seem to be important, so how can we achieve some of that without running out of time and money and falling asleep on our drive home at the end of the day?
This is really where I come in. I’ve spent years running campaigns. I’ve spent years running digital marketing. I have spent years writing code and managing products. The thing is, those are all the same years. While my wife might have a few words about how much I spent fixated on work, she will also probably tell you that I’m actually pretty lazy.
That is because I have the developer gene. That’s that little bit of me that pushes forward the instinct to not do anything twice because that’s boring. When things are boring, I don’t do them. What I do instead is build things that will do it for me. Spreadsheets, software, whatever I can hobble together. If I can do it once and never again, I will do that, even if it takes a little longer. You need to be intelligently lazy here. PLay the long game. Optimize for your future laziness. Or if you’re the kind of person who is otherwise naturally productive, think of all that you could accomplish if you killed the time that you spend on overhead.
The agile manifesto, which is that behind so many software development decisions, both good and kind of horrible, really put this point eloquently. “Simplicity - the art of maximizing the amount of work not done - is essential.” It really is that focus on maximizing what you don’t do. I promise this was going to be about governing marketing spend. What you can never forget is the value of your time. The price of your time to the organization you’re servicing and to yourself. Time is extremely limited and frankly, that shit is expensive. So, when we’re thinking about marketing spend, we have to think about how time plays into it and how we can make an ongoing platform that we can work from and maximize reusing the work we do.
What’s in a typical workflow when you’re launching a campaign? Some people work off a base of what they’ve done before, but each campaign tends to be viewed as this monolithic thing that needs to be assembled mostly from scratch. I stole this basic steps from somewhere, but it’s the rough workflow that I think many marketing teams follow. I certainly see it a lot.
You start off set your goals, set what winning looks like, make all of the content. Creative, copy, code. Tag your URLs. Punch it all into Google Ads or Facebook or whatever subset of that giant screen of logos that you work with. Get the IT folks to place some pixels. Press the go button and kind of watch the data come in.
Organizations that care about any of this will really cycle through steps four and five for a while. Watch your spending via CPC, CPA, or whatever other acronym it is that you care about. You adjust bids, placements, and variants to really promote whatever is proving to be most successful. Then you pray to whatever you believe in that this is going to generate some kind of lift overall and if it doesn’t, you open LinkedIn jobs. This is really the bulk of where the iteration happens in current practice. Step six here is important. Evaluate successes and learnings. Step six rarely happens because at this point, you’re done, you’re on to the next thing. You ought to take a look at what you learned in the last round and bring it into the next campaigns that you work on, but that doesn’t always happen. There’s not always time for that it seems.
So how do we embed that learning, that overall progress into the system? If you are paying close attention, you may have noticed that I’m talking both about iteration and not repeating things which maybe seems a little counterintuitive at first. This is where I say, put on your developers hat or your project managers hat, because this kind of iteration, you need to think about what I really mean when I say this.
This kind of iteration in the campaign steps that we were just looking at, is really good in the context of your campaign, but you’re not really giving yourself that opportunity to be lazy later. Lazy or productive, one of those. Especially if you’re not taking that step six too seriously. You really need to play that long game. You need every win you generate as a marketer to cement it’s goodness into everything that you do. You need to build off of those so that every time you spend a dollar or a minute, you’re improving the status quo. Doing that really requires changing your thinking a little bit about what you’re doing as a marketer.
In order to do that, I think the first thing that you need to do, is start paying attention to your marketing environment. This is a conscious choice and some of you probably do this. What I really advocate for is treating that whole environment as an individual product that you either manage or you contribute to the development of. It’s the platform that all of the marketing activity that you undertake is launched from. It needs to get better over time if you’re going to make any progress.
Okay, easier said than done. I alluded a bit to the pressure that exists a few moments ago. Marketers feel a lot of it. Lots of pressure. You need to keep the important numbers in an upward motion. Everyone in the chain of command has somebody breathing down their neck waiting for those numbers to get bigger. Your bonus depends on it. Your job often depends on it. So, how do you act in the face of that pressure to produce short term results and still play the long game with governance?
There’s an important concept that I think helps here that comes from development. That’s the idea of technical debt. Technical debt, I’ll define for our context, is really all the stuff that gets in the way of being lazy. You need to cut some corners on the tracking and now it won’t handle this next campaigns need. Or you didn’t implement a system that can test itself so now you have that as a to-do for later. The data from salesforce has to be massaged every time before you can align it with the marketing spend data and reproduce that report that your boss loved that one time that you put it together. You end up with so many of these little things that you really get bogged down with the overhead of having to maintain the status quo. Innovation dies. There’s not continuity between campaigns other than the continually chaos you experience. That is not fun.
If you don’t want this to be your life, what you need to do, is work on a structure that lets you really minimize technical debts while still watching your spending and growing revenues over time. Okay, so, what do you do? Literally, what’s the thing you do first?
In my always humble opinion, I think you need to start with some ground rules. Don’t kill yourself drafting a law. Make a basic set of rules about how data should be structured. If you have campaign tagging conventions, write them down. If you don't’, make some up. You can refine later, it’s fine. Just put something down. What are you going to look at really in order to gauge effectiveness here? What are your expectations for spend? What about return? And when you say return do you mean ROAS? Return on advertising spend. Do you mean ROI? Those are very different things so define the things that matter to you in your rules.
The technical debt of ignoring this step is paid for later by the time it takes to build ad hoc reports that reaggregate inconsistent data. Your cart abandonment triggers spelled email with a hyphen but your primary agency spells it without. Big deal. Well, there are sixty things like that so tell me what it’s going to take to make a nice clean report out of that data.
Establish those rules early on, today, if you don’t have any. When they seem good enough, give your vendor's a deadline to start complying with them. Anybody who is tagging data needs to start following this. What you want to do, is as you implement those rules, also alongside them, implement an approvals process for data collection components such as your conversion tags and your campaign parameters so that you can really govern that going forward. If your analytics tools offer the opportunity as Google Analytics and Adobe Analytics do, move toward workflows that really limit the opportunities for error. Make it locked down whether it’s with data import or classifications how things have to look and potentially start rejecting things that don’t match that.
As you build this environment, you really need to kind of cut down on the chances that something is going to break. So this is why I say employee test-driven development. We’re talking about building a significant a product here. One that’s probably going to end up with many moving parts if we’re going to expand the breadth of our reach and meet that opportunity. This can’t be stuff you spend a week or more testing any time you make a measurement change.
So instead, automate. Buzzword of the decade I guess. Automation is going to take all of our jobs, I absolutely can’t wait until it takes mine. The best part about automation taking your job is that you can do new things with all of that time you get back. I mean you came to this event, and you chose my talk so obviously you’re pretty smart. But you’ll survive and move on to something that’s actually challenging and leverages your capabilities as a marketer or as an analyst.
Automate testing to make sure that when you add to your marketing environment, you don’t break the stuff that’s already there. Sometimes I hate saying that because sometimes we make money fixing stuff that breaks that way, but that’s not really scalable and we would rather do interesting things than fix things we’ve built before. That’s an important piece of feedback.
I know some people who are amazing at implementing that. Some of them work at ObservePoint and have some really cool products. Some of them work here at Napkyn and will also point you to ObservePoint. So keep that in mind.
So some of the governance examples I just touched on, focus on the content of your data more than the spending you actually do. In other words, we haven’t met all of our marketing environment’s requirements yet. We need to be able to govern spending. If you came to my talk last year on marketing data, I went on for a while about how important I think context is in your marketing data. That’s still true. As I said before, there can be major differences between the metrics that seem to imply the same things. Revenue is not ROAS. ROAS is not ROI. If you conflate these things because your not looking at the big picture, from your marketing spend to merchandise cost to again the cost of your own time, you and the people you send reports to are going to make some bad decisions. If it’s the latter, they’re probably going to get mad at you, and it is really hard to get lazy when someone is yelling at you to do better.
Okay, so how do we get all of that yummy context into place? Integrations. Choose to integrate data whenever you can. When you’re choosing platforms, networks, ad servers, whatever marketing tech you’ll be implementing next, you need a strong preference for those that make getting data out of them and into wherever you report from easy. Sometimes you’ll have a lot of options. Google Analytics integrates well with the Google stack of advertising products. That has an obvious path.
Some things will be less obvious. Some vendors won’t have a nice clean API or the ability to really automate this, but it’s really important to be able to do that. You need all of this stuff to talk to each other so that you don’t have to spend your time in Excel remediating problems. So whether that’s using built in integrations or leveraging external products to push data together and push data into the platforms that you use to actually analyze it, you really need a strategy there.
Do you put your costs into Google Analytics or Adobe Analytics? Do you consistently and automatically tag adds wherever possible through auto tagging features in a product or some kind of integration? Do the dimension values that you enter between those two activities match up so you can actually equate the money you make from the tag campaign to the money you spent on it? Those aren’t just good questions for you to answer, those are good questions for your vendors to answer. By the way, I would never trust an advertising vendor that didn’t make data accessible and transparent. So knocking out that kind of nonsense is really going to help you for more than just governance.
The last one I think you need to do is to decide that you’re going to test yourself. We all spend time thinking about testing. How is this ad/landing page/email/creative going to perform compared to this other one? How will certain UX changes influence our customers behavior? How can I start to predict that behavior and personalize in real time or serve up and activate ads in real time based on the audiences I define?
You’re a factor here too. As marketers, we need to test ourselves. We need to focus in on our own behavior and take a little more care in noticing, “What are the things we’re doing that breathe success?” When you’re making adjustments to your own process, really think about how you test your customer behavior and apply a lot of those same concepts to your own iterations. Try to think about one thing at a time that you’re changing so that you have some hope of identifying what parts of your strategy are actually making a difference to the money you breathe in. If you can maximize the amount of work that you don’t do, that’s fantastic. That saves everybody money, and you can be lazy or productive, whichever.
So, we’re all full of our own cognitive biases too, so we really need to keep ourselves in check and make sure that this is a system that’s going to promote its own growth and its own successes rather than relying on you to always have good ideas.
So when we think about that kind of step by step workflow we were looking at before for creating a campaign. What is some appropriate revision to that look like? All of our previous steps are still here in this version. They were all important, but what’s really been added here is this underlying concept of our marketing environment that’s represented by that line at the bottom. If you’re familiar with software development workflows, this probably looks familiar. It’s a version control diagram. For our context, what I’m really considering here as the master branch, is our constant tech stack. You know while we have some discreet marketing campaigns that have a beginning and an end, our marketing environment is a continuum that survives that. That’s the product that we’re ultimately responsible for building and maintaining.
As for individual campaigns, there’s still the driver of performance but they are an iterative component of the larger product development piece. The purpose of the underlying product then is to serve as a launch pad for the new marketing activities. As those activities occur, receive feedback from them and improve over time. We have this stage here where the campaign launches off, becomes its own branch here, we assess and add the requirement measurement capability as one iteration. We work on content strategy. We implement and run the campaign and still do that intro campaign optimization based on feedback to improve the spending for this campaign as it’s running.
But then overall, at the end, we analyze the campaign results and continuously throughout as well. You’ll notice that I have two points here where I then take what was learned and incorporate that back into the environment because you want to improve your ROI overall. We’re going to do that not just when the campaign is over all said and done and as that final step that can get ignored if you are busy. But instead, in every step, as part of your process of that iteration, not only implement the changes in the immediate campaigns that you’re running, but implement the learning from those changes into your marketing environment. That might be changes to configuration in your text stack. That might be changes to your settings for how advertising is going to work. That might be adding new fields to salesforce because you’ve discovered new measurement that needs to exist. There are lots of different kinds of things that not only impact the individual campaign but really need to have a global effect.
And finally, start immediately. Right now or as this event ends. Choose to adopt a governance-first approach to your marketing, starting right now. As we wrap up, take a quick inventory, a snapshot of your existing stack. This is your marketing environment, this is the product that you develop. Define the rules that will roll into it first and when they’re going to take effect. Then communicate that to everybody that’s involved.
I really think you should always have a roadmap document for this marketing environment. Whether you call that a measurement plan or something else or a marketing plan globally for your marketing programs. This is where your snapshot is going to move next. Except that you’re never going to complete this of course. This is a living document. It shows you where you’re going, what your next immediate step is. Your going to try to project into the future beyond that, but that iteration is constantly going to push back into this document. When I talk about taking things and moving them back into your marketing environment, as you learn them, this roadmap document is also a piece that you want to update and refer back to as a resource for everything that you then launch off of this marketing environment platform.
That’s your homework. I’ll leave you to it. Thank you so much for your time today. Thank you Brian and ObservePoint for giving me the chance to speak today. I hope you enjoy the rest of the content. I hope you enjoyed my little talk. I hope there’s a lot you can take back with you. If you have any questions later for me, feel free to reach out on twitter or get in touch with Napkyn if you’re not sure how to kickstart that kind of marketing environment product. Cheers! Brian, I’ll send back to you to wrap us up.