Video: An Introduction to Segmenting Website Data

Hello, I’m Anna Lewis, and this video is about segments, using segments to improve your website’s results, and also just learning a bit more about your visitors and what’s happening on your website. This video should be useful whether you’re in SEO, PPC, conversion rate optimisation, or anything that basically involves website statistics.

Where can you see segments? The easiest place to do this is Google Analytics. You can use the default advanced segments, or you can create your own custom advanced segments. These easily show you the data, and they can apply it historically.

Options within Google Analytics that can be applied historically are custom variables, which are really good for choosing the segments that you want and working out which visitor fits which segment to start with. These are set up with a bit of code on your pages.

The other option in Google Analytics, that can be done historically, but is always good for the full segment, is to use another profile and put a filter on that with your segment specifics to pull away that traffic from the other bits that you don’t need to look at on that occasion. For example, the Koozai Analytics account has at least 15 other profiles that have all sorts of different filters on them.

The other way of segmenting the data, if you’re using AdWords, for example, you can use the Segment button, or you can go to the Dimensions tab and have a look at some of the segments they have there. Obviously, if you’re using something like Website Optimizer, that’s its own little interface itself, so you can just have a look within that. That’s sort of the main ways that you would segment your website data for most people.

So, what are you going to look at? Let’s start with sources. This is where your traffic comes from. You might want to group together your email campaigns. Work out which spring email campaigns have performed the best over time, rather than looking at all of your email campaigns together.

What about looking at which links people click within your email campaigns? Take all the information from your email campaigns, but segment by the link type or the final destination.

That’s just emails. You could segment your referring traffic by links you’ve built versus links that have been built naturally, or links from forums versus links from blogs.

So there’s a number of different things you can do to actually get a better understanding of why your traffic is coming to the site and what it’s doing there. From that, you can build on that to improve the site for these people and help gain conversions. Obviously, at the end of the day, making money is one of the biggest reasons for looking at this data and making sure you optimise it effectively.

Within sources, we’ve got keywords. That’s a pretty major part of any SEO or PPC project, is focusing on your target keywords, then understanding how they perform. You might want to group your keywords. You could, for organic things, or even PPC, you could look at non-brand versus brand. Then look more closely at your non-brand segment and have a look and see what’s working, how differently ‘brand’ versus ‘non-brand’ terms interact and what the results are, and what you can do to optimise each individual set of traffic.

Additionally, with keywords, you could group them by your category. So, you think about what your top selling product is or your top referring keyword, group all of those similar keywords together. Within that segment, have a look. Have a look at how these people are coming to your site, how long they’re spending there, how much they convert, and which one actually is the best performing keyword within that section alone, rather than getting it lost within the bigger picture of all your keywords in one place. That works for both SEO and PPC.

Other things to look at, with regards to keywords, is the rank or the position. Within Analytics, you can create a separate profile and add some filters to that so that when you’re looking at your keyword report, you can see what position the result was in when the user clicked it. Now this can show you how well your position threes compare to your position ones, or you could look at it by keyword and see, “Well, actually, this keyword works best in position two, rather than position one.” Also see, actually, when it was in position one for a day, it received 100 visitors, compared to 75 visitors consistently from a position two. That shows you how much you can gain from being in the top position, just by segmenting and having a bit of a closer look at your data.

Position, now that’s AdWords, so you really benefit from having a look at which position works for you, for each keyword type, each ad group, maybe each time of day versus which position. It might be that position four actually has a much better result than position two because position four is at the top of the right-hand section. People’s eyes are drawn to that, rather than position two, for example, where the ad might get lost amongst the others that it’s in between. What you might want to work out is what your ROI is based on a position four compared to a position two. It might be that because it’s slightly cheaper, you might get a better click through rate because of the position. Then, if you get the same amount of conversions, if the conversion rate is similar, then actually it’s much more profitable to be in position four. That’s something that really helps you understand what you can do to help you gain the most money from your budget.

What else have we got? We’ve got split testing. Now that’s something you can do in AdWords or email campaigns, even organically. An example in AdWords, what you can do is head over to Dimensions, and then look at the destination URL. Where you’ve got an ad group, and you’ve got two or four ads, and they point to different destination URLs, you can then see which one actually has the better conversion rate, which one makes the most money, basically. You can work out which of your landing pages is the better one. This can, obviously, also be done through Website Optimizer. Even in Analytics you can look at your landing pages, which is, obviously, another section, just to see which one is performing better.

So landing pages, you can look at these whether you’re looking at your organic traffic, your referring traffic, your paid traffic, any of those. You can segment by what type it is, have a look at your landing page, and then think about the data. Think about your interaction. What’s your bounce rate from each of them? Which ones are performing better? Can you push any up so that more people see the ones that are performing better, or maybe the ones that are there already, getting a lot of traffic are the ones performing well. What more can you do to increase their visibility or improve conversions if they’re bringing in the most traffic? You really need to look closely, segment by several things, and work out what’s going to work best for you.

We’ve got two left. These are areas of the site. That is related to landing pages, really. What I mean by areas of the site is maybe /blog, you want to analyse very separately to /services. If you’ve got two quite distinct areas of the site, or even two different product types, maybe you want to compare different product categories. By applying some segments, you can have a look at how these perform on their own, without getting the information from the other types of areas of the site. You can just look at your /blog, work out what’s performing the best there, work out what you can do to improve it, and increase whatever it is you’re looking to get from your blog.

This works really well for product categories. You can see whether the people coming to your site do well looking for one type of service versus another. You can work out how to improve your website, how to make changes to increase conversions.

Visitors. We’ve spoken a little bit about interaction so far. So, visitors, you want to look at new versus returning. You want to look at their demographics. Now demographics is something that you can set up and track through custom variables. If it’s possible to work out the demographic of a user on your site, then set this as a custom variable. You can then compare male versus female, for example, or different age groups, or whether somebody is a member or not a member, or somebody who’s converted or not.

Using the custom variables, you can then see how you can optimise best for each of those people. It might be that men convert a lot better than women on your site. Maybe you need to put a bit more budget toward them, or maybe you need to work out why women aren’t converting. See what you can do to gain better results by just looking closely at the segments.

The last one on the list is errors. I didn’t forget it. Errors is one of the most important things to use a segment for. When there’s something wrong, when you notice an anomaly, whether it’s a good or a bad one, in fact it’s not just errors, looking for a trend. The first thing I’ll do when somebody says they’ve got a problem with their website traffic is look at traffic sources, throw in a couple of the standard advanced segments, and see the graph. Look for whether something drops off, whether something increases dramatically, and where any sort of these anomalies are. Then you can dig down into why that might have occurred.

If you don’t spot anything there, I’d head over to landing pages, see if anything’s changed dramatically there, with a bit of a date comparison or that sort of thing. Content is another place you can look. If your content on certain pages has suddenly dropped to zero, that’s maybe going to tell you that your traffic isn’t there. But you don’t know without looking a little bit closer.

We’ve sometimes spotted a problem where one keyword has suddenly, over time it’s just increased and you’re getting quite a lot of visits per day, but it doesn’t show up as being a core target. You look at that and you think maybe I’ll just segment by that keyword alone and work out what’s going on.

In this example, we worked out that one person was coming to the site at least 10 times a day from the same computer, the same location, using the same keyword, on the same browser. All those factors, having just segmented by one keyword, we could work out almost exactly where that person was, what they were doing. The client was quite surprised by the amount of information we could tell. It’s really good for working out what’s working and where the anomalies are.

There are other ways. Obviously, I mentioned, that you can segment by browser, by desktop or mobile, or by social interaction, or browser, what resolution they’re running. There are so many things. When you work out an anomaly, you can segment by that and really work it out.

We quite often end up looking at the host name. That’s a really good report for seeing which URL your tracking code is actually being shown on. By creating a segment of the anomaly, you can look at all of these things and actually pin down what’s going on, whether all your traffic is coming from one IP alone, or anything else for that matter.

So, plenty of things to consider and plenty of ways where segments and segmenting your data can be really beneficial. There’s more information about all of this on the blog, and I’m sure I’ll be posting some more in the future too. Keep an eye out on Koozai.com/blog, and check out the links here for some more information. Thank you for watching.