More Web Analytics for Content Editors

Well I'd been writing the previous post for ages and decided that it was time to cut it in half and write one half that dealt with the metrics that content editors use and the second half was going to talk about how to use those metrics to work out the best way of creating new content.  This is essentially the crux of the matter is more or less what I wrote at the bottom of the post.  We want to be able to use the analytics we have on our existing content to work out how we should write or edit our content.  Also we need to work out how we are going to use that information to market our content to get the most visits and the best conversion, whatever that may be (and remember that you don't have to measure conversion in a conventional manner if you are not selling anything per se).  I think this is the process that the content editors should take when they have a subject to write about:


1. Find content on a similar subject

This may sound a little odd at first, but if you are writing a new bit of content, then you want to look back at your old bits of content.  No really, that is a great place to start.  Particularly you want to find things that were around the same sort of subject.  This is where having a good taxonomy will come in useful because you'll be able to search through your archives in a better way.

Remember that taxonomy doesn't mean navigation structure.  Your taxonomy should be a series of words that describe the content (like the ones on the left hand side of this blog), whereas you navigation structure should be grouping similar bits of content into a navigational structure that the user will understand.

Using the taxonomy approach it means you should be able to cut across many ways to find similar content.  Particularly you'll be looking for where the subject matter is similar or the same and/or where the topic has been similar or the same.  Eg if you are writing about the measuring SEO with Omniture tools, then you should look at your other SEO content and your other Omniture content.

2. Find the content in your web analytics software

This might be that if you've set up your content in good groupings in your web analytics tool, you can do step one with step two.  But I'm going to assume you haven't.  So the best way is to work out how the pages are named in your solution.  Remember that although in Google Analytics they are named after the url of the page, in many tools they can be completely bespoke.

 

Now really what you want here is a number of metrics.  You want to see visits, Entries, Single Access (or bounce rate).  In Google you can't do that, but remember in Site Catalyst you can choose your metrics in your pages report, so I'd advise sticking all three of them in there.

This gives you a good overview of how your content was accessed and what people did with it.  Particularly of importance is the entries v visits.  If you have many more visits to entries, then that implies that previously the users found the content from elsewhere on your site.  If virtually all your visits entered the site at that page, it shows that they were on other websites when they found your article.


3a. More people found your page through your site than from other sites


This means that you need to look at the sort of pages that they got traffic from.  This is really easy in virtually all web analytics tools.  You don't need to go looking at all the paths, you just need to look at the previous page report for this page:

In Google Analytics you need to choose your page and then click on the 'Analyse' drop down to choose Navigation summary.  It'll show you how many visits came from external sources again and how many came from internal sources.  As you can see in my example, most of my visits coming internally came from one particular other bit of content.  When trying to drive traffic to a page, you need to explore the possible avenues of getting people to the page by getting those links in other bits of content. 

Can you get more links on this old page?  Just putting a new one at the end as a 'related link' might help. 

Why did the piece of content you are looking at work so well with the other bit of content?  Can you find some similar content for your new piece of content?

In SiteCatalyst of course finding out these paths is even easier.  You can click on your page name, then on paths and then previous page.

3b. More people found your page through other sites than from your site


I'm fairly confident that this is the most likely situation for most of your content.  Given now the increased use of Google, many users will just be searching for stuff and your site will pop up.  Alternatively you'll have got traffic from a news aggregator or even better some nice chap will link to you from his blog.  In this case, you really want to know what those sites are.  This is where you can use the fact that other sites have linked to you in the past to help build links for your new bit of content.  Of course, it does no harm for your research for your new bit of content if you can find related analysis on the web.

And of course if you put a link to their content in your content, it isn't going to encourage them to link back to you at all.  No siree.  Nope, not even if you send them an email and tell them.  I wouldn't advise doing that.  Or if it is a blog, don't put a comment up and tell them that you found it interesting and you've just written something else with a link.

I don't think I can stress enough that we aren't in a 'build it and they will come' world.  You've got to market your website.  If that means going out begging, stealing and borrowing links where ever you can, that is perfectly fine.  The more people who read your content, the more links you get, the better Google thinks you are an pushes you up the rankings.

If you do discover that the majority of the traffic last time was from google - look at the search terms:

 

 

Find out what sort of things that you can be writing about.  These key terms will help you pick up long tail traffic.  Especially if you link back to the content with those keywords in anchor text.  Particularly see how this fits into your keyword footprint.

4. Look at what links people click on in the content


A long time ago I wrote about how active viewer was awesome.  Well HBX may well be near on defunct, but that doesn't mean that there aren't new toys to play with.  In Omniture you have the equally brilliant clickmap.  I have to say, it'd be nice if Omniture could update this at least so that they could give it the right look and feel to fit in with v14, but it does its job still:

I have to say I like this tool better than active viewer.  Probably because Active viewer suffered from one of the issues that you ended up having the same anchor text repeatedly on a page and then it would group them all as one.  With clickmap, they describe the location of the link as well as the url, and you get individual figures.  Admittedly if you try and view it in any other format than on screen it becomes near on useless for dynamic sites and sites without a good url naming structure.

Why is this important?  Well, really this is all about 'persuasion architecture' and 'scent'.  Your pages are very much part of a journey that a user takes.  This journey doesn't begin when someone enters your site at the home page and exits having bought your product.  The journey begins at the very first tentative step of turning on the computer.  The user doesn't want their journey interrupted, but if they feel that it is there is a very easy option in the back button.

So that means that you are looking at building a page that gives the user the right set of links on the page for the place that they came from.  By analysing the links on the previous content, you can work out for that article, what type of things people were interested in.  This is key - you are trying to get your users to stay longer on the site.  They'll get a better impression of your brand and are more likely to convert (buy, sign up, etc) if they are on the site.  If the content has a high bounce rate, the users haven't seen enough to want to convert and give you money.

That back button is very powerful.  Be wary of it.  Make sure you know how you are going to market your content before you write it so that you can get the best links in it.  If you have personas this can be built in here as well.  Writing content on the web isn't just about writing good, readable content.  It's about tailoring it for a very niche audience and providing a very valuable added extra.

Web Analytics for Content editors

Recently (ok, it seems like it was recently, but actually it was almost a month ago now - where does time go) I was reading a post on the web analytics forum about training in Web Analytics for editors.  It was interesting as it was coming in at a time in my current job where I was thinking about that again, having done it in my previous job working at a B2B news organisation.  Whereas now the copy isn't news, but is very long lasting and needs to be changed/updated, whereas there is also new copy coming on all the time.  So I dug out an old presentation that I'd been given and started to write this blog.  The question is, what do content editors need to be looking at in Web Analytics data to help them do their job?

Just to be clear about this - Content editors should be interested in how they affect the company's bottom line.  For my previous job this was maximising their profit through whichever business model they were going for.  For my current job there are a whole host of things, but because the website is not for profit, they can be a very specific group of success events.  Needless to say, for both the web analytics data is of vital importance in terms of calculating revenue and whilst publishers and editors have long be looking at page views (as a proxy for ad views) to show how much money their content is making, maybe they aren't aware of how they can use the data to improve the site.

1.  Every Content Editor should know what a page view, visit and (unique) visitor are.

The Web Analytics Association defined these over a year ago.  I've got my own simplified and not technically correct versions - but this is how they should be understood:

• Page View – A page loaded by a person
• Visit – A session of looking at a website/page by a person
• Visitor – A person looking at a website

Actually it's not even that simple.  That is how it should be.  This is what we really count if you use javascript tags and cookies:

• Page View – A page with a tag being loaded
  
• Visit – A series of page views by a unique user until 30 minutes of inactivity
• Visitor – cookie

Now I'm not going to go into the whole cookie debate.  What I do tend to tell people is that Visitor data sucks.  Yes that's right.  It's fairly useless.  It's the most inaccurate thing you could ever imagine (people deleting cookies, using more than one computer, etc, etc).  Don't use it.  Use page views and visits.  They don't suck.  They're fairly accurate.  Bravo!!!

Well actually when it comes to page views, you have the problem for a page that sometimes someone will click on a link on that page and then come back to it later, so you might get more than one page view for the same page in a visit.  You don't want to count all of them, because you're only really interested in the fact they saw it in a visit, not how many times they saw it.  So lets use visits instead.  That way if you view a page once or more in a visit, lets include it only once.  Huzzah - visits rule the waves.  Before anyone comments - having multiple views of the same page if you sell advertising on that page is good (you have more advertising to sell after all), but in terms of a content editor - you just want to know that the person has viewed the page.



2.  Know what Entries, Exits, Single Access and Bounces are


These are very hard to write about without giving a real description to your journey, so I like to do it in that way.  When you're browsing through the web you end up on a website at a specific page (either by clicking on a link, typing in the url, or having bookmarked the page).  That page won't necessarily be the home page, but it will be your entry page.

When you leave the website, the last page you loaded is now your exit page.  This is the point where you've said "Ah, I'm bored, I'm going to look elsewhere!!" or "Brilliant, I've done what I want, I'm out of here!" or "Damn, who is that at the door, I'd better go and answer it, oh its uncle David just nipping in for a cup of tea and why haven't I finished ironing my shirts, I better just nip away from the computer."

What happens if you arrive at a website and leave without viewing any other pages?  This is what your single access pages are.  These are people who bounce on and off your website.  You can claim your bounce rate is the percentage of the people who arrived at that page, who leave without viewing another page.  Note that you could view the same page repeatedly before leaving and still be a bounce.  This means you need to be wary when setting up your analytics tools if pages are going to have the same page name (Omniture/HBX/others) or the same url (Google Analytics/others).

3. Work out how you've set up your content groups

Every web analytics tool makes use of them.  In Google Analytics, you go to the content breakdown:

Here you can see your page views and unique page views (make them into visits properly and then call them visits dammit.  Do you here me Google?!?!?).

In SiteCatalyst there are other ways of doing this and it will be specific to your set up.  For our site we make use of the breadcrumb trail that describes the levels of navigation on our site (and I think that is far more useful than the breakdown of the url as above).

Remember that when you get to these reports, you can drill down level by level, to give you deeper insight into the visits to each of the sections, as well as giving you 'top level' data.


4. Drill down to individual pages



Now really this is where editors earn their crust.  The above was just proving that they could measure the long tail (use the above content groups along with the top 10 visited pages to show that you have a long tail.  You do have one - I'm certain of it). What we really want to do is look at the pages themselves and see what they are doing.  Firstly we need to think about metrics.

• Visits to a page are important - you want to know how many times it was visited
• Entries to the page are important - you want to know how many times it was the first page a person saw
• Single Access visits are important - you want to know how many times the first page was the only page and you didn't cause a user to delve deeper into the site.

This brings up two new metrics:

• Entry rate - percentage of a page's visits that were caused by the user arriving directly at that page
• Bounce rate - percentage of entries that left without viewing another page on the site

Clearly a page with a high entry rate has a reason for having a high entry rate.  What was it?  Where did the visits come from?  And if they have a high bounce rate, why did that happen?  What can you do to decrease it and encourage your users to continue on in the site?

This all brings me on to the next topic, which I'm going to save for another blog post (this one is too long already).  How do you put all these metrics above into a method of increasing the visits to your site?  This is really the crux of it - the above is just giving you a foundation on how to work this out.