Good Behaviour

An argument often made in relation to website performance goes as follows:

If the conversion rate on a site is 2% then there is a 98% pot of untapped revenue? Technically yes but realistically no.

If 98 out of every 100 visits to the average e-commerce site don’t make a purchase then there is clearly an opportunity, but it’s unrealistic to think that the size of that opportunity is the entire remaining 98 visits - no sane person would think that anyway!! The question then becomes, what is the realistic size of the opportunity?

This is worth asking in light of the fact that site managers and marketers must spend money chasing the revenue and the amount spent often relates to the size of the opportunity.

It’s far more likely that some visitors are there simply to research either products or prices or both. Anything between 2% and 10% percent may admit to wanting to buy and it’s these are the guys that represent the real opportunity. The reality is that even a site wide conversion rate of 8% would be well above average for most e-commerce sites so it would be an impressive feat to reach that but this is at least a more realistic figure to aim for.

The best way to find out why they’re there? Ask them.

Just before Christmas the prolific Blogstorm wrote a post on what a week’s vacation taught him about web design. I was reminded of his post whilst I was away over Christmas and perhaps inevitably I’ve ended up with my own set of comparisons specifically regarding managing visitor’s expectations.

I firmly believe that specifically managing visitor and customer expectations is one of the most important things that can and should be done with a site. Here are 4 basic and probably rather obvious observations on managing expectation.

1. Do what you say you will do

After taking off from Lisbon to Maputo we had to turn back because the wing flaps weren’t working properly. Sitting on the tarmac in Lisbon an announcement advised us that we would have to disembark. We waited and after a while another announcement informed us that we should remain seated with belts undone whilst the pane re-fuelled; still not sure what the problem was and if we would still have to disembark at some point. Eventually we took off again rendering the first announcement redundant.

Meeting expectations of visitors to your site is critically important. In part this can be measured using the Bounce Rate metric. This flagship metric demonstrates how good a particular entry page is at meeting the expectations of visitors arriving from specific sources. However meeting expectation goes further than this. At every step in the visitor’s journey through a site there should never be a corner around which the visitor finds something that wasn’t expected. Don’t leave them guessing.

2. Avoid dead ends and information black holes.

As a result of our delay in Lisbon we were re-routed to Johannesburg before going to Maputo in Mozambique - our final destination but we weren’t told about that until we were over half way to J’burg. The flight was supposed to make those stops in the reverse order. I asked how long we’d be in J’burg and was told 25 minutes. That’s how long the stopover would be if it was scheduled, in fact we were there for nearer 4 hours but we only knew that after the boarding announcement came in Johannesburg.

Every page on a site should have a purpose, if it doesn’t then you have to ask why its there. Try and empathise with visitors to your site and ask what they would want to know as they reach each page or type of page. Then make sure the information is readily available to them in a clearly identifiable format - clever (Web 2.0) technology is not always the solution here.

Some people will look at exit pages to understand if a page has fulfilled its role, this can be misleading as all visitors have to exit a site from somewhere, but knowing the 2 or 3 pages that were viewed immediately before the exit page may help shed further light.

3. Provide relevant information before it has to be asked for.

As already mentioned, on arrival in Johannesburg there was no indication of how long the stop over would be. Additionally there was nobody in the TAP office to answer questions. We simply had to wander around and wait. All the time we were cursing TAP for being so useless.

When a visitor or customer completes an important action such as registering for an email or buying a product or service be sure to tell them know what will happen next and what they should do. Remember that visitors to your site will not necessarily (probably not) read the fine print so you need to make important information stand out, this in itself is not as easy as one might think.

4. Don’t over complicate

There’s no holiday related comparison for this one.

Design your site for your grandmother. Clever technologies are only good if they actually make life easier. If a potential customer has to spend more than a few minutes trying to figure out how to access a piece of content that will show off the product in all its unfettered glory then, unless you are offering a financial incentive, the opportunity is lost to all but the most determined.

Finally I should add that despite my experiences flying with TAP which prompted the above comparisons my thoughts are also, in part, backed up by a piece of work I did for a client just before Christmas in which we provided a link on the category and product pages to a simple piece of content which served to provide additional reassurance to the customer during the research and purchasing process. Prior to the link being there very little traffic had been reaching the content in question. Once the link was installed traffic to that area shot up, visitors were reassured and relative performance improved.

Analytics is not a means to an end in its own right. Even insight that comes from analytics isn’t a means to an end in its own right. Looking at analytics data provides some initial insight but often (in my experience) it has thrown up almost as many questions as it has answers. That’s not a bad thing. The ultimate goal is action that leads to positive results.

Analytics can’t answer all questions so directly asking visitors via usability studies and/or surveys is a good way to find answers to questions that analytics data falls short on. There is also another source of information which often seems overlooked.

Many sites undergo frequent if minor tweaks and updates. Improvement often works better and is less risky as an iterative process rather than as a big re-design. In addition, if a number of issues have been identified, by picking them off one by one it’s easier to assess the relative impact as solutions are implemented.

There have been many occasions when I have been to see clients to go through the findings of an analysis and I’ve had questions as well as answers. I’ve not always had the benefit of usability studies or site surveys to help unearth answers to my questions so I’ve simply had to ask my client. On occasions like this I’ve been met with blank faces or furrowed brows as memories are searched for activity that took place around the time (shown on the trended chart) where change is apparent. Quite often an explanation is clear but this raises another issue, that of change tracking. It’s amazing how many organisations appear not to log this (adequately).

There are literally dozens of events that can have an impact on site performance, ranging from a complete site re-design to a sunny day or public holiday. I’m not suggesting that weather conditions should be factored into every site analysis, although some might consider it important enough, but changes to marketing activity on and off line will have an impact. Additionally, back end technical issues will have an impact; changing the position of a link or even just the destination page will have an impact; changing some wording; tweaking a business rule, the list goes on. Remember also that a relatively innocuous change can’t have a disproportionately significant impact.

With all the minor and not so minor tweaks that continually take place it makes sense to log these in a central data base. This could be as simple as a collection of spreadsheets for each department to a central resource that can be access by all relevant stakeholders. Each change is logged by date and ideally time of day. The benefit of being able to refer back to such a resource and accurately identify what took place around the time of a change in performance can’t be underestimated. In addition the cumulative learning that can be built up if performance tracking were included in such a database would be invaluable. If you’re reading this and you’re not doing it, give it a shot.