Analyzing the Analytics

Everybody is aware of some simple facts in the Analytics landscape: An increase in Page Impressions is good. A high bounce rate is bad. Are you sure? Shouldn´t you educate yourself on additional synergies within those facts? It is too naïve to say that generating more Page Impressions than the day or the month before is a positive development in your digital channels. An important aspect of Key Performance Indicators for example is that they all should be set in comparison with a time interval to be able to judge the growth or decrease of the figures.

In many metrics this is a key aspect. So, again: is an increase of Server Calls good? Simple answer is: you do not know. If you want to find out – compare. Not only with your historical data (and then of course with the same time frame a short period ago: yesterday with the same day a week ago, instead of yesterday with the day before). You also have to compare your data with the market trend.

And I already see some of you readers have three arguments against this comparison:

  1. We are sooo unique. No, you are not – Full stop. There is competition.
  2. We do not have the market data – Valid point. Lots of industries have some public overview of market data available, for example media or finance, but yes, this is hard to find sometimes.
  3. We are not able to compare the data in our tool – Go select a professional solution. Not a tool.  The high-end vendors will very much be able to integrate market data via API into the system and have clear overview of the comparison between your company data and the market information.

Back to the metrics and to start with an easy one; an increase of server calls is good. We all agree to add additional aspects, see above. What if the Bounce Rate is rather high? I would say: perfect – well done (if we talk about a specific landing page with no further call to action). I would say, oh my god, if we talk about an eCommerce shop. The perspective makes the difference, as well as the website goals. Depending on content and goal a high bounce rate can theoretically be the ideal set up: if you own a company that is purely concentrating on services and you deliver all needed info on a specific page, then there is no need for the visitor to click further: he received all info he was searching for. Job done.

Another metric to be discussed is concerning Streaming Media content. Is it bad if a video was interrupted in between or has not been viewed until the end? You simply can not tell. It all depends on the placement of the content. If the relevant content has been seen (and was maybe included within the first minute), then the interruption of the video after this minute is less relevant and alerting because the main message was received. So, if the video is 5 minutes, the completion rate is only less than 20% but you do not care, because the main information has been covered.

How about long duration on a website? Good or bad? Again: it depends. If you have fascinating content on your website, lots of videos and detailed articles a long duration would seem to be more likely than on a news site. But the duration can also be high on a news site – because your editor is using the wrong words or too complex descriptions. If you are able to check the duration per article and check this with the number of words for example (again: good analytics solutions can combine different data sources and get you a more transparent overview) then you see if the complexity level per article has an impact on the duration.

At the end I would also want to touch a more detailed topic: picture galleries. Is it good or bad to add more pictures to the gallery? Does it depend on the area of the gallery, meaning the content or the page? Is there a limit of pictures to show without annoying people to click through too many pics? Try it out! Analytics is your friend also here.

With the given experience and testing on current clients we were able to get some results. Far from being representative official figures, but good to check for your own website. It was obvious to see that see completion rate was always around 45-55%, no matter how many pictures were included in the gallery. That means: the costs for stockphotos and the effort involved can pay out. You might have asked yourself is it worth putting more and more photos to that gallery.

The answer in short is: yes. If you include 40 pictures it is most likely the visitors will click around 20, so 50%. If you include 15 pictures it is most likely the visitors click up to appr. 7 picture, so roughly 50%. The quantity is less important. It is important to include your most important pictures or messages within the first half. See the descriptive screenshot below to see a range of galleries from 3 to 40.The completion rate is quite the same, as said. So, how long should a picture gallery be: as long as you want it to make sure you cover all your stuff within the first 50%. Is it relevant in which parts or channels of a website or in which interest group this is shown: no. They all have around 50% completion rate.

PhotoCompletion_ComparisonGraph

So, how long should a picture gallery be? As long as you want it to make sure you cover all your stuff within the first 50%. Is it relevant in which parts or channels of a website or in which interest group this is shown? No. They all have around 50% completion rate.

So what have we learned?

  1. We are not unique – there is competition
  2. Market data can be hard to find but it is very valuable
  3. You need to be able to compare your data
  4. Find answers to questions that have not been asked yet
  5. Go and start with something. But start!

Comments

  1. says

    Hi Ralf,
    Web analytics is something that really needs to be analyzed carefully and it can indeed give you some important insights about your website that can help you in improving it to the next level. But unfortunately majority of us do not try to understand the facts from Analytics and just use to see whether we are getting traffic, referral traffic etc, but do not tend to spend time in drilling into deeper data.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>