A Breakfast Guide to Web Analytics

Anyone who goes to Breakfast networking meetings will understand the need for a short talk on the subject of their business - and at 7 in the morning, it doesn't need to be heavy going. So this article is based on my breakfast guide to web analytics....

Runninga website without looking at and using the stats it generates is like trying to win a race blindfolded when the competition has 20 20 vision.

So how do you design a process that is easy to maintain and results in continuing improvement of your website’s performance?

The first time your looked at your web stats, the chances are you were overwhelmed by the sheer quantity of data. “Where do I start” and “What do I do with all this information” are common reactions.  Hours are whiled away first being horrified at the Big Brotherness of what you can find out about your visitors, soon followed by amazement at the phenomenal variety and imagination used by visitors when using a search engine to find your site.  The result however is probably that you learn nothing very useful or meaningful about anything.  Worst of all, no action is taken!

So the first thing to realise about web stats is that the greatest challenge they pose is INFORMATION OVERLOAD. 

 

Web stats are a bit like 42 – they are the answer to life, the universe and everything there is to know about your site – but the key is knowing what questions to ask.  Fans of the Hitchhiker’s Guide to the Galaxy will remember that a very large computer, viz the earth, was needed to deduce the question.  Luckily web analytics is not quite that difficult.

The first step in dealing with this problem of information overload is to DEFINE YOUR SITE’S GOALS.

 

You should of course already be clear about these before you build your site but if you aren't, then now is the time to think through exactly what is it that you are trying to achieve on your site and how you want to lead your visitors through it.  Of course, many sites have more than one goal.  No problem, define them all.

 

To make things easy, let’s say the purpose of my site is to take orders.  So I took 20 orders online yesterday!  A meaningless statement.  If I were selling jumbo jets then that would be fantastic.  A site like amazon.com would no doubt be less ecstatic.  So CONTEXT is vital.  Raw numbers are out, always report data compared with last week, last month or compared with a target or some other ratio or fraction that gives the information meaning.

Last February orders were down 50%.  This might be in context but it is pretty useless information, it wouldn't have been last March though!  So data needs to be reported in a TIMELY fashion.  In other words it has a short shelf life and wilts quickly.

 

So we are now reporting results that are in context and timely, but there is still something missing – ACTION.

 

The concept of KAIZEN could have been invented for web analytics.  It was actually invented by Japanese manufacturers in the 1940s to improve their industrial base.  No matter, it is ideal for our purposes.

Kaizen refers to a cycle of continuous incremental improvement.  So first we measure a metric, then we report it, then we analyse what it means, then we take action (if necessary) to improve the value of that metric.  Then of course, we repeat the whole cycle at whatever schedule is appropriate for the website.  It might be monthly for some sites, weekly for many, daily for others.

 

But now we need to think carefully about what it is we want to measure.  Remember back at the beginning, we defined our site’s goal or goals. These of course need to tie in with our business or organisation’s goals.  The idea of KEY PERFORMANCE INDICATORS is well established with respect to running an organisation – and they are equally relevant and appropriate to running a website.  So each site needs a list, very often a very short list, of KPIs.  Every site needs a KPI list that is specific to it and its goals.  As far as possible, website KPIs should correlate with the overall organisational KPIs so that the running of the website is linked in seamlessly with running of the organisation.

 

Although every website needs its own KPI list, there are some KPIs that are relevant to a wide range of different sites, so let’s have a look at a few of these.

 

 

 

TRAFFIC – most sites want to know whether traffic is going up, going down or staying static.  We can delve deeper to look at, for example, the ratio of new and returning visitors and how many visits returning visitors make.

The number of CONVERSIONS is another core metric appropriate to all sites.  Conversion will be defined in many different ways – a transaction on an ecommerce site – but every site must have a purpose and when that purpose has been achieved, then a conversion has taken place.  For an information based site, it might be as simple as any traffic that stays on the site for more than 20 seconds.  Many sites define the visiting of a certain page as a successful visit – and thank you pages are almost certainly goal pages.

 

PAGE STICKINESS – this is one of the fundamental KPIs and one that is often overlooked.  It is simply the proportion of visitors that arrive on a page and then navigate onto another page on the site.  It is the converse of bounce rate, but the problem with defining bounce rate as a metric is that if it goes up, things are going wrong (almost always anyway) and it is far better from an information assimilation point of view to define all our KPIs so that an increase is good.

It is vital to know the stickiness of a website’s home page. For the majority of sites the home page is the most popular point of entry.  It is tempting to try to improve conversion rates by working on the buying funnel but an increase of 1% in the numbers of visitors moving on from the home page has far more potential to improve sales than a 1% increase in numbers completing the buying process.  Of course increasing both by 1% is the best solution!

For any site that is using Pay per Click to generate traffic then landing page stickiness is crucial in maximising return on investment (ROI).  With a PPC campaign, you have control over the keyword used to attract the visitor to the site, you have control over the ad content they click on and you have control over the content they see on entering your site.  The result should therefore be that visitors like what they see and conversion rates should be high.

It is difficult to define what are acceptable or good values for page stickiness but I usually start to worry if it falls below about 55% or a home page and anything below 75% for a PPC landing page.  These are only rules of thumb though and especially for the home page, the value will depend on a lot of factors, including the type of site, the market area and what the site is designed to achieve.

Revenue and all the KPIs related to it are crucial for all transactional sites.  Keep in mind though that the purpose of the site is profit not revenue, so define KPIs with this in mind. 

 

ATOS – average time on site is a KPI that many web analysts swear by since it provides a pretty good indication of how valuable visitors to your site think your content is.  It is made even more valuable when you segment your visitors so that you start measuring ATOS for visitors who used a particular keyword or visited at the weekend for example.

 

REFERRING SITE strictly means any site that refers traffic to your site which of course includes search engines. However, I tend to exclude search engines from other referring sites and treat the two separately.  Looking at the traffic from a particular referring site, both quantity and quality (as measured by ATOS for this group for example) takes the guesswork out of the decision whether to renew paid listings.  If you know how much traffic that site delivered and how much of it converted, the decision is simple arithmetic.

There are an almost infinite number of KPIs that can be defined.  Eric Peterson’s Big Book of KPIs does a great job of listing an awful lot of them, but the crux of all this is defining the very fewest you need to answer the core questions of how well is your site working and how can it be improved.

Before we leave the topic of KPIs and Kaizen, there are just a couple of things to keep in mind.  As you compile your web analytics reports, always keep RECORDS of what you tested out.

If you think conversions will rocket if you change the colour of your landing pages to blue (and colour does make a huge difference) and it turns out to have the opposite effect, record it!  It is very easy to have the same great idea six months down the line having forgotten you have already tried it out.

Try out new ideas ONE STEP AT A TIME.  When it comes to the action phase of the Kaizen cycle, make sure you can correlate a result back to an action, and this means changing one thing at a time.  So if you are trying to improve the home page stickiness and what to try out a different colour and also moving the position of an accreditation logo to a more prominent position, do one , monitor results and try out the other and monitor those results.

Finally a quick word about DATA CAPTURE.  Web analytics is becoming an incredibly competitive market and likely to become more so.  Google Analytics has shaken the market up and has brought web analytics into the mainstream and make it far less nerdy.  It is a very powerful product and free is a hard price to beat!  You do need to realise though that you are giving Google access to all the information there is about your site including, if you have a transactional site, exactly how much money you are making.  Google already knows an awful lot about us so perhaps it doesn’t matter! Perhaps it does.  Your decision.

Google Analytics uses page tagging, that is you need to add a small amount of javaScript to every page on your site, then log into the GA interface and pick up the results.  Page tagging is now the most popular technology to capture analytics data but there are others.  ClickTracks for example offers a log file analysis product which is a one off purchase which you run on your own machine.

For sites that need a level of customisation, then there is a wealth of products as you go higher up the market. 

But remember, how you capture your data is irrelevant so long as you have the ability to capture the data you need.  It is always how you analyse and use that data that is important.

So finally, the key to making web analytics work for you is to ask the right questions, DISTIL the data into manageable and meaningful KPIs, and finally, when it comes to web analytics