There is lots of data to analyse despite Google's Secure Search update

A lot has been written recently about the introduction of Google secure search means for websites. In fact, I've been at the fore of writing about it myself on the Digital Transparency blog that I've been contributing to recently, so I know all about the subject. Well, I know a little bit about it, which is what I'm going to share with you now.

In case you've missed the memo, Google recently changed the way that it did search quite drastically. They now make it so that if you are logged in to one of their products then your search will be done through a secure environment. You may remember that when Google plus first came out I theorised that they were going to use it to make search more personalised based on what sort of stuff you have plus oned. I still think this will happen, but the first thing that Google needed to do was obfuscate the url in the search bar, as this is the way that they are going to make it personal. If you put it in the url then you decrease the security for the user because this information gets passed over when you click on a link. They obfuscated it by making it so that the referrer doesn't have a search term associated with it at all.

So what does it mean? Well it means that in Google Analytics you now have (not provided) as a search term and in SiteCatalyst you have Keyword Unavailable.

In fact if you look at the graph above, it is quite a lot of visits that it is providing - up to 10% of the search visits for this blog, but many others are seeing more and it does really depend on your audience. My audience is slightly more technical (I think!) and more likely to be logged in to something like Google Plus or Gmail than 'average' audiences. But remember that no audience is average and the figures you are going to get will vary.

As Avinash Kaushik points out in his blog post on the subject - there is a wealth of information that you can find out about the (not provided) data in Google Analytics. In fact there is a wealth of information that you can find out about the visits from all your tools. You can find out exactly the same set of information about the visits as you would be able to from any other segment of visits.

Of course one of the points that Avinash makes is that it seems like it is a large value because it appears near the top of your keyword list, but it isn't actually because it is comprised of lots of small keywords. It is worth remembering though that this data could be slightly biased because of the segment of the users. This is called 'sample bias' and is something that you should all know about.

Of course this isn't the information that we want from the data. We want the original stuff back. Analysing it in this manner is useful, but it is only providing a bit of detail that we knew we could get anyway.

Given that we're not going to get the original stuff back, what else can we do? We can still use the data that we have to continue doing conversion analysis (with the data bias assumed) as Avinash says. We can do it for the terms that we have and we can do it for those that we don't have.

What we can do is start looking at how we combine together some different data sets. A recent change to the way that Google Analytics works was an introduction of some of the web master tools reports. You may be sitting there and thinking 'So what' at this point, because this is data that you already had, but this isn't quite true. In webmaster tools you only get one months historical data, whereas in Google Analytics it will be for as long as your connection has been set up.

This much longer data set means that you can compare it to your visits in a much more efficient manner. For example I can see here that the number of queries that my search terms appear in doesn't seem to have gone down recently, despite a small dip in the number of visits from searches.

The other advantage of this is that you can filter the reports by your search terms as well to produce some graphs showing the data for just a sub set. You can try to improve for a particular search term and then see how often you appear in results because of that. You can also see the effect of news or changes in search popularity.

The advantage of these reports is that I can quite handily also look at not just keywords, but also of the pages and how often they appear. This can provide me useful information on which types of story are generating search terms over a period of time. Those that are doing well can have more posts written about them and vice versa. This is useful when writing new posts - I can tell if they are gaining traction in search engines or not. My last post on Conversion Funnel Analysis in Google Analytics didn't seem to be getting much search traffic to start with, but I can see with this report that it is starting to appear in search queries now.

But of course you shouldn't treat these reports in isolation. What search terms are generating queries to a certain page? Well Google Analytics has hit its limits here, but you can go back into your webmaster tools and start looking at that data too. This has the ability to break the search phrases down by page or vice versa (over a smaller time period):

So we have two different data sets that will give us lots of different bits of information that we can use in lots of different ways. It is a shame that we've lost some data, but there is still loads for us to analyse.

How to do Conversion Funnel Analysis in Google Analytics

Last time I suggested that I should write a blog post about funnel analysis in Google Analytics, because all my conversion funnel analysis blog posts had been successful. So this week I am writing a blog post about conversion funnel analysis. It's almost like this blog has a plan week on week, isn't it? It doesn't, I can assure you that next time I'll be sitting with the computer in front of me wondering what on earth I should be writing about again, but there you go!

In case you've never come across this blog before, one of my early blog posts was on conversion funnel analysis: When, How, What, where I discussed what conversion funnel analysis is and what you should do with the data. I then followed it up with a post about how you do conversion funnel analysis in SiteCatalyst, which not only showed you how to do it in SiteCatalyst using fallout reports and so on, but it also expanded a bit on the principles to show that it wasn't as simple as it may seem. If you are interested, I also wrote a blog post on the sister website to this one, Digital Transparency, on how conversion rate analysis is about targeting as well as testing, which touches briefly on funnel analysis.

How To Create Funnels In Google Analytics

Creating funnels in Google Analytics is relatively easy. Relatively easy in the sense of the process is easy to set up, but it does require a little bit of thought. More of that later. Firstly though - this is how you set it up:

• Click on the settings button in the top right hand corner of your page
• Click on Goals in the second horizontal menu (next one to assets)
• Click on one of your free 'goals'
• Choose URL destination
• Check the box for the funnels at the bottom
• Add your goal and funnel stages

This process means that you need to set up your goals and funnels in advance and you need to think about the funnels that you are going to want. There is no retrospective data processing involved in this.

Note that you will be asked whether you want the first step in the funnel to be compulsory. If you check this box it effectively means that the goal will not convert unless the user has gone through this step before converting. It's a bit of an oddity, because whilst the users have to go through step one to get to the goal, they don't have to go through step two or step three. This is a vital bit of information because it relates to not just your conversion funnel but all your goal reports as well. As we'll see in the next section it isn't vital that you do it for your conversion funnel.

What Do The Reports Look Like?

Getting on to the more exciting part, you really want to know what the reports look like. Well you can access them in the left hand menu in the Conversions tab.

So these reports don't really visually represent the way things move through in the way that SiteCatalyst's fallout report does, but it does give you some interesting information. Because I didn't select that my steps were required in the report we have the options of seeing how many visits enter into the process halfway through as well as how many drop out halfway through. 

This relates back to our option for having the first step as compulsory. If you selected it then your process will be limited to just those who also went through the first as well. The second step in this case will only include those that went through the first step. An oddity of how the funnels works means that your second step could quite possibly be lower than your total for the goal (if the users bypass it).

This is an interesting concept that I expanded upon on in my last post on Conversion funnels. Not all those visits that get to the goal will have come through your funnel route. This is especially true where there are large volumes of content that your users could arrive at halfway through.

But this process is very useful when you have a payment process (shopping cart, credit card details, delivery address, confirmation, etc) to be able to look at how people drop out of your process. This can help inform you where in your process people are dropping out and you can look at improving this and monitoring it over time. This is the vital part of the process - you need to change something and monitor how it changes over time.

What Should You Do With The Reports?

Well normally I'd sit here and tell you to segment, segment and segment some more, but in this case we can't do that because there isn't an option to put your advance segments on here. What we can do though is put in some custom filters to create alternative reports. It's a bit of a fix for something that should be done automatically in Google Analytics.

So if you create a new profile in Google Analytics for the top segments that you are going to be interested in using the profile builder. Don't forget to make sure that you have a standard filter that doesn't exclude any of your data, otherwise something may go wrong and you won't have the raw data in the background.

Using a profile of your data is important when looking at funnels. Each of your segments is likely to react in a different way when going through funnel and one of your jobs is to try and work out a way of getting each of your segments to get through the process as best as possible. This may mean that you need to have more than one process for each of your segments, but this is all about targeting as well as testing.

I'd also recommend creating a diagram like the one I suggested in my last post on this subject:

Creating a visualisation like the above will allow you to work out if there are steps in the process that you are ignoring. Are people moving in and out of the process all over the place or is it really a traditional funnel? Does one of your sources into your goal provide more of the traffic than others. Should you be focussing on a different journey for your users or the one that you've set your funnel up for?

Don't forget that you can set up several funnels for the same conversion point (if you have enough spare goals). Don't be afraid of setting up more than one that are very similar. You'll get lots more data that will give you a better clue as to what you should do to improve the process.