5 Tips for Using SiteCatalyst Version 15 segments

With an ever increasing number of people going onto version 15 of SiteCatalyst I thought that it would be a perfect opportunity for me to write a blog post about segmentation, which is the major functionality improvement. It looks like it is something that should be easy to do, but if you follow my easy instructions below, you'll get the best out of it. Remember, the data you get from it is worthless if you don't do anything with it!

1. Page Views Based Segments

It seems to make sense to start with page view based segments first, because they are the lowest denominator in the world of web analytics.

But in the case of segments, I find these things the most difficult to deal with and I'll try and explain why in the hope that it will make it easier for you to understand, as perverse as that sounds.

The main reason that I have trouble with page views based segments is that I can't really relate them to visits, which is the metric that I use most frequently, but I think I have some diagrams that will help explain the situation:

I'll explain in a minute how to create the segments themselves, but lets get our heads around the values first. Your segment only includes a subsection of all the pages on your website. The visit and visitor do not get counted unless they land in your segment. This has a couple of implications:

• The entry page to the whole site may not be in the segment, so the segment's entry page will be the first page in that segment that the user arrives at
• This also means that the referrer when the user does eventually get to a page in the segment is going to be from an internal source (not the original entry source)
• The user may leave to pages outside the segment before returning - in pathing reports SiteCatalyst will not represent the leaving part
• The exit page to the whole site may not be in the segment, the exit page of the segment is the last page the user viewed in their visit in that segment
• A visitor may have visits to the site which don't touch the segment affecting the visit number reports - a New Visitor is only new to the segment

The best way to think about this is if you have two completely separate websites - one for your segment, one for everything outside your segment. If you were looking at analytics reports for one of the sites, you wouldn't expect any visits to the other to be included, nor would you expect to know about what the user did before getting to the website.

We have three ways that we can create segments:

1. Creating a defining rule to show a number of pages to include (or exclude)
2. Creating more than one defining rule to show a number of pages, of which the rules have overlap
3. Creating more than one defining rule to show a number of pages, some of which may not have overlap

Creating the segments is very simple itself. Once you've clicked the add segment button you get a screen a bit like the one below:

You just need to drag your container from the left hand side and put it in the box on the right hand side. You have two options here - you can either have an include or an exclude. If you use an include then it will only include pages that you specify in your filter, excludes obviously do the opposite.

One note - try not to have double negatives because SiteCatalyst doesn't like them.

You can then click on the 'Page View' or the little pencil icon in the corner to edit the details about your container to choose your options:

Here you have the choice of using any of your standard or custom traffic or conversion variables. Any pages where the values in these pages match your selected filter will be included in the segment (don't forget to give it a name!).

There are several options for how you set your filters:

• Use greater than and less than for numeric values
• Contains, starts with and ends with are great for situations where you want to do a search type filter
• is null is great for situations where you can't work out which pages don't have certain tags

2. Visit Based Segments

Visits based segments are much more intrinsically obvious. You deal with them all the time in the world. There the staple of Google analytics segments, so I won't dwell too long on them.

However there are some things that you should know about them. If you create a segment based on visits, then it will include the whole visit, not just from the point of that it qualifies for the segment.

The other thing to note is that this may affect the way that your conversion reports work. If your conversion report is set to a length of time different to a visit, then it's possible that your reports may be slightly misleading. Remember that in conversion reports, the value of the conversion is related to the time of the event, not of the time the value was written to the report. So you could have a report that is showing you values from previous visits in your report, if you event is included in the segment.

For example, my segment could be people who viewed the home page. But I didn't view the home page on my first visit when I came through a campaign, so my first visit isn't included in the segment. In the second visit, when I am included in the segment, my conversion on my campaigns report, set to 30 day time out, converts against the campaign from the first visit.


3. Visitor Based Segments

Visitor based segments work in a similar way to visit segments, but instead of including just the visit, they will include everything that the visitor has done. This leads to some interesting scenarios:

The visitor might have done the action that puts them in the segment during the time period of the report. For example I set up a segment for people who have bought one of my products as a visitor segment. If I am looking at a report for December 2011, it's possible that I am looking at visits and visitors where they bought that product in January 2012.

It's also worth remembering that the rules for the custom conversion reports still hold - if you create a visitor segment based on the value in one of your eVars, but the time out for that report is set for visits then subsequent visits by the visitor will convert against the value for that visitor.

This has some good effects. If you create a segment for people who come to the site through Facebook, you can look at the referring domains report to see what started the visit that actually caused the conversion. Don't forget that this visit from Facebook could have been after the conversion.


4. Event based variables

You can also add events in as your containers. This sounds a little odd, but remember that you can set events on any pages (or on clicks) that you like, so creating something that only shows when the event occurred could make sense, although it does bring us nicely into the next section...


5. Combinations of the above

Event based segments only really work where they have one of the other three containers involved first. You can add a container that is for visits and then specify that the only visits you want are the ones where they had an event in them. This is great for looking at just those that bought something or those who signed up to emails and so on.

Obviously you can also have visitor segments which only look at particular visits or particular pages and you can have visit segments that only include a number of page views.

However they don't work the other way around, you can't specify a subset of pages and then say only give me visits who came from search engines. Follow the heirarchy and you'll be fine.

2011's Predictions Revisited - How Wrong was I?

At this time of year, it is traditionally that people write their predictions for the year coming up. They usually make a series of bold claims that subsequently start to look ridiculous or they make a series of claims that are so mind numbingly obvious that everyone gets bored after a couple of seconds. Of course the next year nobody bothers to check on their predictions to see whether they were wildly off or particularly astute. Given that I did the wild predictions last year, lets revisit those claims and see how stupid I really was.

1. The Fall of Google

Ah.

This is going well already, isn't it?

So Google didn't fall last year at all, in fact, it went from strength to strength. It is still the number one visited website according to anyone that matters (Alexa, Compete, Comscore, etc).

But it hasn't all been going Google's way. Why did I think that it would be their downfall last year? The reason I thought was spam. Particularly I was looking at spam that was 'user generated content' within affiliate websites. Sites which offered 'reviews' to give value add, which were becoming increasingly nonsense and no use to the user at all. I thought that users would soon start switching off because of this.

However what I didn't foresee was Google making some major updates to the way that the algorithm worked:

• Google Panda was an algorithm update in February. One of the main differences was that newer content hit the top of the rankings sooner and pushed older content further down the rankings. It also seemed to adversely affect affiliate websites and websites that do scraping of content. It also reduces the impact of mass produced pages.
• Google Plus and Plus one is the start of a major change to the way Google creates its personal algorithm. Individual users are going to have their own personal algorithm based on what their contacts have been doing.
• Google changed the default search to secure. This will mean that they can transition those who have Google accounts to use their personalised data for search. Particularly this will impact those who are using Gmail, Google Plus and other systems that frequently run in the background.
• More Personalised Paid Search. Whilst all the organic updates were going on, one big change that was often missed was the move towards personalised paid search. This was always the fall back of those sites with poor reputations, but it may not continue in that way as users push down the paid search ads from undesired websites.

So Google certainly haven't been quiet this year in making sure that my prediction hasn't come true. One could argue that they've made themselves much, much stronger than they were in the past (which was pretty damn strong).

2. The Death of Online Newspapers

Erm...

Yes...

I'm going well so far.

So it turns out it wasn't the death of online newspapers. The MailOnline hit 80m unique browsers in a month in November 2011. This is up 73% on the previous year. The Guardian was up 59% and the Independent was up 36%. So how did they do it?

Well the Mail decided that rather than being niche it was going to go as mainstream as it possibly could. 55% of their unique browsers now come from outside the UK, although 67% of their page impressions come from inside the UK suggesting that they are attracting a lot of people who are just floating past (probably through search). This is a pattern that is matched at the Guardian.

So maybe some more newspapers have gone behind a subscription model? Well if anything it has gone the other way around. The News of the World was News International's attempt to see if it could make a tabloid work behind a subscription model after moving the Times there. Unfortunately that experiment went by the wayside as they were forced to close amid a huge hacking scandal that hasn't quite reached its zenith yet. It remains to be seen whether the Sunday Sun will follow the same route when it arrives.

3. The Mobile Web's applications find wider use

I'm on safer footing with this one at least. This was an important year in mobile application building for one big reason. The big reason is that Adobe decided that they weren't going to update their Flash for mobile browsers. This doesn't sound like a particularly exciting move to the average bystander, but it has far reaching implications.

Firstly this is an admission by Adobe that they'll never get Apple to have Flash in Safari on the iPhone, therefore missing out on one of the biggest markets in mobile. Notably they are going to continue to help developers with Flash on applications.

Secondly they are going to start helping developers work towards developing for HTML5, which is standards compliant. This means that as mobile browsers get more sophisticated they will be starting to show what is on the actual web page, rather than developers having to code for different devices.

It's an interesting move because the shift in the mobile world is a significant one at the moment. The growth in the Mobile market over the last twelve months has been to Android:

Data from Comscore via fiercemobilecontent

Android of course being open source means that you can develop any applications you like for it and do whatever you want with it, including creating new browsers. My current browser of choice on the mobile is Dolphin.

4. The boom of use of Analytics tools

Last year at this time I said that there were 94 jobs in the UK for Web Analytics on Total Jobs. This year there are 119. Is that evidence of an increase in use of Analytics over the last year?

There was a continuation of the acquisitions, most notably as Adobe started buying companies that not only had technology, but also did consultancy. Google also introduced a 'Premium' version of the tool, which you can pay for to get support.

Who knows what will happen, but the industry is facing a major crisis in the coming year. As the rest of the world watches, Europe is introducing new regulations on cookies as the US watches on. If the US follows the EU example then it could well spell the end of Web Analytics as we know it. If the EU experiment fails then we're likely to find hundreds of US websites flooding the area that the EU companies lose competitiveness in.

Certainly it is an interesting time to be part of the industry! It's make or break time.