Prove your Marketing Worth with Multi-touch, Multi-channel Attribution
[Tim] Thanks for joining us today. I just wanted to give you some quick info before we jump into into our attribution webinar.
Introductions and Housekeeping
My name is Tim. I manage our demand generation here and I work closely with Mark, who will be leading today’s webinar. Throughout the webinar, please send us your questions as we’ll have some time at the end to answer a few of them.
Just submit them through the message box that you see on your screen. We’ll also be asking a few polling questions during the webinar to learn a little bit more about you. So if you can take a quick second to answer those, we’ll leave them up for a few seconds to give you a chance to answer. And a bit more about our talented speaker today, Mark Sullivan has been leading demand generation at CallRail for about five years now.
He’s seen the company grow from a scrappy startup to the almost 200-person company that it is today. He’s full of insight and very passionate about helping companies grow. So certainly feel free to shoot him an email after the webinar if you have any questions. And with that, I’m going to kick it off to mark.
[Mark] Thanks, Tim. Hi, everyone. I am excited to talk about this subject today. I, as Tim mentioned, have been here at CallRail for almost five years and I’ve seen it grow.
And I can tell you that having data and understanding about what is producing outcomes, conversions, and new customers, new revenue for CallRail has proved incredibly insightful and we’ve been able to continue growing at 50% year-over-year for the past several years as a result of that and the other things that we’re doing in the rest of the company. So today’s webinar is a little bit different.
You might be used to webinars where they’re very promotional, kind of promoting products. And we’re actually going to be spending a lot of time educating you all today. This is not really a highly promotional webinar. This is really helping you understand the concepts that will really lead to transformation in your career, if you’re not already using them, when it comes to attribution.
Poll Time: How familiar are you with attribution?
So before I get started, I actually want to run a quick poll. Our first poll question is really just giving me an idea of you all as an audience and how familiar you are with attribution. So let’s launch that poll question. If you can answer this, this really helps me make sure I tailor the entire talk to you all. So how familiar are you with marketing attribution principles? Just trying to get an idea of what levels we have on the line today and really want to understand what people were looking for in attending this webinar.
So we’ll leave it up for just another minute. We’ll leave it up for another second. But one thing I want to say, this is a really challenging subject. So as much as the content we’re going to get into is fairly dense, please, please, please, do not be intimidated by this. I’m going to give you an approach that I think is going to be quite digestible to begin looking at your marketing efforts and your company’s marketing efforts and really understanding them through a lens of multi-touch.
I don’t want anybody to be shied away from the contents we’re going to be going over. It’s not as intimidating as some of the language we’ll use would make it seem. So okay, I think most of you have voted. Let’s close that poll and let’s show the results just so everybody can see. So 75% of you are somewhat familiar. So that’s most of you.
And to be honest, when I started my journey into attribution and really diving into multi-touch, that’s probably how I would’ve answered that question as well. So some of you are not at all familiar and this will be a great primer. Again, don’t be intimidated. And then we have some experts, it looks like, in the audience as well. So definitely feel free to ask questions, any of you all, as we move forward. There will be some time for Q&A at the end if I’m able to get through this dense content quickly, so okay.
Turning attribution into a real life scenario
Moving on, thank you all for answering that, okay. So I’m going to start at the end here. These are my two daughters. That’s my four-year-old feeding my one-year-old. That was taken just a few months ago. This is my beautiful wife, who I met many years ago when I was living out of the country.
Create side by side screenshot
And this is us on our wedding day. And this is probably three months ago. I was at a party and I ran into an old college friend and that college friend, he saw my wife and he saw pictures of my kids and he said, “Gosh, you really turned out okay, Mark. How did you turn from this crazy guy in college and how did you end up like this? How did this crazy person I knew that I didn’t ever think would settle down, how did this person attract such a beautiful fiancé and wife and have kids?”
And really what he was asking is, “How did you meet her?” He was telling me that he’s been dating women and he’s been trying to find a mate and he would love to have a family but he’s having trouble figuring out what he’s doing wrong. So really like embedded in his question, what he’s trying to do is he wants to figure out, “How can I change what I’m doing to have a similar outcome that you’ve had?”
And so he says, you know, “How did you meet her? I really want to know how you met her because I’m not meeting the right people.” So basically, I start diving into it. And I tell him, “Well, after university, when you knew me, I moved to Nicaragua. I lived there for four and a half years and I had joined a Nicaraguan Softball League with a few friends, didn’t know what I was doing. And I partied the night before our second game and was hung over and really hungry. And so I went to a sandwich shop before the game to get food and there is where I saw this pretty girl that I had not seen before in this small surf town that I lived in and I decided to talk to her.”
So that’s what I told him. So I’m going to pause really quick and I’m just going to set some definitions. So we’ll get back to my story in just a moment.
What is an attribution model?
And for the purpose of this presentation, what we’re really really talking about with attribution models is it’s a way to give credit to what causes an outcome, right?
It’s cause and effect. Attribution is used not just in marketing. It’s used in machine learning. It’s used in algorithms. We’re basically trying to determine which set of rules determine the preceding events and what credit they should get for a particular outcome. So in the marketing world, we’re really speaking about conversions. Now, those conversions can happen throughout the funnel, they can be a download of an e-book, for example, at the top, they could be a phone call, little bit lower in the funnel, and they could ultimately be a customer conversion as well.
So really, for the purpose of this webinar, that is what we’re working with, a set of rules to determine which preceding events get credit for a particular outcome. Now, most of us are familiar with attribution models that look like this, first interaction or first-touch and last interaction or last-touch. So basically what these mean, for those of you who are not very familiar at all:
First-touch is really out of all the data points I have on a particular web user or customer, what is the very first interaction that I have in that sort of chronology of data points that I can attribute with credit for bringing that person into my funnel.
And then last interaction is really the last interaction before the outcome that we’re looking to attribute. And I’m going to give 100% in that model, 100% in the first-touch model. So basically, these are single- touch attribution models. And these are the most popular among all marketers and really among a lot of marketing attribution tools that many of us use day in, day out and we’ll talk about some of the problems with those.
Back to the attribution love story
So getting back to my story, let’s take a first-touch attribution model and really use that to determine what my friend should be doing to meet a mate. Essentially, the first data point that he had was that I moved to Nicaragua after university, right? So in a first-touch model, sounds a little silly, but basically, what my friend should do is he should move to Nicaragua.
Now maybe that’s true, maybe he needs to get out of his comfort zone but I’m not sure he needs to move to Nicaragua. So this might seem a little silly but this is…a lot of what we’re using as marketers, many of us are using first-touch attribution models that might be saying the equivalent of move to Nicaragua in my story with my friend. In our last-touch attribution model, I’m going to take the last touch before actually the outcome, meeting my future wife, and I decide to talk to a pretty girl I didn’t know.
So in that model, basically, what it tells my friend is that you need to talk to pretty girls or you need to talk to people, potential mates that you don’t know. Excuse me, I’m drinking some water. I’m recovering from the flu here. I’m trying to avoid coughing to spare you all of the hacking that my family has been exposed to in last few days.
So in the last-touch model, this makes more sense, right? Talk to girls you don’t know. And the reason I bring this up is just because no attribution model is perfect. And I want to make the distinction that it’s less about getting the perfect one, it’s more about understanding the relationship between them.
Poll Question: Who is using first-touch? Who is using last-touch? What other models?
So I think Tim has another poll question that we’re going to go over really quick. And I’m just kind of curious, really, I want to know how many of you are using first-touch, how many of you are using last-touch, and then how many of you are using other attribution models? And if you don’t use anything, just put other. That’s fine. So this is a quick poll question. We’ll show the results again just to share with everybody on the line.
But I’m curious, you know, there’s a lot of surveys out there about usage of models. And I think AdWords puts out the most data with regards to who is using attribution models. And really, we’re sort of boxed in with a lot of these tools with their default attribution model and if we start learning about it, like Google Analytics, for example, we may want to change that attribution model from the default one to one that maybe makes more sense for our business.
So I’ll give you guys a few more seconds to answer this question and then we’ll move on. Very interesting results here that we’re seeing. I think about 90% of you have voted. So let’s close the poll and let’s show the results. So 50% of you are using first-touch.
This is very common. First-touch is a very, very common…It’s actually the default attribution model in a lot of the tools we use. Last-touch is also a pretty common one, especially with PPC advertisers. They’ll use something called last non- direct click or last AdWords click. And so last-touch is a pretty popular one there.
And then other, some of you might be using more advanced models, multi-touch, time decay, or you’re not using anything at all. But this is interesting and this conforms with my understanding. First-touch really is the go-to attribution model. So let’s close that poll and keep moving. Okay, so here’s what I want to ask you.
Which attribution model is better?
In my story, which marketer do you think gets paid more? Which one do you think is more valuable to the company they work for? The one that is using a first-touch model that says to my friend, “You need to move to Nicaragua to find a mate,” the one that says, “Talk to girls you don’t know and you might find a mate.”
And I’d say between these two, it’s fairly intuitive that we’d say the second one, or, and I’m going to give a slightly different version of this. The marketing attribution geek who really understands that we can’t definitively say what is perfect, who says we need to capture more data and move to a multi-touch attribution model but you know what, go ahead and talk to girls you don’t know because that seems to make the most sense.
Embedded within that is really the understanding that nothing is perfect and also that we can make assumptions and deductions from the data that we have even though we don’t have the perfect model. So driving home, it’s never going to be perfect. Don’t worry about getting the perfect attribution model.
What it is really useful for when you start comparing different models is it’s useful for exposing your buyer’s journey and really helping you understand which channels, which efforts, which advertisements are actually bringing people into the funnel and which ones are helping push people over the line. I want to take a moment before we dive into sort of the meat of the presentation to look at a sample report.
And this is a sample report and I’ll orient you really quickly and then I’ll go over some of the metrics in this report. And by the way, we’re going to send out this deck to everyone so don’t feel like you have to actually take in everything you’ll see in this section or in the subsequent sections. We’ll send this out and you can sort of read over it and come to the same conclusions that we’re going to go over rather quickly here.
So essentially, we have the channels on the left hand side. This is a sample report for July 2018. We have our average customer revenue at the top of $2000. And then we have a few columns that… we have a first block of columns that’s first-touch and a second block of columns that’s last-touch. And you’ll see some of the metrics that we’re going to go over. I’m just going to give some definitions real quick.
So here’s some definitions of lead. This is essentially any known person interacting with your brand. In this sample report, that can be someone who fills out a form who basically…they give you their information to tell you who they are. It could be a phone call, an email, phone call. It could be a form-fill, it could be an in-store visit or could even be a trade show interaction where they come to your booth.
A channel is really the source of the lead interaction. Think of it in terms of how did the eyeballs get on my brand that turned into an eventual lead? And then cost per lead, CPL, some of us are familiar with this if we work in PPC. It’s basically how much did you spend during, this would be July of 2018 divided by the number of leads we were able to capture.
Marketing acquisition costs. This is a more advanced metric but I think it’s a fairly good metric for really understanding what your advertising spend is producing at the bottom of the funnel. You can get more advanced and do customer acquisition costs but that includes, like, personnel and other expenses that you may not have access to as a marketer.
So really, this is your advertising spend divided by the number of customers in that time period. And then ROAS. This is something you’ll see in Google Ads, for example, and other platforms. This is a return on ad spend. In order to get the return on ad spend, we have to actually know what our average customer revenue is.
So this calculation is essentially how many customers did we get? So if we got five customers and the average customer revenue is $2000, then 5 times 2000 is $10,000 divided by the amount of spend. So if we spent $2000 in that month that would be 5X return. And so it’s usually displayed as an X after…usually a single or a double digit. So going back to our report, there’s a few questions that I want everybody to think about as we’re looking at this:
- Which channels are performing the best?
- What does first-touch tell us versus last-touch about those channels?
- What decisions would you make based on this report?
This is the most important question for any marketing report, any marketing attribution data.
You really want to be able to make decisions. A report is meaningless if you can’t make decisions based on the report, even if those decisions are we’re going to keep doing what we’re doing. If you can’t make decisions based on the data you’re seeing, then you need to go back to the drawing board and figure out a better report. So think about these as we go over this. I’ve highlighted a few things. This particular format for this presentation on our webinar doesn’t lend itself as perfectly as a conference where I point this out with a laser pointer.
I gave this at SMX a few months ago and was able to point these out. So I’ve highlighted a few things that I just want you to look at. If you look on the first-touch line, take a look at paid search. And particularly, look at the return on ad spend, the last column in the first-touch block. This is the very first highlighted section. And then the last-touch, you’ll see it’s very different on the return on ad spend.
So remember those numbers. Moving down to paid web referral, take a look at that and look at the performance on the first- touch. So in a first-touch model, it’s performing very differently than on the last-touch model. So that’s something I would be curious about. Organic. So organic as a whole, if you look at the actual percentage of total projected revenue or just the projected revenue, it’s very similar to paid in the first-touch.
It’s very different in the last-touch model. And then organic email, not doing much on the first-touch and that would make sense, right? You actually have to have a lead, you have to have their email before you can actually email them. And some of you may be wondering, how is organic email? How is email even showing up on the first-touch? And that’s a good question.
And really, that happens for a lot of different reasons but one of the most common ones is you’re emailing people that have left the company and their email is forwarded to someone else at that company and they fill out a form. And so it looks like it’s a brand new lead coming from that channel and that does happen quite often. But you can see it performs much better in last-touch. So when we think about these questions, when we look at that first-touch model, what the first-touch model really tells us with that sample data is, “Give Google more money.”
Paid search has the highest return. And if I can go back here, paid search has a 5.2X return on the first-touch model. Now, it makes sense if we’re marketers and we’re really trying to prove our value. It makes sense that we might lend ourselves towards basically using a model that only exposes our value but there’s dangers in that too because we’re short-changing investments in organic.
So if I’m a marketer, I’m doing both paid and organic, I’m not only responsible for paid search, for example, then I’m going to be much more interested in both of these models and what they say about the buyer’s journey. Paid search is great at bringing people in at a 5.2X with the first-touch model. It’s not so good at the last-touch model. Organic really closes the loop.
So it’s almost like in basketball. I got basketball in the brain from watching March Madness but if you see a beautiful alley-oop, it’s like paid search is basically passing the ball and getting into position. And then organic is the person who’s actually dunking it in the alley-oop. Sorry for anybody who doesn’t like sports references. It just happened to come up.
So paid web referral, again, is not producing. So paid web referrals, 0X return on the first-touch model, 80X in the last-touch model. That could be something that’s really more of a mid-funnel paid web referral. It could be a listing site where people are researching your company and they see reviews and they see all those positive reviews. And so they’re already sort of… you already know about them.
They’ve already visited your website through other channels but then it really is effective at closing them and making them become a customer or a lead. So that’s another one that I see that’s very common and that looks different in those two models. And then I mentioned this earlier, organic channel revenue in the first-touch model is roughly the same as paid channel revenue and very different in the last-touch model.
And then organic email isn’t producing. We talked about that already. So let’s take a look at the last-touch. What the overarching message is is go all in on organic. So if I were using a last-touch model, I might say paid is not producing anything except for paid web referral and that only cost us, I think it was $300, in the month of July. Why don’t we only do that and nothing else for paid and everything organic?
But what that would be doing is hurting our funnel, right? That would basically be starving at the top of the funnel if we only went in organic. So just to kind of recap what the last touch model shows, paid search has a negative return. It’s actually 0.9X, paid web referrals is producing incredible results, and organic channels are accounting for 82% of the revenue.
Organic email is producing. Let’s keep doing that. So I just want to point out that there are stark differences between these two models and neither one of them are wrong or right. It’s the combination of them that really exposes the buyer’s journey. So if we took a more holistic approach to our marketing and we said, “We want to align our efforts and our investments in our marketing department or on our team according to the buyer’s journey,” we might come up with these conclusions.
And this is where I think you become a much more valuable marketer. Paid channels produce awareness. Paid channels are doing a…they’re doing a great job at bringing people into our funnel, making people aware of us. Organic channels produce lower funnel activity and conversions. So if you start thinking about those efforts and those channels according to the buyer’s journey, you’re going to begin to make optimizations on the content that you’re using for organic knowing that you can maybe surface that CTA a little bit higher in that organic content so you can actually get more of those conversions.
For paid channels, you may want to look at the content that performs the best in those channels to bring people into the fold and then use organic in the mid-funnel as well. And also, paid web referral and organic email accelerate customer acquisition. So they’re really producing a lot of value in the last-touch model. So in that sample report, we essentially started at the finish line.
This is Haile Gebrselassie. He’s one of the best marathon runners, some people think the best marathon runner of all time, and he is famous for how much he trains and he’s an incredible trainer. The reason I put Haile Gebrselassie up here is because I want to make sure that we understand what the finish line looks like.
But we have to put in a lot of work to actually get to the finish line. And meaningful attribution reporting, whether you can make a decision on it, on those reports, it takes a lot of work. Now, here’s what I want to go over. I want to make sure you all understand the concepts and some nitty-gritty about how to accomplish this with whatever tools that you’re working with right now so that you can begin to approach multi-touch with a more holistic view and become that marketer that actually is way more valuable than the average marketer just using one of these attribution models.
So there really are four things that attribution marketers do differently. One, they understand identity resolution. This is one of those phrases that may scare a lot of us off. Don’t be worried about it. I’m going to talk about how to do this. But it is a challenging concept to understand and it’s absolutely critical to understand because if you don’t have identity, you don’t have a narrative glue to tie together all the touchpoints in the previous interactions with your brand.
Attribution marketers also collect the right data. They don’t need all the data in the world to make these decisions. They need a certain amount of data and they collect it consistently for every single interaction. They also define their channels according to their business. So this is really important. A lot of the tools we use, Google Analytics included, they do not have the best channel definitions.
They work maybe 80% of the time. But you’ll see this, if you’ve ever seen in Google Analytics, all the social refer URLs, they don’t show up correctly. That’s because the stock definitions in Google Analytics will only count linkedin.com, for example, or it may catch some of the other ones but it won’t catch the shortened URLs or t.cos, the Twitter one. Google doesn’t have really a vested interest in making sure that those platforms are attributed because they want the money in their world.
So the best marketers, attribution marketers, actually define their channels really well with custom definitions. And then the last thing they do differently is really, they produce these reports, similar to what we just went over. They factor in the money for the reporting. They factor in both how much are we spending and how much are we getting back. And sometimes, you can’t know the exact customer amounts.
So we’ll talk about how to get an average customer revenue figure that makes sense for your business. So let’s talk about identity resolution. So here’s the challenge with identity resolution. The challenge is really tying together anonymous web visits with lead acquisition events. So those phone calls, those form fills, any way that someone becomes a lead. So tying number one to number two, and then number two to number three, number three being customer acquisition.
This is the challenge. If we don’t actually resolve the identity for all these interactions that are happening, the data is meaningless. And so most of the tools we use, they actually do this with cookies. So they do this for us. Cookies are essentially that they’re stored in our browser cache. And so if I visit a website one day and it’s the first time I visit that website, it actually assigns a unique user ID, user cookie to my browser.
So when I come back the next day, it actually knows, it stores the information about who I am with that user ID, that number. They don’t know who I am in terms of my identity yet but they know that I’m the same person because there’s an identifier in my browser. And then if you go to the lead stage… Excuse me. The lead stage, really, this can be an email address, that’s the most common, a phone number as well.
A lot of our customers here at CallRail that are using us for call tracking and analytics, they’ll use a unique identifier of a caller ID for their inbound phone call leads. And they’ll use that until they get an email address. And so it’s really important. The jump between number one and number two is the hardest to get right because of some limitations with tools that we’ll talk about in a moment.
But basically, you want to be able to tie anonymous web traffic so you can get all that chronology of interactions. The first-touch, you can’t get the first-touch unless you know before they ever became a lead, how did they come to your website. So that is the biggest jump from one to two. And then the customer acquisition, this is really like, depending on the complexity of your company and your marketing team, your marketing department, but you really want to try to tie the actual transaction amount or revenue amount to the person that purchased it.
That can get a little bit complicated with multiple people being a part of the same company. If you’re selling something like a B2B SaaS application or even any buying process where multiple people are involved. But we’re really going to focus on one and two. So I just want to hammer home how important it is to understand this. Don’t worry about making sure that you can develop all the solutions that I’ll talk about to solve for these challenges.
But you need to understand that identity really is the narrative glue for marketing attribution. The only way to tie these visits together is to have some unique identifier and a change as you move down the funnel. So the good news is most analytics tools take care of anonymous stage identity resolution. Google Analytics does this with a GA client ID, for example. The bad news, however, is that most analytics tools do not have access to lead or customer data.
And Google Analytics actually forbids having personally identifiable information in Google Analytics and they’ve been cracking down this more since some changes in email regulations called GDPR that went out last May. So it’s really important to not pump information into Google Analytics if it’s identifiable to a person, so no names, no full caller IDs, or anything like that.
But don’t freak out. Do this instead. Really, what you need to be doing is you need to begin…pick an analytics provider that assigns this, let’s say Google Analytics, for example, the GA client ID. And the GA client ID, you can actually capture that with hidden form fields on form submissions.
So basically, leverage the identity resolution of other tools that are already doing it and they’re doing it quite well and capture that data so that you can basically pull in that first-touch and other interaction data associated with that user ID. So if you have a form submission on your site, you can actually set up with most form providers, a hidden field to actually capture something that’s either in a cookie like the Google Analytics cookie or any other tool you’re using.
You can have it capture a user ID in a hidden field. For phone calls, like here at CallRail, you can actually use custom cookie capture. So the way that call tracking works, you can actually track calls at the session level. So if you have session level call tracking, it basically means that you’re basically getting all the data that Google Analytics would have for a particular phone call.
And you can see how they came in, the visitor timeline, all that stuff that you can set up custom cookie capture. All you got to do is give it the name of the cookie you want to capture and you can begin capturing that user ID as well. So you can do it with phone calls. And then with live chat, if any of you use that on your websites, you can actually integrate usually directly with your analytics tool. If you can’t, you would have to probably involve some development resources. This sounds more complicated than it is but hidden fields are very, very common.
Most form providers have an entire support page on it so don’t be intimidated by it. The second thing I want to go over that attribution marketers do differently is they collect the right data. What do I mean by that? Well, let’s talk about the marketing touchpoint for a moment. This is really a foundational concept of attribution. The marketing touchpoint is essentially any interaction with your brand that we can reasonably track as marketers.
So think back to my story, when I was giving all the data points about how I met my wife, that’s all we have. Let’s stop there. All the data that we have to work with is what we’re talking about here. So those are all touchpoints, right, any interaction with your brand that you can reasonably track. So it’s important to collect the right data for your funnel.
So for example, you may be using a combination of different efforts to produce a conversion, display, email, social, organic search, paid search. You really need to understand what you’re doing and then collect the data for each of those. Make sure you look at all the calls to action and that you’re collecting data according to where they came from.
So the minimum data points you actually need and there’s only three that you need, to begin getting reports like the sample report we went over, you absolutely have to have a timestamp, you need a referring URL, and you need a landing page URL, you need the full string. So let’s back up a second. So I am visiting a website and let’s say I make a phone call.
What I need to have as a marketer to attribute that interaction, the… like the phone call event and the visit that produced it, I need a timestamp, referring URL…timestamp for the web visit, a referring URL, and the landing page URL. The reason you need the referring URL is because you want to know where they’re coming from.
Now, for some channels like email, it doesn’t often…if someone is using a client, a desktop email client like Outlook, it may not share any referring URL. It may just show up as a blank referring URL. But if you’re tagging your emails with UTM parameters, which I’m not going to get into in this presentation, but basically, they’re little things appended at the end, little strings appended at the end of a URL.
It doesn’t change where you’re going on a website but it actually gives you data as a marketer when you read the web logs essentially of people visiting your site. If you’re tagging those links and those emails correctly, then you can actually use the landing page URL with the full string of it to suss out, was this an email campaign or was this really a direct visit with no referring information at all?
I mentioned two other data points here that I call them bonuses but I think as you get into this more, you’re going to want more data and more dimensions to attribute. And the first one is really what is the call to action or the offer type? So this is really the content dimension. Is it a phone call? Is it a form?
And even on the form side, is it a demo request? Is it an e-book or is it an appointment booking? Basically, like if you start capturing that data in addition to the stuff above, you can actually produce reports similar to what we saw earlier where you’re replacing the channels with a different dimension like, you know, are people coming in through phone calls or our demo request or appointment bookings?
And you can look at first-touch and last-touch in different models in those as well. So the more data you end up capturing over time, the more you can actually attribute and produce reports that are really meaningful like that. And then the form page URL can be really helpful. This is kind of the page…and it could be the page URL that they were on when they called if you’re using a tool like CallRail.
But basically, what was the version of the offer type that actually produced the result? I mean that can be helpful if you have, you know, if you’re doing appointment booking and you might have several different places on your website where you have an appointment booking widget. You might want to know what piece of content actually produced that appointment booking.
So that’s an important one as well. And by the way, I’m just going to stop for a second. If you have any questions about this stuff as we’re going through it, I realize this is dense and we’re going through it quickly, so please save those or put them in…actually, I think you can put them right in, right, Tim?
Okay. Yeah, you can put them right in and we’ll get to some of the questions towards the end. So I just want to remind everybody to ask questions. So on collecting the right data, don’t freak out. Do this instead. So for anonymous first-touches, this is probably the most complex part of the entire presentation that you’ll probably have to involve a developer if you’re not familiar with this.
Now, you can ping Google Analytics’ API and you can get this stuff. So there’s really two ways to do it. You can either set a cookie on your own to capture this data that you can pull out and put into a lead capture hidden field or you can basically just get the GA client ID in the hidden field and then use something on the backend to get that…to backfill that data to understand where they came from originally.
So on the lead capture touches, retrieve the first-touch values from the cookie you set up, so that’s if you did the step above, and then pass them in the hidden fields. And then the other one that I mentioned is basically get the GA client ID on the lead capture events, if you’re using Google Analytics, and then use Google Analytics’ API, excuse me, to pull out the very first-touch information from Google Analytics that you can get.
So for other touches, you really need to understand how much data you need for your attribution model. And then so you need to think about this critically and set up to collect, retrieve, and pass those as needed. One of the questions I get sometimes is how do you attribute something like a trade show or how do you attribute something like a trade show, many of us might not be doing but maybe an event like an offline event?
There’s no way to initially tie a GA client ID or some unique identifier that we’re using from an analytics tool to someone we meet at a local networking event or some trade show we’re doing. But if you get the email address at that trade show or at that local networking event, what you can essentially do is as soon as they make an appointment on your website or they fill out some form where you have their email address and you’re capturing that unique identifier, you actually can backfill all the touches previously as long as you’re getting the unique identifier you need.
Feel free to ask any questions about that but that’s a question that I get quite often. It is possible but it’s only possible after you have both of those events happen. So marrying an offline event with an online event where you have something like an email address that marries the two. So the third thing that attribution marketers do differently is they actually define their channels.
Every business is different and it takes some work to define your channels. And when I talk about channels, I want to make it clear that I’m really talking about how eyeballs get on your brand. And more specifically, I’m talking about sort of bounded marketing efforts and investments.
This is a perfect example. Here at CallRail, we invest a lot in making sure that customers’ voices, our customers’ voices, are present on platforms like, I call them Yelp for software platforms, basically the G2 Crowd and Capterra. We really want to encourage our customers.
So we send all of our customers invites to leave their feedback on our platform so the rest of world can see. So we put a lot of investment in there and so that really is a marketing effort that is…it’s very different from the other things that we’re doing and if I didn’t define that as a specific channel, it would show up just as a web referral, right?
So people coming from G2 Crowd on the CallRail page clicking over to our website. If I didn’t define that channel for our business here at CallRail, it would just look like referrals are fantastic. But that’s actually a very specific type of referral. So I want to stress that the best marketers, they actually take time to define, with logic, the channels that are important for them to track because they want to understand what’s working and what’s not.
I’ll go into some details here and then we’ll move on but you want to pick probably a broad channel. So in the examples I gave earlier, we had paid search and then your sub-channel might be Google ads, for example. And then if you want to go a third level, I wouldn’t go much more than three levels unless you just have a crazy amount of volume and crazy amount of channels. You could add something like this was a specific campaign or a geographic region.
But you really want to try to stick to two levels at most, and for some of you, if you can keep it simple and just do one channel, then do that. And then you want to define each of those channels. You want to basically set the definitions. So you use the touchpoint data that you captured in step three, or was that step two that we just talked about?
And you want to use that data to categorize each touchpoint and the channel that produced it. So for example, if a referring URL contains Facebook and the landing page has paid in the UTMs and the UTM parameters, then the touchpoint channel might be paid social, Facebook. So that is an example of a definition, channel and sub-channel definition. And these definitions, you can put into any analytics tool that you’re using, Google Analytics being the most popular.
You can set up these definitions to make sure that you’re getting a channel grouping that makes more sense. So again, don’t be intimidated. Do these instead. Analytics tools, like I mentioned, have decent predefined channels. Use them if they work, customize them if they don’t.
If you’re investing heavily in social and you’re using Google Analytics for a lot of this, you need to invest…you need to look at all your referrals and see what Google Analytics is lumping in with referrals that you know is social. Google Analytics also does not do paid search very well. So if you’re doing like Search Network campaigns, it’ll capture like the big ones.
It’ll capture its own platform, obviously, Google ads. It will also capture and categorize correctly Microsoft Ads (formerly Bing ads). But if you’re doing sort of Search Network on Microsoft and even on Google, you might see some search providers in your referrals that really you know are paid search and they’re coming from your investments there. So make sure that you take time to define those and use the ones that work to short circuit some of your workload.
And then your business likely has the unique channel like the listings that I talked about earlier, the review sites. Those are really important to us. We put a lot of effort in trying to invite our customers to leave reviews there. So that was an important one to define for our business. And there’s several more I could go into but think about your business and figure out which ones those are.
And don’t forget about offline and unpaid channels. I think that there’s a tendency to only focus on paid ones but they’re going to be efforts that you’re investing in time and energy that may not have a line item in your advertising budget that you’re going to want to track. So make sure you don’t forget about those. And then don’t have too many, especially at the broad level. It gets really hard to maintain the definitions.
And I can tell you this as someone who… we have like a sheet of I think it’s 89 definitions for all of our channels and maintaining that becomes really, really challenging. So don’t… Only include ones if you absolutely need to. And the last thing that attribution marketers do differently that I want to talk about today is really they factor money into the reporting. So they use everything that we’ve talked about in the first three points and then they factor in money.
They really make it meaningful, right? So the way you do this is fairly simple. You want to track spend at the channel level and I’ll show you in a moment, an example, sort of template and I can show you this, share the templates with you if you want to see those and use them for your business. But you want to track spend by channel and just aggregate the spend monthly. Don’t get super complicated with the date ranges.
Just pick monthly and say, “How much did we spend on paid search Google ads this month?” And then track that spend so that you have that spend for that month. If you try to get too fancy and granular, you might have a hard time aggregating and getting a report that’s meaningful in the end. So start broad and then get specific as needed. Don’t forget offline channels.
You’ll want to sort of define those as well. Often those are paid events or events in general, networking events and trade shows and stuff like that. Determine your average customer revenue. So for those of you who don’t have access to actual transaction amounts at your company or the actual total customer revenue, you can get an estimate. You can talk around and talk to different people.
If you’re an agency, for example, and you’re working, and you have clients, get them to give it to you. Because once they give it to you, you say, “Okay, this is what I’m going to use in the reporting.Does this make sense?” “Yeah, I’d say like that makes sense.We get about $2,000 per customer.” “Okay, fantastic.That’s what I’m going to use.” And if you get it from the people that you’re reporting to, like the consumers, the report, then if there’s issues with it, then it came from them.
And if they don’t like what they’re seeing or they don’t believe it, then you can have that conversation. You say, “Well, I used what you gave me.” So it’s important to sort of get some proxy metric for average customer revenue to be able to calculate the projected revenue you’re producing. And then you use this data, the cost data, the spend data for cost per lead, for marketing acquisition cost, and then you use the actual average customer revenue data for the projected revenue and the return on ad spend.
So this is an example of basically a sample data showing you how I built a tracker here at CallRail. I think I built this about a year ago but this works great. And I have a budget there. And this is not actual data for us. This was just sample data.
But it basically…I have it set up to show you when you’re over or above budget and it kind of helps us plan as well when we use sheets like this. So I’m happy to share that with anyone. So tools to use, obviously, Google Sheets is free, WordPress plugins, there are some free and some paid ones that are pretty good. I tend to default to Google Sheets to sort of kind of prototype what I want and then see if there’s any tools that fit what I want.
Supermetrics is something that’s…it’s a fantastic tool. If you don’t know about it, it’s a great way to connect to a bunch of different ad platforms. They maintain connections with all the major ad platforms and even some that are not major and I’m kind of surprised at their growing list. And it’s fairly inexpensive and you can basically automatically import data into Google Sheets with their Google Sheets add-on.
So you don’t even have to think about it. It can refresh daily even. And then CallRail. So obviously, this is the very tiny promotional segment. But basically, we’re trying to solve this for you guys. And I can’t speak to exactly what we’re about to release or building to help in regards to multi-touch attribution, managing spend, defining channels, but I can say that we’re very interested and we’re aware of this problem.
We’re acutely aware. I mean, this is a challenge for large companies as well, really trying to attribute proceeding interactions to ultimate outcomes is something that you might think big companies have figured out or medium-sized companies but it’s really a challenge. And so we’re working on tools to release in the near future that are going to go after this and really solve it at a really affordable price point.
So stay tuned and make sure to keep getting our emails because there’ll be an announcement in the next two months on that. Bizible. So Bizible is a solution and they were just acquired by Marketo a few months ago and they range around $1500 to $3000 a month.
There’s no way to sort of upload revenue data. But it is a good solution if you’re using Microsoft Dynamics CRM or Salesforce. So just recapping, I think we’re getting close to the end here. We’re going to get to your questions. These are the four things you need to do.