Segmentation, Part 2: Techniques, Nuances, & Amazing Features Tutorial

2.2 Introduction

Hi everyone, my name is Avinash. I am the cofounder of Market Motive and the faculty for Web Analytics. Welcome to part two of our journey on the Analytics Data Segmentation Awesomeness. This video will contain a lot of secret tips and tricks and exposes the hidden sub nuances that are necessary to understand in order to unlock the true power and awesomeness of what is available to you inside analytics when it comes to segmentation. We are also going to cover a lot of amazing features that you may not be aware of that were not there in the past but are available to use now that really unlock a new level of analysis that was just simply impossible really, in the past. So, with this video you are probably going to want to have your notepad ready and take some notes for yourself. You might even come back and watch it a couple different times because there is slight sort of technical bent to it. But without this video it would be very hard for you to unlock the full power of segmentation. So, please do the attention. If you're excited let's go.

2.3 Conditions

The first thing I want to talk about, is this idea of conditions. Conditions are very important to any analytics approach that includes segmentation. Let me give you a specific example. So here, as an example, I'm going to focus on the language of the person, just as an example, but it could be anything really, whereas you can notice that I have a number of options available for me when it comes to the condition I want to apply. Now assume for a second that we were doing this for our keyword analysis and the keyword that I was interested in typing here is Avinash. Or a language called Avinash, it doesn't matter. Right? So it's Avinash, and we've typed it here. Now depending on the condition that you apply, it can dramatically change the segment that you end up creating. As an example, you could pick a condition called exactly matches. Doesn't matter, right? When you pick the segment it will only pick visits where the exact word Avinash is present. And it's going to ignore all other visit where anything else is present. Including Avinash Kaushik, even that is gone, because it is exactly matching Avinash and nothing else. If I, instead of saying exactly matching as a condition, if I choose my condition contains Avinash, in which case it will match these criteria, it will match Avinash Kashek. Kashek Avinash, I like Avinash. And so essentially what it's doing is looking for the mention of the word Avinash. And it will pick a much wider pool of people than you had originally if you had chosen exactly matches. Let's look at one more example in which case we choose the segment, starts with, so that's the one we're going to choose. In this case we will pick Avinash Kaushik, Avinash rocks, Avinash stinks, Avinash is so amazing. Every single place where the word Avinash is present first. So as an example, it's not going pick I like Avinash or Kaushik Avinash. Both of which are segments that you saw which would be picked up earlier. So starts with will apply that segment. So there are three of the most important ones you should think about, which is exactly matches, contains, starts with. And then it is easy for you to understand what is ends with and what is does not contain, does not start, does not end, etc., etc. But there is one other one here called match a regular expression. And that's a condition. And so let's look at that one. And so that one is going to look something like this. In this case, I'm saying I would like you to match a regular expression where I would like the search keyword query to only contain two words. If somebody had searched for I like Avinash, I don't want that, I only want covert queries where two words were present. In which case, you would use this regular expression that you see here. You would type that here, and then you would be able to pick up the search queries will be two words. In fact, actually, you can type in this regular expression in this particular render for language and chose matches regular expression and it will pick languages that have two words in them. So it's won't pick up the language English, it will pick up French Polynesian, as an example, but because that language has description has two words in it. Regular expressions require to have a little bit more technical knowledge. And if you are not familiar with how to create regular expressions, I encourage you to download this ebook, a PDF. If you go to this URL, it will download a PDF onto your computer that walks you through regular expression and all the basics and the complexities around it. So it's very important to understand that conditions will increase or decrease the sizes of your segment. Conditions will ensure whether your segments are accurate or inaccurate. It's very important to understand what happens when you click on that drop down box.

2.4 Using "And" & "Or

The next thing that we're going to look at are and and or, and what happens when we pick either and or or when we do segmentation. Take this simple picture. If I apply the or condition to my segment, essentially, what I'm saying is I would like to have 1 or I would like to have 2, which means give me everything that is inside this blue box. That's, at its core, what happens when you apply the OR condition. Another condition that you will apply often is AND. And so what happens when you apply AND? When you apply AND you're saying, unlike the 1 case above, I don't want all of 1. I don't want all of 2. I would like to have where 1 and 2 meet. So essentially my segment will only give me this little overlap. I'm sorry, not this orange and not this purple, but everything inside where this overlap is. So it's very important to understand that when you use AND or when you use OR, you get very different results because you are looking at very different things. Let's solidify this concept by looking at an example. In this case, I am saying I would like to look at people who have page views greater than 2. So, Pageviews > 2. Now, I can choose a condition here, AND and OR, and that will allow me to expand or contract this segment. So I can click on the AND button, it creates this AND line here you see and what I'm creating essentially is a segment that says, I would like to filter page visits but page views are greater than 2, but less than 10. So what I'm saying is create a segment for me where I would only like to focus on people who have seen more than 2 pages, and less than 10 pages, so you can see how the condition gets applied here, and it helps me focus what I'm looking at. Let's look at another example. In this case we're saying, I would like all the visitors, their country and territory, all the visits that come from India and those people during the same time period have also come from the country and territory of Israel. Also I can apply an additional condition that they should have come via google.com. So you can see how I am applying another two conditions at the same time and using the and parameter. Now, you can imagine that the end result of this exercise is going to be very small. Because the same person won't come from India and Israel all at the same time, and using Google, so it is a very, very small segment for me. But it's just a way to illustrate to you what happens when you use the and condition. It applies a much higher bar, and you can have a very small condition. Or you can apply the same thing but slightly different rather than clicking with and, you can say, I would like all visits for people who come from India, or people who come from Israel, or people who come from Indonesia, and they also use the source Google. So there's an and, and an or going on. But as you you can imagine as you look at these three or conditions, you're not getting a small population of people, you're getting a very big population of people because they could have been from India, they could have been from Israel, they could have been from Indonesia, it doesn't really matter. It's a very big condition and we will get more number of people so. Whether you use or, or you use and, it will greatly impact the size of the segment you end up creating in the focus area that you have.

2.4 Using "And" & "Or

The next thing that we're going to look at are and and or, and what happens when we pick either and or or when we do segmentation. Take this simple picture. If I apply the or condition to my segment, essentially, what I'm saying is I would like to have 1 or I would like to have 2, which means give me everything that is inside this blue box. That's, at its core, what happens when you apply the OR condition. Another condition that you will apply often is AND. And so what happens when you apply AND? When you apply AND you're saying, unlike the 1 case above, I don't want all of 1. I don't want all of 2. I would like to have where 1 and 2 meet. So essentially my segment will only give me this little overlap. I'm sorry, not this orange and not this purple, but everything inside where this overlap is. So it's very important to understand that when you use AND or when you use OR, you get very different results because you are looking at very different things. Let's solidify this concept by looking at an example. In this case, I am saying I would like to look at people who have page views greater than 2. So, Pageviews > 2. Now, I can choose a condition here, AND and OR, and that will allow me to expand or contract this segment. So I can click on the AND button, it creates this AND line here you see and what I'm creating essentially is a segment that says, I would like to filter page visits but page views are greater than 2, but less than 10. So what I'm saying is create a segment for me where I would only like to focus on people who have seen more than 2 pages, and less than 10 pages, so you can see how the condition gets applied here, and it helps me focus what I'm looking at. Let's look at another example. In this case we're saying, I would like all the visitors, their country and territory, all the visits that come from India and those people during the same time period have also come from the country and territory of Israel. Also I can apply an additional condition that they should have come via google.com. So you can see how I am applying another two conditions at the same time and using the and parameter. Now, you can imagine that the end result of this exercise is going to be very small. Because the same person won't come from India and Israel all at the same time, and using Google, so it is a very, very small segment for me. But it's just a way to illustrate to you what happens when you use the and condition. It applies a much higher bar, and you can have a very small condition. Or you can apply the same thing but slightly different rather than clicking with and, you can say, I would like all visits for people who come from India, or people who come from Israel, or people who come from Indonesia, and they also use the source Google. So there's an and, and an or going on. But as you you can imagine as you look at these three or conditions, you're not getting a small population of people, you're getting a very big population of people because they could have been from India, they could have been from Israel, they could have been from Indonesia, it doesn't really matter. It's a very big condition and we will get more number of people so. Whether you use or, or you use and, it will greatly impact the size of the segment you end up creating in the focus area that you have.

2.5 User, Visit, & Hit

The next thing to look at is a very interesting new concept called user, visit, hit. It's very, very tricky, very, very important, so I hope that you're paying attention but it is amazingly cool. You got a lot of fun, so a user is very easy to understand. It's me, or it's my friend, John Marshall. And so, when does a user essentially mean a person. That's what we're talking about. There could be many users who come to the site, so we have myself and John and Michael and Scott and Lilah. And we have all of these people, who come to our website, so when you think about users, it's all of these people who come to our site. When it comes to visit, we're focusing on one engagement that they have on our website during one session that they have on our web site. So, I had visited this site for 10 minutes and there's my session. And over time, I can have multiple sessions during the visit to the web sites. So, multiple visits to the web site. So, in the month of December, I have visited the websites one, two, three, four times. So, those are visits. So, users are the things on the top. Visits are the sessions that people have on our site. So, what is hit? Well, a hit is that little piece of engagement that happens inside a session, so in this case, I saw two pages. I submitted an order for $100 and I saw another page. So, each of those boxes that you see inside my visit is a hit. So, that's what a hit is. A user is a person, a visit is one session that they have on my website, and hits are things that happen inside those sessions. And during multiple visits to the site, I have many hits recorded for each session. And as you can see, some of them I have conversions. I purchased three times during this time period. And once I had visited, I only had one hit and I bounced away. I did not buy anything. So, that's what we're talking about, right? And as you can see, as we look across different users. Each of them have different number visits to our site. They have different number of hits. Some hits are longer. Some hits are shorter. Some people have conversions. Some people have gold values. Some people don't. Some people come to our website after longer gaps in time. Some come more frequently, etcetera. So, to summarize, a user is a person. These are all sessions. And every single thing that happens inside a session is a hit. The cool thing about Google Analytics is that it allows us to do segmentation on all of these things. In the past, it was only possible for us to allow segments inside Google Analytics at a hit level, or at a session level. A hit level would be a page, as an example. Or an event, or a heat level custom variable. Those are all hits. A session would be the visit for the person. Those are the only two things we were able to do. But now, we're able to also create segments at a user level, which is an amazing new possibility. So, let's go and create some segments at all of these levels so we can solidify our concept. As you can see in this case, when I am analyzing the behavior of people who come to my website and I click on this drop down, I have an option to create a segment that is on either one of the three levels. I can create a segment for transaction at a user level. I can create a segment for transaction at a session level. And I can create a transaction at a hit level, so I have three different options that I can avail of when I create a segment for behavior where a purchase was made on my website. So, let's see what we can do with this technique. I'm going to create a segment that is focused on creating revenue for the website. I've said I'm interested in revenue per user of greater than $125, in this case. So, remember what I'm saying. I'm not saying that I would like to create a segment of people who placed an order for $125. What I'm saying is that, I would like a segment that during this time period I'm looking at, a day, a week, a month, or a year, it doesn't matter, you choose but during this time period I'm going to apply this segment to a person should have placed an order totaled of $125 with my business. So, in our scenario that we looked at a few minutes ago, what does that mean? That means that these two users, John and Michael, are out of the game. Because while they did convert, John's order was only for $10, and Michael's order over the entire time period, was only $70. In this case, for me or for somebody else in green, the order was $100, plus $62 plus $25, in addition of all these three visits means the green person is going to qualify for this segment and not these two other people. So, this segment is really good to create for websites where people make multiple purchases. And we want to focus on a particular revenue target and it doesn't matter if you buy in one session or you buy it across many different days, or you buy it in 2 orders or 50 orders, it doesn't really matter. We're focused on high-worth individuals, and that is a new capacity that you can do in analytics, and that's what this segment will accomplish. Let's look at a different scenario. In the same case, I have said I would like to focus on revenue per session. Per session is greater than 60. What does this mean? What does this means is that during one visit I would like to know people who placed more than 60 dollars worth of revenue. Created $60 of revenues. In this case, what happens is that all of these sessions that are on our websites get cancelled out. So, everything gets cancelled out for the orange person, for the green person, session number three and four get cancelled out. But now, all that is left are two sessions that will be included for the green person and one session that will be included for the blue person. And that's what you're doing when you apply a per session condition and create that segment. Now, let's look at a different one, let's look at a focus on per head. We would like to focus only on the head where we got $100 of value from the engagement on our web site. So, what does that look like in our scenario? In our scenario, that looks like both of these people will be out. And all of these other hits will be out. The only hit that will be included in the segment is the one that you seek over there clearly. That happened for the green person. And so with a hit, it's very, very granular. It excludes all that other stuff. And we just get that particular hit reported back to us, so when you focus on these conditions, it really helps you understand visits versus users. So,in this case, we are looking at a segment where we had $100 worth of e-commerce revenue and we have people who came to us during the email campaigns, and we can focus it either on visits where only the visits initiated by the email campaigns and the head e-commerce greater than one hundred. Or in this scenario, we could focus these two visits get excluded, and this visit gets included because it qualified for the one hundred dollar session. Or we could focus on users. In which case, when we focus on users and say we would like a segment for users that came from email and had a credit in $100 session, then our scenario would look like this, where we only have one $100 session, but because we're focusing on user, all of these visits will be included in our segment. So, that's what's happening when you switch from visits to users. It's very, very important, because you are focusing on completely different areas, and you're going to get completely different results depending on what you end up choosing.

2.6 Sequence Segments

The next thing that we would like to look at are called Sequence Segments. These are very important and give us a new power when it comes to understanding behavior on our websites. So what are sequence segments? Sequence segments are very simple. They allow us to analyze a step by step behavior of what is happening on our website as you might imagine. With a name like sequence so let's actually look at what a sequence segment looks like. In this case, let's say we are running patagonia.com which is one of my favorite websites and so I would like to buy this jacket which is very nice and I have added it to my cart and I have started a checkout and here I'm in a checkout process and for whatever reason I don't complete the checkout, I just leave. So, I added something to the cart and I abandoned it during the checkout process. Maybe I just hated this page. Right, because the page is awful. When it is not but let's say the page was awful and I abandoned or, something else happened, and I abandoned. Now for Patagonia, it's very important to understand, hey these poepoel are really important. They already found the product and add it to cart and now they are leaving. So, we should figure out what is going on with these people and the only way that you can analyze people who who did this particular sequence of action is to create a sequence segment. So this URL will be very important to us, that you see at the very top, in our sequence segment creation. So please make a note of this when your are creating sequence segments of this type. So what we're going to do is, first, we're going to create a step where people have made it to that particular page, shipping.jsp. This is the URL that you see on the top, and we took this particular stem. This stem and we typed it here, or copy pasted it here. And then what we do is we say we would like everybody who came to this page start at the checkout, this is a first page of checkout. And then we say, and it's followed by, and that step is followed by the second step of not placing an order. So, by choosing transaction per session equals zero, we are saying we would like to focus only on people who start the checkout process, but do not complete the checkout process. Do not place the order. And you hit the save button. This will allow you to focus on the behavior that we just observed on and allow us to understand how many people are doing this, and then where do they come from, and what pages did they have, and what products did they have in their cart, and were they coming from page search, were they coming from email campaigns or organic or were they from Pakistan or India or Bangladesh or Russia? What was going on, allows us to do very deep analysis for people who go in this particular sequence. And it allows us to do some really great analysis. It's very important to note that when you create step one and step two in sequence, you have two options in terms of conditions. One is is follow by, which is that one I had used earlier. You can it here. I've used is followed by, and in this case it says is immediately followed by. So there are two different scenarios. In my guess I had said that after adding to checkout, people could have come to this page. And then maybe they went back and then they monkeyed around here, changed the quantity, and then came back to this page and then did not convert. I was fine with them doing that. But when you choose the condition, is immediately followed by this. Then you are really restricting to only the people who were in the check out process, and did not go back or do anything else, but they just failed to make it through the Conversion process, so it's a very very good way to force on a much narrower focus in terms of your analysis. So, sequent segments are very important, and you can easily see How it is entirely possible for you to create a sequent segment of people who are using immediately followed by abandon shipping, abandon but after going to shipping plus billing, or after going to shipping plus billing plus payment. So these are sequences that are in your process, and you can use this segment to create sequences that that really help you focus on groups of individuals were exhibiting this behavior and analyze their performance so you can make more money. You can also combine these two things. So in this case, at the top you can see that there is a segment I've created for one particular ad content. And people who came to my site from a campaign that I had targeted to them, and that they happened to be coming from Bing. So you can see that I have created sequence segments. I'm using multiple sequence segments in order to focus on a very sophisticated type of segmentation strategy. A one that is very important to my business and I just wanted to have this here to show you the amount of creativity that you can add to this process and do something really cool.

2.7 Cohort Analysis

The last thing I wanted to share with you is this new way of creating segments called cohort analysis. And I want to be very careful. A lot of people have very different definitions of what a cohort means, and I would like to share with you the definition that is in Wikipedia, because it is a very good definition. So a cohort is a group of subjects who have shared a particular event together during a particular time. For example, people born in Europe between 1918 and 1939. Or survivors of an air crash. Or truck drivers who smoke, between the age of 30 and 40. Cohorts may be tracked for an extended period of time in a cohort study. So what it is is a group of people whose behavior you want to observe over a long period of time. That's what cohort analysis. And this is something that we have not been able to do in the past with Google Analytics. In fact, you still can't do it with pretty much any analytics tool in the marketplace except for Kissmetrics, which is a wonderful web analytics tool where you can do some really cool cohort analysis. But cohort analysis is not possible with most analytics tools in the marketplace. So what is cohort analysis? So, here's a good example of what cohort analysis is. In this case, I am creating a segment where I am saying that I would like a segment where the first visit to my website was between July 7th, 2013 and July 14, 2013. So for whatever reason, that week is super important to me and I would like to have people who visited my website during that time because I was running a campaign during that time. I can add to that and say, in addition to that time, I would only like to focus on people who came to the site in response to this email campaign that I did, and for forever reason, the title of my email campaign is Email March 24. ]So now you can see on the right hand side, I've created a cohort of people who first visited my site during this time period and they came from this campaign, so I've collected that pool of people. And now I can go and apply that segment to my web analytics reports and try and understand the behavior of people who first came to my site, and first, for the very first time in their entire life, were introduced to my website during this time period from this particular campaign that I had done. And I can analyze the behavior completely separately from other people who might have come to my website, or even separately from people who might have come on the campaign before or after this particular week that I'm interested in analyzing. And that allows me to segregate people into different segments depending on the cohort by a commonality they share and do some really amazing analysis.

2.8 Thank You

Those are the five different techniques that I wanted to share with you. They are supremely important if you want to do segmentation to the fullest power possible. Doing things that you are not able to do like user level segmentation, like sequence segments, like cohort analysis. In addition to knowing the impact of and or when you applied those conditions to your segments or other conditions that you can apply to your segments like contains, matches exactly, starts with, ends with all regular expressions. So this is a foundational element to allowing you to unlock the power of segmentation. I encourage you to watch this video one more time so that you can truly internalize these techniques. Thank you.

  • Disclaimer
  • PMP, PMI, PMBOK, CAPM, PgMP, PfMP, ACP, PBA, RMP, SP, and OPM3 are registered marks of the Project Management Institute, Inc.

Request more information

For individuals
For business
Name*
Email*
Phone Number*
Your Message (Optional)
We are looking into your query.
Our consultants will get in touch with you soon.

A Simplilearn representative will get back to you in one business day.

First Name*
Last Name*
Email*
Phone Number*
Company*
Job Title*