Why I'll be taking GooglePage Rank off my toolbar

You kind of new that at some point someone over at Google was going to do something a little silly and wipe all the good marketing that they've done over time into the dust bin. Ok, ok that is a wild exaggeration, but news broke recently when sites saw changes in their page rank, about how Google had started to penalise some sites for buying and selling paid links.

Whilst the searchenginejournal got on their high horse about this straight away, they do make some valid points:

If Google are going to keep changing pagerank of a website, why should I use this as a measure of how good my site is?

Firstly, it doesn't reflect how much traffic a website is going to get from Google. PageRank, as we keep being told is only a minor part of why a website will rank for a search term.

Secondly, half the pages on my site don't have a PageRank anyway - they are just too far removed from the home page to pick up any of the 'juice'. All these page get traffic from search engines (usually far more than the home page). If PageRank were to actually mean anything, surely they wouldn't get any.

I think I am slowly starting to form a similar opinion to Jim Newsome, that maybe Google should have sent out some warnings to those sites before doing creating the drop, I'm actually going to do a bit more a shoulder shrug on this one. Our PR has fallen. Who cares?

Lets measure our Search Engine Optimisation work in a completely different way:

What do we want our SEO to do? We want it to send more people to our website.

How do we measure this? Lets count how much traffic it has sent to our website using our web analytics tool.

Has our SEO work been a success? Well lets look to see whether our search engine referred traffic is going up or down. More importantly lets do it in 2 ways:

Firstly - lets measure how many visits search engines have created, then lets use this as a comparison to the total traffic to the site.

Secondly - lets measure how much money this traffic has made us (through either conversions or if your website doesn't sell anything then you'll need to have page impression generated).

Then, and this is the most important step, measure this over time. Trending is the most important thing you can do.

Hey - lets delve deeper in. Take out your brand terms if you rank number one for them (they are the easy terms). Segment by search engine. Segment by search term. Start doing the analysis of which terms you are missing. But all this still needs to be trended.

Getting your users to engage

There was a bit of conversation recently around the measurement of engagement within the web analytics industry and I think that this has been brewing around this company for a while too (although under a different name). Whilst Avinash and Steve have been doing their impressions of Frankee and Eamon, this is actually quite an important issue.

Whilst a couple of weeks ago I wrote about how you can use HBX to find out which links people are clicking on a page and before that how you can use active viewer to have a little 'heat' map of where people are going, this doesn't give an overall figure that you can put infront of your manager and say this is how 'engaged' our users are.


So how do we do it? Do we get Research to do a survey so that we can find out how much our users love our site? Can we measure it through HBX? Should we go and ask some of them ourselves?


Well, to cut Avinash and Steve short, yes is the answer to the last three of those questions. But in order, here is why:


Do we get Research to do a survey so that we can find out how much our users love our site? There is no way you can find out what your users think of your website without asking them. What is the best way of doing this? The best way is to set up a survey on your website and ask them. "Are you engaged with our website?" Answer yes or no. I know this sounds simplistic, but when you ask them the same question in 4, 6, 8 or 12 months time, you can find out if the efforts you have been putting in to the site have made a difference. Hell, you can even say "Are you engaged with section A of our website?". Ok that is a bit simplistic too, maybe we should look back at Avinash's blog and question what we want them to do on each part of the website and then ask if they felt like they'd achieved this: Applying for a job (finding one first); reading lots of relevant articles, etc.


Can we measure it through HBX? Again this is something we should, and can do (here I am agreeing with Steve). How do we do it? Well, sitewide we can look at average pages per visit (page views divided by visits). Well this is ok, but we know that some of our users consume loads of pages and some hardly any, so lets do it differently. Let's look at the number of visits consuming more than one page (ie they have at least in part engaged with the site).

Let's not pretend though - this metric is pages per visit, not engagement.

Another method that we can use is to measure the length of time that our users stay on the site for. The theory behind this one is the longer they stay, the more they have consumed (bearing in mind different readers have different reading speeds).

This metric is now showing not only an average, but also how long for each of the 30 second time periods. Now my feelings on averages should be known by now!!! What we want to do is look at the percentage of visits that last longer than one minute. These users have interacted with our sites long enough for us to say they've spent a while on a our site. Unfortunately our average will be skewed by those people spending hours (and there will be some) on the site. Remember though - this metric is not engagement, lets tell people what it is.

Should we go and ask some of them ourselves? Well now that we've got a survey result telling us what people think on average (going up or down over time) and we've got some HBX stats on how much our users consume. What do we do with these results? We've made a couple of changes. They've worked ok. They've made a difference to our metrics above, but what else? What you need to do now is go and ask some specific people. Get them in a room, write to them in an email or ring them up and ask them what they think. They'll be candid and they'll let you know what actions you should take. These usability studies will tell you where, when and how your users are engaging.

Using Link Analysis, you can work out what links work best in your articles

Do you know where people are clicking in your article? Do you know which links work best on a page? HBX's clever link analysis tool can tell you all of this, so that you can work out which links work best in giving your users related information, whether that be linking to other pages within the site, or even to external sites. So how do you do it and what do you do with the results?

Getting the results is easy, working out what to do with them is the difficult bit. To get to the page in HBX you go to Navigation=>Link Analysis=>top links per page, as shown in the image below. This will then give you a page telling you all the pages you have where users have clicked on a link and how many clicks they have done.

This information only really becomes useful though if you drill down to the next level. We want to know the exact links for this page that have been clicked. Well by clicking on the little grey arrow next to the page name (and then on links), you can drill down to the next level. This will tell you all the links that the users have clicked on and which ones are working best.

How does this help you though? You can't put any more links in it, because you've saturated it? Well next time you come to put links in a story, you can think back to this experience. "Hey, I wrote a story on a similar subject x weeks ago," is what you'll be thinking, "why don't I look back at that story to see which links drove traffic onwards in the site and which didn't."

Why not also check out doing it in active viewer as well. Then you'll be able to see a visual effect based on the actual page.

Britney's Dog Menaced by Sharks

A common theme around the Business at the moment is moving from writing articles for the magazine to writing articles for the web. More specifically what we are finding is that articles that are going into the magazine are also being put onto the website and being asked to perform. Whilst Joel Achenbach from the Washington Post is worried about this, that shouldn't be the case for RBI's writers.

"Good writing remains good writing regardless of platform. The Web tends to be a chattier place, more off-the-cuff, but it is still a place where readers appreciate a well-crafted sentence, a nuanced thought, a fully elucidated thesis and commentary undergirded by fact, honesty and a generosity of spirit.


And the readers who don't like that stuff? A buncha jerks."


Well ok, maybe not, but we can look at it a different way. Yes good writing remains good writing regardless of platform. Readers do appreciate well-crafted sentence. The problem on the web is that we can find out whether our readers think that it is a fully elucidated theory and commentary undergirded by fact.


How do we do this - well the simple fact is that if users like your content, they'll link to it. The more links you get to your article, the more people like it.


Ok it's not that simple either - the people who read your magazine probably won't be the ones who are linking to you. The surfers who stumble across your website probably won't the first time either. Even those that do link to you, may not be giving you the response you are looking for. Some websites will give you more traffic than others in driving people to your website. Experience will only tell you which ones have worked best.


Getting that experience is easier than you thought. Look at the articles that are getting visits at the moment.

Traditionally we have only been interested in page views and visits (maybe). Lets look at it a different way: You get the most visits to your article if you can get the most people entering the site at your page. Gone are the myths of being high on the homepage driving visits to your article. That column that says 'Entries' is now your friend. These are the people who have come to the site for the sole purpose of looking at your article.


So where did they come from? Well that little grey button next to the page name will tell you. Click on it and choose referring URLs. You'll get a page showing you where the visits arrived at the article from. Use this information next time you write an article on a similar subject - these people probably would like to see that one too. Tell the owner of the website. Send them an email or post a link. You could even reference them in your article, they're probably writing around the subject too.

Is HBX active viewing awesome?

Yes it is, according to Stephane Hamel. However he was looking at it from a slightly different point of view than from how we will be looking at it within a publishing company. So how will we be looking at it differently?

Well for a start most of our site isn't form based application, it is content driven. However some of our site is form based and this is where active viewing can become powerful in working out where our users are leaving our form pages. If a page is too long then we'll see constant drop out throughout the page. If there is a question that a user doesn't understand or doesn't want to answer, then we'll see a big drop off there too. Everyone leaving at the first question? They probably don't see the benefits of filling in the form.

For the content pages on our sites HBX can be even more useful, especially when looking at home pages. So you want to know what to put up on your home page? Well lets look at what is there already and how it is doing. By overlaying the links that our users have clicked ontop of the actual home page itself using the links button at the top of the active viewing screen (). New stories have had no clicks on, whereas the older stories are still ranked.

Also worth noting that the other stories that users have been clicking on appear in the boxes at the bottom and the nav at the left. It's also worth noting that these are ranks of links that have been clicked on and are still on the page. In some cases, the most clicked link may no longer be on the page. So if we know which stories are being clicked on we can work out what to put as the lead story and which stories should be located around it.

Whilst this is looking at a day so far so that we can work short term into what we'd like the stories on our home page to look like, we can also look long term. Which navigation gets more clicks? More importantly which navigation doesn't get many clicks? Abbreviations and acronyms in navigation are often confusing. You can usually find if a navigation item is confusing because it will have an unusually low volume of clicks.

Next level down for active viewing? Well lets look at the links that users have been clicking within articles. This is a key method for our sites for engaging the users in terms of stickiness and for giving the users what they want. There are of course many SEO reasons as well.

And then there are the other reasons you should use the active viewing, like wanting to know how many site page views/visits/unique users, individual page views/visits/unique users, more importantly individual page names and a load of other things that I don't have time to go into here.

So what if we can't measure conversion conventionally

If you don't own an eCommerce website, you have to come up with some other success criteria for your site, which don't necessarily have to relate to actions. An interesting post from Avinash Kaushik on this subject references Google Analytics, but it is just as easy to do through HBX.

These metrics recently came to a head when I was asked about the lifetime returning visitors metric in HBX. The lifetime returning visitor metric is a record of anyone who is visiting the site for the second time using their current cookie. If the user has already visited the site twice then they will not be added into this metric again. Hence reporting lifetime visitors as a percentage of all visitors is a little bit meaningless (you are only counting the percentage of visits that are arriving for the second time). A better method would be to look at the percentage of new visitors as a total of all visitors.

Daily/weekly/monthly returning visitors are visitors who are not new visitors (ie they have a cookie) but have not see the site before in that time period.

A better way of looking at this metric is to look in the Loyalty report showing the number of visitors who visit once, twice, etc during a time period. It is important that this is done over a relatively long period though - given a proportion of users will respond to emails every 7 days

Another important metric that we have been measuring is the number of page views per visit. However we need to get away from averages (the extremes will skew your average) and really we want to be looking at distribution of data. So a good way of measuring this is the number of one page visits (or if you want your graph to go up to the right, the percentage of visits viewing more than one page). Reducing this bounce rate should mean that we will get our users more involved in our content.