Marketing Analytics Fundamentals | Joel Witten | Skillshare

Marketing Analytics Fundamentals

Joel Witten, Data Science Manager at Facebok

Play Speed
  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x
10 Lessons (41m)
    • 1. Introduction

    • 2. Billy learns about Soks

    • 3. Overview of Marketing Attribution

    • 4. Tracking Technology

    • 5. Deep Dive into UTM Codes

    • 6. UTM Codes with Google Analytics

    • 7. Attribution Methodologies

    • 8. Evaluating Marketing Campaigns

    • 9. Project Overview

    • 10. That's a Wrap!

20 students are watching this class

About This Class

Use your marketing budget wisely. Make smart business decisions.

The barrier to entry to market your product has never been easier. Companies such as Google and Facebook have created online platforms to allow you to target who you want, where you want, and when you want. In order to make the most of these platforms, you need to learn the basics about digital marketing, and how to evaluate how well various campaigns are performing. 

In this class you'll:

  • Learn about marketing analytics technologies
  • Deep dive into structuring URLs with UTM codes to better track your campaigns
  • Various attribution methodologies
  • Practice setting goals and learning about tracking in Google Analytics
  • Dive into quantitatively evaluating marketing campaigns

This course will give you an overview of what is out there, and so you can better understand what you don't know. You'll leave understanding some of marketing lingo, and you'll be on your way to become a modern marketing manager.


1. Introduction: Hey everybody and welcome to Marketing Analytics Fundamentals. My name is Joel and I'll be your instructor for today. A little bit about me. I head up the data engineering and analytics team that Venmo, where we help the product team, the operations team, etc, make data-driven decisions and help move the company forward. Before that, I was a data scientists at Squarespace working on lifetime value and marketing attribution, as well as helping optimize customer service and a lot of other data-driven tasks. I've also taught marketing statistics at NYU and Data Science at General Assembly. Right now, at SkillShare. What I've learned throughout my time is that companies are really becoming data-driven. Every single action and moment that a user has on an app or a website, is trapped and used by various product and design, and operation seems to optimize the product and make the experience as good and fluid as possible. Marketing analytics is really no exception. The modern marketing manager is really tactical. They understand technology and understand how to optimize various marketing channels. Throughout this class, we're going to learn a lot about the basic technology. We're going to learn about implementing some of the modern features in marketing attribution. We'll walk through some real-world examples and they'll be an interesting project to the end that you can go and practice your analytic sales. Today's curriculum, we're going to start off with a marketing attribution example. We'll talk about the history and evolution of marketing attribution, we'll go through some common tracking technologies, we'll walk-through UTM codes a little bit deeper, we're going to demonstrate the implementation of UTM codes through Google Analytics, we'll talk about some of the attribution methodologies that are common to nowadays, and finally, will evaluate campaigns and tie it all together. Let's get started. 2. Billy learns about Soks: Soks Inc. is a company that sells socks, let's say $10 a month. It's subscription service and you get socks every month. You'll keep getting socks until eventually you call them and you quit or you just you cancel your credit card or your credit card fails or something along those lines. It's a typical subscription company. Now, Soks Inc. has been has been grown over time, mostly through referrals. But they want to step up their game. They're not growing fast enough. They want to do is they want to start launching various marketing campaigns, and they did. Imagine Billy is a customer who is a potential customer is completely unaware of Soks Inc. and he's surfing the web, minding his own business. On one of the websites he typically visits, he sees a banner ad for Soks Inc. He's never heard of soks Inc and he doesn't know what the company is. He doesn't even necessarily click on the ad. He just sees that Soks Inc. Is a company and it likely sell socks. Now what this Ad does is it prime's Billy, he's now aware of the company and awareness is the first step in a customer journey to eventually purchasing the product. Now let's fast forward. Billy is you know, sitting back at his computer, he's looking to buy some socks and has completely forgotten about Soks Inc. and at the very top he sees an ad for socks, and because he's already been primed by the banner ad, he clicks it and purchases. Now on the flip side, Sandy is the Marketing Manager. She's got various campaigns running and she wants to understand their performance. She has just spent $10 thousand on Google AdWords. She has spent $5 thousand on podcasts, and she has spent $2.5 thousand on banner ads. What she's trying to do now is understand which of these campaigns actually did well and which ones didn't do well? Let's find the ones that did well. We'll put more money for them and the ones that didn't hopefully either make them more efficient or cut them out. But the question for Sandy is, how many users did Google AdWords acquire. Well first did she, did Google AdWords acquire Billy? Well, you could argue that, well, since Google AdWords was the last place Billy clicked on before he ended up purchasing you could say yeah, that AdWords acquired Billy. Though you could also say that the banner ads acquired Billy because if he had and seen the banner ad he would have never known about the company and the one that Google ad had popped up, he probably wouldn't have clicked it, so Sandy has to make a decision around where the attribution goes for Billy, should he go towards AdWords as you go towards banner ads. These types of decisions are really important because they help dictate which of your channels are doing well and which of them are under performing. 3. Overview of Marketing Attribution: Let's talk about marketing attribution. The purpose of marketing attribution is to understand how effective various marketing campaigns are, and to sign a quantitative number to measure its performance. As a marketing manager, you may have dozens of campaigns, across multiple channels and you want to make sure that the ones that aren't performing get rid of quickly and the ones that are performing, perhaps you double down and spend more money on. As a manager, you only have so much time to spend. You may as well be spending it on the channels that are performing the best. Before we dive into marketing attribution, let's go through a little bit of the history. Since the early 1990's, marketing performance is measured through something called marketing mix models. What is a marketing mix model? Well, it is a historical sales data, as well as marketing spent to try and understand the lift attributed to marketing. Let's walk through an example, you've got sign-ups on the bottom left, and you can see that sign-ups going up and down, it looks like they go up at the end beginning of the month, and they come back down and up and down, and on the right-hand side is a chart of how much money you spent on marketing. You can see that really you haven't spent anything since early February through mid March through mid April. You're spending some money on marketing and the marketing spent drops back down. A question that you might want to answer is, well, how much lift in daily sign-ups was there as a result of the money that is spent in marketing. You could break your marketing spend down even further. Like maybe I spent some money on TV and I spent some money on radio and I spent some money on billboards, and when you throw it into marketing mix model, which is a really complicated statistical or linear model, it'll, spit out how much it thinks each channel contributed to the daily sign-ups. Out of the marketing mix model, you might get TV spend increased sign ups by 10 thousand users. Radio spend increased sign ups by 5 thousand users and billboards had 1000 user impact. The pros of marketing mix models and they give you a great top-down view. They're especially good for non-digital channels, such as TV, radio, etc. The downside of marketing mix models first other very complicated. It's not readily accessible to a just a general analyst or marketing manager. You often have to hire consultants or external vendors and use their products. But another downside is you can't attribute individual users to various channels. We don't know which users came in through TV or which users came in through radio. A final downside of marketing mix models is that you can't really attribute lift to channel so you don't spend a lot of money out. If I'm spending a $100 thousand on radio and $1 million on TV and I want to try out $10 thousand on online ads, your marketing mix models not going to pick up that impact. There are certainly pros and cons, but it hasn't quite a complicated model to wield. On the other hand, you've got digital attribution. Digital attribution is when you attribute user acquisition to specific channels, often through the use of device data URLs. Looking at where did users come from? What websites where they coming from? What is their URL when they landed on your page? What ads they see? How many times they visit your page? Using all this information to better understand your customer and better understand which of these factors helps predict customer conversion. Which in the end is a customer making a purchase on your page or maybe signing up for something, that's what customer conversion is. What are the pros and cons? But one of the pros are, with digital attribution, you can really easily identify performance of digital channels. If I wanted to spend now my $100, my $1000 on Google AdWords. I can do that with digital attribution because I know exactly which users clicked in through my AdWords. I know it was 12, something like that versus in a marketing mix model, you'd never be able to see that impact. There is a downside of course, to digital attribution is that you can't really apply this to offline channels like radio or TV or taxis. You've got to use something more traditional and that's often marketing mix models. Now why is digital attributions so important, we'll take a look at this slide. This slide says that internet is one third of all advertising. That means in 2016, of the $500 billion spent on ads, about 170 billion was spent on the internet, which is huge, and take a look at internet and advertising spending over time. We see that print was really big in the 1980's, even bigger than TV. Over time, it evened out with TV in 2000. It looks like it was more or less the same. But since 2006, internet advertising has been eating away at print advertising, and the reason is pretty simple. When you are a marketing manager and you have a $100 thousand to spend, and you're trying to think, should I spend this on trackable media like YouTube or Twitter or Facebook or Google? Or should I pant and put this in some sort of magazine and hope that the users who see the ad in a magazine will then go and purchase something on my website. I'll never be able to track. I don't know who even saw the ad and I don't know if any of those people went on to purchase, but you really need to have a deep belief that print ads are working. As a marketing manager, it's a lot safer play to just say, let me put it on the internet. In addition, if you're a small company, you've got $1000 to spend. It's really hard to get into print advertising or it's hard to buy a single spot, you might as well spend it on the internet. There's really no barrier to entry. Of the $170 billion spent on internet advertising, Google and Facebook, are two-thirds of it and that's not particularly surprising either. Google and Facebook have really great information about you. Facebook understands who your friends are, what you're looking at on the feed, what's your favorite movies are, food and preferences and Google because you're probably using Chrome, knows what websites you visit and what you're searching for and can really track you throughout your entire internet experience. These companies have done is taken that information, opened it up to advertisers and allowed companies, small business owners to target specific demographics of users, allowing them to have really powerful marketing for quite a low barrier to entry. 4. Tracking Technology: So now we'll talk about some of the various tracking technologies. Some of the problems that they're trying to solve, and some of the issues. Of all the tracking techniques that we'll talk about, cookies are the ones that you're probably most familiar with. Cookies are small text files, they're placed on your computer by your browser, and they store some generally non-personal information about you. They'll store things like, how did you get to a specific website? Where did you come from? How many page views did you see? It generally tracks your progress throughout the internet. These cookies are usually anonymous and they're often encrypted. It's not like anybody can go away and see what you did. These are really useful for companies to understand. I heard this user made a purchase. They visited twice before, once from an ad, once not from an ad. Cool. So we know that they came from an ad as your first click. Another tracking technique is a device ID. Device ID are generally mobile specific. Apple has one and it's called an IDFA and Android has one called an Android advertising ID. The device IDs are used by apps to understand when users do certain actions. In apps where you don't need to login and you clicked around, the app is tracking what you're doing, but they're tracking it using your IDFA. So they know that later when you load the app backup, it's the same IDFA, it means it's the same person. Lastly, UTM codes. UTM codes have been around for awhile since about 2005. These are the descriptors that you can add into URLs, to understand where users are coming from. Sometimes you might receive an e-mail and it'll say, "Hey, click here on this blog", and you'll click on the blog and then the URL will be really, really long. There will be a lot of percentages and underscores and what this URL is doing is it's telling the company, where this user's coming from. This user is coming from an e-mail clicking on this link to this blog, and it helps a company understand, how users are coming to their website. Now, there are still some issues with the tracking technologies we described. The biggest one is that they're all device specific. The device ID's just for the mobile, the cookie is usually just for your browser. So if you're using a browser on your mobile phone, it doesn't sync up to your browser on your desktop and the UTM codes are also just for your specific to your device. If you for example, wake up in the morning and you check your Facebook and you see an ad you click on it but later in the day you go and purchase from this company, but on your desktop, technically the company doesn't know that you saw the ad and then made a purchase, so they might think that the ad wasn't working, even though in reality it was. Some companies are trying to solve this problem. The two biggest ones are Google and Facebook, because you're always logged in with Facebook on your mobile phone and probably on your desktop and you're always logged into your Gmail on your mobile phone and your desktop, and you're probably even using Google Chrome. Google Chrome can track you say with Facebook across your devices, cross device attribution. This is really useful for marketers, so they can have mobile advertising and they can also have desktop advertising tied up together and understand the efficiency about both devices. 5. Deep Dive into UTM Codes: All right. In the previous section we talked a little about various tracking technologies, but the one we're going to spend the most time focusing on are UTM codes. UTM codes and URLs allow you break up various marketing initiatives and track how well certain links are bringing people to your app or your site. A lot of times e-mails you receive from various from companies around content marketing or promotions or whatnot, they rely really heavily on UTM codes in their URLs to track you. On the right is an example of an e-mail I received from the Gothamist, I receive this daily, it's a newsletter about what's going on in New York City, and when I clicked on their image, the URL below is what I ended up going to, so what does this mean? What's going on here? We're going to talk about this a little bit later in the section. We'll talk about how to structure UTM codes, but in general, you might have various marketing initiatives that you've launched. You've launched some advertising in Google and you've launched some advertising on Facebook, maybe you've partnered with a site, and they've sponsored a blog post by you or something along those lines, you have various campaigns that you've done, and you want to track how efficient they are. The best way to track how efficient they are, is you wanted to see how many people are landing on your website from the various links, how many times are people clicking on ads and get into your website or how many people are clicking on the sponsored posts. The UTM code is what breaks up the URL, and when you land behind the scenes, Google analytics or whatever analytics tool you're using, takes this URL, breaks it up and stores it in an easy to understand manner. What are the ways to actually structure UTM code? Well, you take your campaign and you have to break it up into three to five sections. First, you need to use the campaign source, and the campaign source is the highest level descriptor of the campaign, so are people coming from Google? Are people coming from Facebook? Perhaps from Spotify or online radio, you want to get pretty specific from where users are finding your website or app. The next is the medium, what are they seeing? Are they seeing a display ad? Are they seeing an e-mail that they're clicking through? Are they going on Google AdWords like cost-per-click or Bing or something like that? There's where you want to describe the medium through which the users are finding your website and clicking on this link. The next important one is the campaign name. This is the high-level campaign descriptor. You might have various marketing campaigns on Google. For example, if I have one, maybe I'll call it marketing class on Google, but then maybe one day I'll launch an analytics course and I'll call it analytic class, and one day I launch a Photoshop class, that campaign could be Photoshop class. Those are the three required and recommended UTM codes that you input into your URL, and we'll go through how to exactly input them, but those are the three main ones. You can get even more granular. Perhaps marketing class, I actually want to bid on specific words. When you search for online classes, I want that to come up. I can add online class to the URL as well, so if you search for online class on Google and an ad came up that's specific to online classes. When you clicked in it, I would know that you came through an online class in my marketing class campaign. Now, the last one is campaign content, also not really necessary, but if you want to get really good detail. Maybe when you search for online classes, I actually have various ads, so some of my ads might say sign up for classes or another ad might say free classes. How do I know which ads specifically is doing better? Let say to do that, I could add even more detail to my URL, and one of them could have a campaign content of sign up for classes and other could have free class, and I can know which one brings users in better and which one converts users better. Let's go through an example. You head up marketing for socks and you want to start advertising on Google AdWords. You've launched a new line of extra warm socks that you want to sell. What happens is when someone searches for the word warm socks, you want an ad to come up. Let's think about this. For just for a moment, I'm going to pause here, but what would you call the UTM source? If you say Google, that's correct because we're doing an AdWords, people are coming through Google and source is really the high level descriptor of where people are coming from. Next is the medium. Well, what is the medium? CPC would be a common medium to use, cost-per-click because it's Google AdWords, every time a user clicks, you get charged, so it's a cost-per-click and that's a common medium to use for these types of bidding systems. What about a UTM campaign? The campaign is extra warm socks. Because this is your new product, you want to generally name your campaign after the new product that you're launching, so if your next product is going to be shirts, you might want to call your UTM campaign the shirt campaign, but because this is your product, extra warm socks, that's what we're launching, that's what you want to name your campaign because that's what you want users to eventually go and buy. How about the term, warm socks? Because we said when we users search for the word warm socks, we want certain ads to come up and drive people to buy the extra warm socks, and finally the content, what should we use for content? That is correct, buy some really comfortable socks, because essentially this is the ad text that's going to pop up, and we want to send a conversion. Now, let's review that URL that I brought up earlier. This is the Gothamist, I clicked on the Gothamist and this is the URL that it took me to. Now, we can dig in a little and see that they're using the UTM parameters in their URL, so I bolded where it says UTM medium, UTM campaign content, term and source, the percentage twenties in that URL, they're pretty confusing, it's just a space. Here I replace the percentage twenties with underscores, so we can better read what it said. Let's go through. What source did the Gothamist use in this URL? Their source is CM, which very likely means content marketing, and that makes sense, I receive this on a daily basis, it's probably content marketing. How about the medium? The medium, we can see is e-mail, and that's correct, I receive an e-mail. The campaign, let's look at the campaign, this is daily Gothamist, daily winter storm warning in effect, as New York City may see 6-12 inches of snow, possibly more. That's a really strange one. Let's keep moving on. How about the term? The term is Brooklyn Heights tenants fear landlord excavation, and that's exactly the article I clicked on, and finally, let's go to the content. It's also daily Gothamist daily winter storm warning in effect is New York City may see it 6-12 inches of snow. The Gothamist is seems to be doing something a little bit weird here, they're using their source properly, they're using their medium properly, and they using the term properly, but it looks like they're improperly using their campaign and the content because they're using links from a long time ago, 6-12 inches of snow was many months ago, and the winter, it's not the winter now, so something is going on in the URLs and every single article you click on, takes you to the same one. What's important when you build UTM parameters? The most important thing is to say can stay consistent. If you have e-mails, make sure the medium is all e-mails, if you're using content marketing, make sure the source is content marketing, and make sure that the other parameters, the campaign, the content parameters are properly set because this could lead to really confusing analysis in the future. 6. UTM Codes with Google Analytics: How do you actually use UTM codes in real life? Let's say you own a small business or you're the marketing manager of a company, how do you actually implement UTM codes and track them and understand conversion and all that stuff. Well, you use Google Analytics. That's the best tool to start off with. In Google Analytics, it's really cool. If you ever create UTM URL with all of the UTM source and you campaigns and whatnot and you go on a website, Google Analytics will automatically break up the source, the campaign, the medium, the content, all that stuff in its UI. All you need to do is navigate into there, click on "All Campaigns", and then you can see the various campaigns automatically populate without you having to do anything. All you need to do in order to make this live is create a URL and put it in the wild, put on an ad or put it on a blog, or put it in the e-mail and you'll be able to track how many users are coming in through this URL. Of course, just learning on the website isn't everything. You need to be able to add goals. Once the user learns, what should they do? Do you want them to purchase something or do you want them sign up for a newsfeed or do you want them to just land on a specific page and navigate somewhere? With Google Analytics, you can actually build in this goal and you can track this goal across the various campaigns. Let's go through an example, where on my website with, I want to see how people land on it and then navigate to the page where they can submit their e-mail address. Something really basic, but let's start off with something like that and I'll show you all how to do it. This is my website for Lucida, which is our analytics course. Let's say I have a URL right under the description of our class and I want to understand how many people are coming from SkillShare to my website with What I want to do is go into Google's campaign URL builder. Their campaign URL builder lets you type in your website URL, lets say you type in your campaign sources of SkillShare, lets you type in the medium. In this case, I'm going to say PR. Why not? Lets you type in your campaign name, which is marketing analytics and that's it. Then on the bottom, you get a code that's generated. You can copy the URL, you put it wherever you want and then users can land on it. Here I'm landing a great, takes me the same website but in addition, Google now has information that, hey, I came from SkillShare for the UTM campaign marketing analytics. Great. How do I actually go in and visualize and see how many people I'm getting from all of my campaigns. I don't actually have a lot of campaigns running, but let's take a look anyway. In Google Analytics, you can scroll to Acquisition. You could then go to Campaigns right here, you click "All Campaigns". I scroll to the bottom, I see, cool and automatically Marketing Analytics is populated. I can break it down by source, you can look at this from SkillShare and I can go back to the Marketing Analytics and then you see a lot of information about this. I'm not going to go through what everything means here. But some of the important metrics are the number of sessions and especially your goals, your conversions. You can set different goals because what we're interested is not really how many people are landing on the website. That's meaningless unless they actually do something, is how many people eventually convert to it goal. Let's work on setting up a goal. Imagine my only goal. I want people to come here and then I want them to stay updated, fill out the information and click "Submit". I just want the people to get this page, which is with Google Analytics actually allows you to really easily build in goals. I go into the admin panel, click on "Goals". I could then say "New Goal". What's my goal? Is it a template, making a payment registering? No. Is it creating account? No, is it engagement? No. Instead it's something custom. The goal description is clicking "Update Me". Great. It's destination, it's the URL that I want people to go to. Click "Continue" and let's get the goal details. The goal details is the page that I want users to visit. It's this page right here, "Updates". I'll paste that into here and I want to assign a monetary value to this as well. I'll just make it more interesting, every user who clicks is worth $1. I'm sure you're worth a lot more, but let's pretend $1. We can verify this goal to make sure it's real and it says this goal would have 57 percent conversion rate based on your data for the last seven days. That's great. Google Analytics can actually retroactively tell me what percent of people completed this goal. Let's save, and that is it. Now, if we go back to our campaigns, we can see the performance by now, a new goal. We're not getting a lot of information right now, is not coming in live. But eventually as users come in, these percentages over here will continue to update and continue to be reflected in real time. Great. Hopefully you've learned how to do this and you can see how simple it is, so why don't you go ahead and just try it at home? 7. Attribution Methodologies: Now we've talked about UTM codes, how to structure them. We've also gone through Google Analytics and understood how to actually look at performance of various campaigns according to their UTM structure. Let's dive into the Attribution Methodologies. How do you actually know and attribute the user conversion to specific marketing campaigns? You've actually got a quite a few very choice. We're going to go through three today. We're going to first talk about last click attribution. It doesn't matter what marketing channels you came through, it doesn't matter that you discovered it through a banner or ad, you then landed on the page on your own volition, and then lastly, you landed on the site through Google AdWords. We're going to give all the credit to the Last Click, which is Google AdWords. The opposite is First Click Attribution. It doesn't matter what you did after the first visit as a user, what after you visited for the first time, everything you did afterwards gets no credit. If a Google AdWord leads you to the site and then you click on a lot of ads afterwards, only the AdWord is going to get the credit. The last one we'll talk about is Multi Click Attribution. We're going to assign a value to every single click on a user's journey, so they click in the beginning on an Adword, they click at the end, on a banner, every single touch point is going to get a little bit of attribution, we're going to spread the wealth around. Now that we know each of these attribution techniques, let's dive into one you'd want to use one of them. Let's see how we can apply the various attribution methodologies to quasi real-world data. Imagine in the first line you've got a user who clicked on a banner ad, he then went to paid search and he then found the app or your website through organic search and end up making a purchase goods worth a $100, so the question is which of these channels actually gets the credit for the $100? Because the user landed on the app through organic search last, and if we want to do Last Click Attribution, organic search gets all of the credit, the other channels gets absolutely no credit, organic search gets it all. Now what if you instead wanted to think about this in a First Click Attribution? Well, in this case, slightly different user found out about the company, let's say searched through a banner ad, then found out about it again through a Google AdWords paid search app, and then finally went back to the company with through an email blast signed up, spent a $100, but which channel gets the credit? Under this scenario, the first channel because we think First Click Attribution gets all the credit, so you could pretend like banner ads, essentially where worth a $100 in this case, because the user was worth a $100 and paid such an email, really had no contribution. There is another methodology type called all channels get equal credit, where the user came in through all click three channels found out about the website through the three different ways, finally made a purchase, and now each of these channels gets $100. It's not a very common technique, but it's there. The last one, which is often the most used is the Multi Click Attribution. In this case, the user found out about the website through a print ad, came back to the website through social media, came back to the website through paid search and made a payment for $100. In this case, you take the $100 at the user's worth and you split it evenly because it's Multi Click Attribution. The idea behind there is that all channels get equal credit, so the print ad gets $33 a credit, the social media gets $33 of credit and paid search ad gets $33 a credit. In this case, if the marketing manager has spent $30 on a print ad, $30 on a social media, and $30 and a paid search ad, would end up being like this is actually positive, all three channels came back and were actually brought in a customer that was worth more than how much they spent on the marketing. 8. Evaluating Marketing Campaigns: We've almost got all the knowledge to evaluate a proper marketing campaign. We understand how to structure the UTM codes, how to put them into Google Analytics. We also know the trade-offs about various types of attribution methodologies, the first click versus the last click. We got to go a little bit deeper into the metrics. How do you actually evaluate the marketing campaign? So there are actually quite a few costs metrics that marketers use. But what I recommend is to focus on the value of your customers. So there are common metrics like Cost Per Click, Cost Per Impression, that are used. Cost Per Acquisition is really what you're aiming to measure. So let's dive into that. Cost Per Impression is the total number of $ spent on marketing divided by the total number of impressions delivered. So let's say you're in Cosmo and you see the articles at the top are right below is an ad for I'm not even sure what this, but something where you can get free shipping on orders over $99. What's happened behind the scenes is this advertiser has won this spot. They're showing you this ad, and as a result, they're paying Cosmo a tiny little bit in order to show you this ad. So this is the cost of impression. They might have another deal where you got to click into it and then maybe they'll pay for clicks, but generally banner ads are Cost Per Impression. The next is Cost Per Click. When you run a Google AdWords campaign, you don't really get charged if Google shows an ad to somebody, but then they don't click it. As a company and as a marketing person, you only get charged if the user goes ahead and clicks on the ad. So that's why the measure Cost Per Click is often really heavily cited when you talk about things like AdWords. The way you calculate this is the total number of $ spent on, let's say, Google AdWords marketing divided by the total number of ads that were clicked. Now, lastly, is Cost Per Acquisition. This is the one that I think is the most important, and really it's the one that you want to optimize your marketing channels to. This is the number of $ you spent on marketing divided by the total number of users acquired. The reason you want to focus on this is because the other metrics like Cost Per Impression or Cost Per Click, are all trying to somehow get at the fact that through clicks and through impressions, you'll eventually bring somebody on and you'll acquire them. This is a metric, it skips all that. This is literally how much it's costing you to bring out a person. If you can get this, you don't need to worry about any of the other metrics. But how do we know what a good or bad CPA is? What if I could bring on a customer for $5 or $12 or what not, how do I know if that's good? $100, $1000, is that a good CPA? It all depends on your customer lifetime value. The users are worth more than how much they've purchased. If a user goes and makes a $2 purchase right now, it doesn't mean that they're worth $2. They could be worth a lot more. So let's talk a little bit Socks Ink. If Socks Ink is a subscription business, and we know that users are buying Socks for $10 a month, they probably stick around for more than one month. Very few people probably sign up for the first month, pay the $10, then leave. So there are actually worth more than $10. The way you get the true worth is you take a look at how long users stick around for. So let's say users on average stick around for six months. Well, obviously if you acquire them, it'll be worth way more than $10. It'll probably be close to $60. That's how you think about customer value and how you think about CPA. Let's go through a more detailed example. So setting the background here, Sock Ink, we say we know it's $10 a month, and we know that users stick around for six months on average. So we've launched some new campaigns. One campaign is called Warm Socks. We've spent $2,500 and we've acquired 40 users. Another campaign is called Winter clothing. We've spent $3,000, we've acquired 80 users. Finally, the last campaign is called Knitted Wear. We've spent $5,000 and we acquired 100 users. So which ones of these are actually worthwhile? Which one should we keep and which one should we get rid of? So we know the average user is worth $60 approximately, they stick around for six months, they paid 10 bucks a month. There's a couple more nuances here, but let's say simple. We can evaluate these campaigns by simply multiplying the number of acquisitions by the value. So you can see under Warm Socks at 40 acquisitions, well, that's $2,400, Winter Clothing at 80, that's $4,800 of value and Knitted Wear had 100 acquisitions, which is $6,000 of value. That doesn't mean that they've already paid that much. That just means that we expect them to pay that much in the future. So when you acquire customer now, even though they're only paying $10, we really want to think of them as being worth more. Now that we look at these results on the right-hand side, what should we do with the Warm Socks campaign? We spent $2,500, but the customer value that we acquired was only $2,400. We should get rid of this. As a spark marketing manager, if you have access to this data, we have the Cost Per Acquisition, you got to get rid of the Warm Socks. How about the Winter Clothing campaign? We spent $3,000 and we got $4,800 of value. Do we keep it? Absolutely. We got $4,800 of value, which is more than our spend. We probably didn't get that much money upfront, but over time we expected to keep spending more and more. Finally Knitted Wear. $5,000 are spent for $6,000 of value, do we keep it? It's less efficient than Winter clothing, but the value is greater than the spend. I'd say absolutely keep it. Right now you're acquiring users at a CPA that's less than their value. So you're always ending up positive. If you could spend more money on the Winter Clothing, make that channel as big as possible, make the campaign as big as possible. But Knitted wear is also a good campaign, even though it's not as efficient. 9. Project Overview: We've learned quite a bit so far. We've learned about attribution, we learned about lifetime value that cost per acquisition. Now it's time to put all of those, all of that learning to the test with the project. Attached, you'll find an almost realistic dataset, this is a little more curated than you'll probably get in the real world, but you'll get dataset with acquisition data that looks like below. You'll see three columns. One that says the first click channel, one that says the last click channel, and one that says total acquisitions. For example, we know that Banner ads followed by AdWords resulted in 50 acquisitions. We know that AdWords followed by Banner Ads, resulted in 85 acquisitions. You'll get some basic attribution and acquisition data in one sheet. In a second sheet, you're going to get information on how much the marketer spent on each of the channels. Adwords, the market spent $500, Banner Ads, they spent $800, etc for all the different channels. Your goal is to apply first-click attribution, last-click attribution, and multi-click attribution to the data and understand which of these channels did well and which of these channels didn't, which of these were the most effective, and which of these channels were the least effective? You should use either Excel or Google Sheets or really any other tool at your disposal, and what this is supposed to replicate is how you're going to analyze the various marketing campaigns in the real world. In the real world, you're going to get some click data and you might get some attribution and conversion data, and you're going to have to make decisions yourself on analyzing, and most of the marketing managers. I worked with work very heavily in Excel and Google sheets evaluating different campaigns and evaluating which ones do well. Good luck. 10. That's a Wrap!: Well, congratulations, you made it to the end of the Marketing Fundamentals course. Let's go over some of the topics that we covered. We talked initially about the purpose of digital marketing attribution. We talked then about the various marketing technologies, the Cookie the UTM codes, the cross device attribution. We talked about UTMs and how to structure them. We then implemented some of the UTMs and goals in Google Analytics. Finally, we did a little deep dive into Evaluating Campaigns. All of these techniques are critical for the modern marketing manager. The modern marketing manager has access to a lot of resources, a lot of analytics, and the ability to really fine tune and optimize their marketing strategies. So there's no reason not to. But we didn't cover it all. Obviously, there's only so much we didn't cover in these 45 minutes. We didn't dive into the various marketing channels and performance. When should you use display versus the other types of channels. We still need to evaluate, not cost per click channels, so TV offline channels. How do you really do lifetime value? That's a whole course in itself. We can still dive into attribution models and Google Analytics and finally, how do you manage your marketing budget? We are always adding more content and eventually we'll be building out a longer course in Analytics. If you want any more information, come visit us at Don't forget, you can post your answers to the project in the discussion below. Feel free to contribute and help each other out, and I'll be jumping in as well and answer any questions that you have. Good job at this class and see you next time.