Actionable Analytics

Posts Tagged ‘ROI’

Drowning in Data

The concept of ‘drowning in data’ cannot be understated when it comes to web analytics. Apart from the sheer quantity of information available, the situation is worsened because the tools we use are so terribly fast and effective; it has never been easier to slice, dice and peel (?) your way through such huge mountains of click-stream data. But just because it’s there and easy to access certainly doesn’t mean it’s easy to make sense of. I believe most companies that fail in this arena do so because they simply don’t know what to look at, but rather flail around in the data following endless and infinite pathways that, whilst ‘interesting’, ultimately lead nowhere fast.

This post describes how to get your head above the water and start swimming in a straight line. The answer lies in what I call a ‘Measurement and Optimisation Framework’, which might sound complicated but is, in fact, simply a strategy for: what you should be measuring; how to do it; and what you should do with the information once you get it.

Developing a Measurement & Optimisation Framework

The process of developing a measurement and optimisation framework is simply about answering 3 key questions:

1. Why does my website exist?
2. How can I measure the success of that existence?
3. What can I do to make it more successful if I find it isn’t achieving what I want it to?

For very simple websites (such as a personal blog), you could probably get away with just spending an hour or so thinking about this. For more complex business sites it could take some time! Following is a brief summary of the top-level part of the process through which I would typically take a client in order to get this up and running:

1: Define your site’s KPIs

How can you fix something or make it better if you don’t know what it was meant to do in the first place? Not setting proper objectives and goals is the most serious and fundamental mistake anyone can make, and not just in web analytics!

Most companies fail to do this because they assume that they intrinsically know what their site is for and what needs to be done to improve it. Take the example of a site selling CDs – it’s for selling CDs, right? What could be more complicated than that?

But, think about it for a moment, who is it trying to sell CDs to? Is it trying to achieve the lowest price possible or is it selling at a premium because it caters to a niche? And where does the company want to be in 5 years time, and what does that mean in terms of the brand that needs to be built? Is it important that people tell their friends about it? Oh, and how does the profitability work? Do we need to reduce the cost per sale by increasing the number of repeat buyers and therefore reducing media spend? And what about our other sales channels? Sales on the web cost much less than those that go through the call centre, so do we need to persuade some of those customers to get on-line? etc etc etc…

The point is, what your website means strategically is not necessarily all that easy to articulate. You need to get a really firm grasp on what your companies corporate goals are and work downwards. For big companies this generally means using something like a Balanced Scorecard approach. The system you use isn’t necessarily important, the point is that you align the goals of your site with the strategic goals of the company or, better still, the strategic goals of your customers!

2: Set targets

Once you have defined how to measure success (your KPIs), you then need to determine what that success IS. Again, this goes back to your corporate objectives: if your site is there to generate advertising revenue, how much revenue do your shareholders need next year? And what does that mean in terms of the number of visitors you need and the number of pages they need to look at? This is how you set targets.

Even if you can’t get anyone in your company to give you these targets, you should make them up yourself! It is incredibly difficult to optimise something to work better if you don’t know what ‘better’ means. If you are not able to prioritise which areas of the site need the most attention at any one time, you will drown – and you cannot do this without a sense of the goal for each KPI. Just do it!

At this point in time you might be able to produce something like this:

Typical Web Analytics Measurement Framework

[Please note: I doctored this a lot to protect the identity of a client, so it won't necessarily make complete intuitive sense and is provided more as a visual example]

3: Guide your analysis with a KPI dashboard

Now you know what your KPIs are and how to measure them you can produce a dashboard report showing where they are against where they need to be. This is incredibly important because it is the guiding light of your analytics and tells you exactly what to look for. If, taking the example KPIs in the chart above, I produce my weekly or monthly dashboard only to find that my unique visitors are dangerously below target but that all other KPIs are OK, then all my analysis for that week/month will be guided by a very specific question: what drives unique visitors and how can I improve the volume?

By investigating this you might find, for example, that you have saturated your search market and therefore need to optimise the site for different, non-branded keywords – or that the TV campaign you tested sent lots of high quality traffic and should be repeated. The point is that, without the KPIs, targets and the dashboard, you have nothing to focus you and, more importantly, have no solid way of telling your marketing director why they need to spend more on TV!

4: Optimise, optimise, optimise!

Remember finally that web analysis is not about understanding, its about doing. If you think your job is to report figures to someone else so that they can make sense of them, then you are not an analyst. The output of everything you do is about making changes to your site, media strategy, internal processes or whatever. Analysis and optimisation are essentially the same thing!

Empower yourself!

So what’s the benefit of all this? If it isn’t already obvious think about these two possible scenarios in which you are presenting your ‘analysis’ to your wider team:

• You hold a meeting in which you present 30 charts of data from your analytics tool, moving through geography, time on site, hour of day, browsers, screen resolution and lots of other fascinating charts. At the end everyone agrees that it was really interesting and goes back to their jobs.
• You hold a meeting in which you state that you can make the company an additional £1.5m per year in sales revenue and then proceed to present a road-map for implementing changes to make it happen, with a full ROI justification of likely costs. You get promoted and paid more!

Which one would you prefer?

With recent and continued advances to Google’s excellent and free analytics tool, one of the key questions that I seem to get asked these days is whether there is any real value in paying companies like Omniture and Webtrends for the commercial (and expensive!) services they provide. 

It’s probably already obvious that I’m a fan of Google Analytics (be prepared for gratuitous bias); for lots of clients I really don’t see how spending the money on something like Omniture would benefit them. However, this isn’t always the case, and I think a more systematic way of making this decision is often called for. 

This post is therefore an attempt to help make decisions about whether or not you should put your hand in your pocket, and I have chosen Omniture Site Catalyst as an example. 

Monetizing the Incremental Value of Site Catalyst

Now, it is undeniable that a tool like Site Catalyst does some more stuff than Google Analytics, and certainly that it has more dedicated and human support. However, it is very easy for clients to get blinded by the way sales people position these extra features; they don’t stop to think what they might actually use them for. Conversely, GA extremists will flatly deny that there is any use in these additional features (or sometimes that they even exist), likewise failing to provide adequate reasoning. 

It seems to me that there is a more simple way of stating the true question: 

Site Catalyst does various things that Google Analytics doesn’t. What benefit do these things provide on their own (i.e. in isolation from any of the things that both GA and SC can do)? And – can the entire cost of Site Catalyst therefore be justified based on these incremental benefits?

So what does Site Catalyst do that Google Analytics doesn’t?

Following is a list of the key things that I believe SC does that GA doesn’t. It isn’t meant to be completely definitive, but [in my honest opinion] everything else is pretty much cosmetic:

Weighing up the cost benefits

Real-time data – this basically means your stats update more-or-less straight away rather than after about 24 hours or at mid-night. Personally I find it hard to think of companies that could truly benefit from this, but if you think you might then you need to work out exactly what financial benefit it gives you over and above waiting half a day. Also check out Avinash Kaushik’s blog on real-time data.

Importing external data – at first glance, this is a fairly major thing that GA doesn’t do. In Omniture you could import a lookup table of postal codes and then use this to carve up sessions into sales territories. This can be pretty valuable, but what you really need to ask yourself is: ‘how much benefit does this give us over and above exporting the data to excel and making the table ourselves?’ How much extra work is it really to just do this outside the tool? This also applies to a lot of other stuff, such as the functionality that lets you add targets to KPIs – and also to most of the Genesis integrations.

Custom variables – you actually get 2 of these in GA, but then you get loads in Omniture. Yes, for some companies this is valuable, but are you one of them? Again, I’m not denying that these things are important; I’m saying that you need to make an actual financial calculation about the benefit you get from using them over not using them. ‘Nice to have’, ‘convenient’ and ‘handy’ are not good enough reasons! Another function with similar ramifications is the ability to link metrics with dimensions that are not available in the out-of-the-box package.

Creating paths and funnels on the fly – very nice, and I wish GA did this, but I would have a hard time selling it to a client and I also couldn’t say that it is critical. I’ve certainly never seen it as a barrier in GA. Monetize it if you need it!

And, seriously, that’s pretty much it! Like I said, everything else is cosmetic or falls into a similar category. The main point is that you don’t get swayed by the sales spiel, and you calculate the return on your investment not by asking what analytics per-se can do for you, but asking instead exactly what can commercial analytics do for you that the free stuff can’t?

But wait…

Having said all that, a big word of caution  – GA can do a lot of stuff that Site Catalyst can, but a lot of the time it isn’t necessarily easy or straight forward to do, so much so that you might not even know or believe that it is possible in GA. What I’m getting at here is, you may need specialist expertise (a decent analyst) to be able to match GA with Site Catalyst on some levels of functionality. Again though, monetize this properly – you would have to pay someone to use Site Catalyst, so how much more would you have to pay someone to get the most from GA and how does this weigh up against the cost of SC?

Finally, it is worth also noting that I haven’t even touched on what GA can do that SC (on its own, i.e. without Discover etc) can’t, and believe me there is plenty of stuff!