Context is everything in web analytics, not numbers.

Web analytics is not all about numbers, in fact, numbers do not mean anything without context. Yes, numbers don’t mean it all. Let’s go back to basics and let’s take a look at the very well known business statement: your revenue does not mean anything, your profit does. Revenue does not mean anything because your revenue can be as high as you want it to be but it will never be an indication of how much money you are truly making. Having said that, your profit in itself and taken out of context is not as useless as your revenue but can be a useless number when taken out of context.
If your profit is £100K, is it an indication of how healthy your company is? Not at all. How does £100k compare to last month? To last year? Let’s say that it is actually up 2% YoY. Ok, things are starting to look positive. But hang on, the country’s GDP in which your company operates grew by 8% YoY and one of your competitor’s profit grew by 6% YoY. Does your £100K profit (up 2%YoY) look as healthy now? Not anymore I’m afraid…

I hope that this illustrated how important it is to look at numbers with a lot of context. As a web analyst, to get context, you can’t rely on just your web analytics solution. The good news is that there are tools that can give you just that. There are two tools I like using a lot: Google Insights for Search and Hitwise. Google Insights for Search is good as it gives you for FREE volumes indication of what the world has been searching for. It won’t give you actual volume numbers, Google actually normalizes the data and presents it on a scale from 0-100. Based on my experience, I have had discrepancy issues with the tool where I extracted data for a particular week on one day, came back the next day, extracted the exact same data for the exact same week and the results were different to an extent that meant I could not use the data anymore. So, I would bear that in mind and not take the numbers literally but look at trends instead. I mainly use the tool for Year on Year and Week on Week comparison to try and spot demand shifts.

The other tool I really really love is Hitwise. No, the two “really” in a row are not a mistake but a true statement of how I feel about this tool. Amongst so many other things it gives you an indication of your site traffic for a given market, or a “category” and for a set of competitors. Let me describe this a bit more: unlike a tool like Comscore for example,( great tool by the way!) Hitwise will not actually give you actual traffic numbers (at least it did not do this in 2009) but it will give you your site traffic market share as you can see on the graph I created below. For the purpose of this exercise, let’s say that you work in the automotive industry and that you are the web analyst for brand B. You are interested in knowing how your brand has been performing against brand A and C over the past few weeks.

Gain context with Hitwise

This graph is giving you a really good view of what has been happening in your market over the past few weeks. Something happened with brand A as its market share has increased over the weeks and it is now outperforming your brand! What have they been up to? Have they launched a massive TV campaign,? Have they gone viral on the web? Have they just launched a car which makes coffee and croissants in the morning? You guessed it, it could be anything but at least you now know that they are up to something and this could be one of the reasons why you have been getting slightly less traffic than usual over the past few weeks. Actually, one of the smart things about Hitwise is that when there is a sudden increase or decrease in market share, the tool will highlight for you in the graph major events that could have had something to do with it.

Using tools like this in conjunction with your web analytics solution can help build a better picture of the “WHY” as opposed to just the boring “WHAT”. Web analytics is complex in a sense that it requires merging different data sources together to try and make sense of what is occurring on your site on a given date. The tools described above are good at providing you with supply/competition and demand information which is a big step forward but there are so many other factors to be taken into account which can also have a major impact on you site performance. I will be talking about these in my next article so keep an eye on the blog!
Thanks for reading, I hope it was interesting and useful.

Posted in Secrets of Web Analytics | 1 Comment

How to provoke change with data

I wanted to follow up on my previous post Web Analytics and Change: Do these two concepts really go together? just so we don’t stay on a dark note and think that web analytics and change are two incompatible concepts. To me, for web analytics to be able to drive change, two things need to be happen. First of all, time needs to be allocated for this and second of all, processes have to be in place for this.
Let’s go back to the obstacles I raised in my last post. The first obstacle was the amount of people coming to you with very different questions about the site, the customers…To tackle that, I make sure I give new starters basic training on the analytics tool that is in use in the business. I also ask them about their job so I get an understanding of the type of data they can benefit from. That way, I can create custom reports for them straightaway, (and people love custom reports, it makes them feel a bit special!) Obviously, the same principles can be applied to existing members of the team. By doing this, you will save yourself and your team members some time. Time not spent on answering random questions on a regular basis means more time spent on higher value analyses that can lead to change.

The second obstacle I was describing was the amount of time required by the data collection process. In my opinion, there are two different approaches for this. The first approach is to review on a regular basis the amount of data that is really required; after all, businesses change constantly so reports should reflect that. The inconvenient of doing this is when you come to do an annual review and when you have altered your reports multiple times in the meantime, some of the data will be missing and it can be annoying. The second approach which is more practical than the first one consists in automating your reports. When I say automate, I mean automate, automate and automate again, so automate big time basically! Unfortunately and as described in my first post, not all web analytics tools have an excel plug in easy to use built into them. Now, depending on how many hours/days a week you spend on reporting, there could be a solid argument to justify switching to a more adequate analytics platform. Again, time not spent on reporting means time spent on analysing the data properly,and making recommendations that can drive change.

The last point I made in the previous post was the fact that it was very common for drastic changes to take place on a site without web analytics being taken into consideration at any stage. Now, this is a hard one to tackle because for this to change, it requires web analytics to be put at the centre of the decision process. This is hard to achieve because this means proper processes to be in place. Having proper processes in place requires the whole of your business to take web analytics seriously. It’s very easy to organise one meeting to talk about the data that will be required for project X. However, it’s an other thing to have these data driven meetings happening on a regular basis with key stakeholders. If not done already, I would really recommend you to read the web analytics business process white paper by Eric Peterson which gives some very solid grounds on this: The Web Analytics Business Process.
I would be very interested in hearing from anyone who has already implemented bits of what Eric is recommending in his white paper or any other web analytics processes of some sort.
Thank you for reading

Posted in Secrets of Web Analytics | 2 Comments

Web Analytics and Change: Do these two concepts really go together?

Yes, I understand that this is quite a controversial and rather unexpected choice of topic for a web analytics blog, however, this is a problem I have been encountering from the day I started working in web analytics. Sure, it is the end purpose of web analytics to generate actionable insights, however, how realistic is this? There are in fact quite a few obstacles that can lead to not drive actions from the data gathered. The width of data available leads to extensive time spent on collecting it and then ample time reporting on it. And sometimes, when the valuable data is all there, you still can’t act on it, even worse, big changes do happen but without any analyses beforehand…

Because there is so much data available in a company’s analytics tool, one might be tempted to take a look at all sorts of data. When I say one, I mean a web analyst but when you are part of a larger team, you will find that many people will also want to know things just for the sake of it (sometimes); as opposed to try and answer real business needs. After all, it is only natural that people want to know more about their customers even if the answers you supply won’t give them anything to act on. The issue is these sorts of “distractions“ do take time, a lot of time. Yes, if you start counting the number of times people asked you about a KPI or to carry out an analysis for no real business reasons , then you might end up feeling quite sad. Don’t’ get me wrong, I too actually quite like making all sorts of useful and not so useful discoveries on a site, the only trouble is there are only 24 hours in a day, well 12 hours for the crazy people like me who sleep at night.

Collecting data is an other task that can be very time consuming. Of course, most web analytics tools aim at making this process much more straightforward than it used to be. You will always be able to automate basic reports from your analytics tool (all of them) that will drop yesterday’s sales and visitors into your inbox. However, if you want to dive a little further into the data, chances are you will have to use an excel plug in built into your analytics tool. This is why you have brilliant tools such as Report Builder: http://www.omniture.com/press/754 to help you out. Unfortunately such tools aren’t part of all analytics packages, which increases the time of collecting data by a great deal. So this long process of collecting your web analytics data does not facilitate the process of acting on the data.

The other thing you might find is that some of the changes needed to improve a website      performance require structural changes that are too costly and time consuming to implement. Yes, sometimes getting the data and figuring out what would improve engagement or conversions on your site is the easy bit but what needs to be changed requires too much turn around. For example, when crucial information on your website relies on outdated third party data centers that do not send information in real time.

Just when you thought that things couldn’t get any worse: everyday, a whole lot of site changes take place, the only thing is most of these were never driven by data… This is a very common thing – people within a company decide there’s a need for a new site because the current one is outdated and does not look fresh anymore. The problem is making your site look nicer, in most cases, is just not enough and can end up being a waste of time for everyone involved as well as a waste of money. It is not rare that web analysts don’t get asked to carry out a full study of what the real site problems are because there is too much focus on making the site look newer. However, what you can be assured of is that 5 minutes after the new site was launched, tones of people will ask you for RESULTS i.e improved KPIs. And then comes the challenge of reporting on the new site and on the impact that the site redesign has had on your site performance.One of the questions I keep asking myself is why should traffic and/or conversions increase just because your site looks nicer? When I say “nicer”, this of course does not include usability improvements.

So, not wishing to be pessimistic about the ability of web analytics to drive change  but these above facts happen constantly in real web analytics life. The question is how can we as web analysts avoid such situations? Well, in my next post which I will publish within the next two weeks I will try to answer this so keep an eye on the blog if you’re interested! But in the meantime, please do comment on this and tell me how you feel about this.

Posted in Web Analytics Challenges | 2 Comments