Good Behaviour

I whole-heartedly agree that visitor engagement is a concept that needs to be considered as an aggregate of several elements covering both data and, crucially for me, context. I also think it is quite subjective.

I have recently done a piece of work for a client who put a new site redesign live at the beginning of January. Looking at the data before and after the live date there were three very clear changes:

1. Performance to the required goal has dramatically increased

2. Average time spent per visit has increased by c. 50%

3. Average pages viewed per visit has more than doubled.

I was only interested in comparing the data within this one particular site and not with others in the same industry since I accept that competitors design their sites slightly differently, may have different goals and different acquisition strategies - and so may view engagement differently. For this purpose I was interested in our little world only and I will try to justify that later.

Looking at the post redesign data I was initially tempted to think that if overall site performance, as measured by conversion to one specific goal, had increased at exactly the same time as a change occurred in the average visit length and pages viewed per visit and that the occurrence of that change was at the time of the site re-launch then it could be said there is a correlation between the three.

In an effort to try and filter out as much noise as possible, I looked at one referring source which has been a constant over the past 10 months - pay per click marketing. I also know from looking back over the ppc performance data in this particular business that seasonality in market demand appears to have a limited impact on conversion.

Looking at just pay per click (from Google only) the results were the same - a marked increase in conversion occurring at the same time as a marked increase in average visit length and PVs/visit.

Theo Papadakis talks about the idea of positive and negative engagement in an article recently posted on Occam’s Razor by Avinash Kaushik. I like this idea and would consider what I have seen here as an indication of positive engagement.

Looking further into content popularity it became clear that the new internal search function had started receiving much more traffic and now forms the backbone of the site’s navigation. This element of the site functionality was given much greater prominence in the new re-design.

So, what can be observed?

1. It’s easy to see that all 3 key changes occurred at the time of the site re-launch (in this respect we were lucky to have such a marked even to punctuate the data)

2. The same behaviour appears to be the case with a single source of referring traffic.

3. The increased conversion, average visit length and PVs / visit appears to be linked to a change in the sites primary navigation

4. Seasonality in the business cycle can in the main be discounted

This all points to the suggestion that visitors who convert tend to spend more time on the site and view more pages per visit. Given that the conversion goal is a positive outcome for us, then a simultaneous increase in average time on site and average pages viewed per visit must also be positive suggesting that visitors are more positively engaged with the content.

What do we do with this engagement?

I don’t propose to use it as an indicator to drive change in its own right. I see it as a “soft”  indicator; I prefer to think of it as a stalking horse. One which will prompt further investigation should a significant change occur. Additionally, where we have other referral sources I would like to use it to help assess relative value.

One final factor that will have to be taken into consideration and which cannot be accounted for so soon after the site launch is the novelty factor of the new site itself. This particular site sees a high proportion of returning visitors and customers, because of that we will have to see if the new re-design has prompted repeat customers and visitors to stay and look around partly out of curiosity. This should be born out in time.

My view is that engagement should largely be considered on an individual site by site basis. That is why I prefer only to look at engagement in the context of one particular site over time. It may be interesting to compare with other businesses but brand recognition and loyalty will most likely skew results to some degree regardless of site design.