Everyone understands the importance of web analytics.  Web analytics suites help you to gain the important insights you need in order to optimize your web marketing efforts.  For example, if you’re trying to maximize ROI on your PPC campaigns, analytics data will help you see which keywords are yielding not only the most clicks, but also which ones are resulting in the most conversions.  Analytics data can also help you gauge the importance of your content, e.g. what percentage of people visiting your landing page go on to make a purchase, as opposed to “bouncing” off after a few seconds.  Ok, so no need to underscore the importance of web analytics to a seasoned professional, I know.

Now you probably already know that I didn’t write this article just to preach to the choir…hold on to your stepladders, maties, it’s coming…

Although most pros understand the importance of their clickstream data, many are not taking advantage of it as they should.  The reason is that the analytics tools available to them restrict their ability to ask the right questions.  After all, any decision support tool is only as good as the questions it allows you to ask.  Web analytics programs are insufficient in this regard for the following reasons:

1.  Even though all the data you need to get an answer to your question resides in the log file, the analytics program doesn’t analyze or examine the relationships which are relevant to your question.  For example, let’s say you wanted to know how many people who clicked on a particular ad came from the UK.  Most analytics suites will allow you to ask “What percentage of my visitors for this period came from the UK?”  Unfortunately, this is about as far as most free analytics suites will go  - there are aren’t many free web analytics suites that allow you to take it a few steps further and ask “What percentage of visitors who searched for keyword ‘Madame Tussaud’s’ and clicked on the ‘Soho Hotels’ PPC ad came from the US?”  Try using your Google Analytics suite to answer a question like that – you’ll find that it’s just not possible.  And it’s not because the data you need to answer the question isn’t there in your log file – it’s all there, it’s just that your analytics suite severely limits the kinds of questions you can ask of it.

You may be asking “If the data is all there in my log file why isn’t my analytics suite providing the tools I need to get the answer?”  Well, there are a number of reasons.  First, take a look at what you’re paying for it (cheapskate!) – if you’re like me, you’re probably using a free tool like Google Analytics to track your personal blog or the website of your Aunt Zoe’s manicure shop.  Yes, you get what you pay for – and with good reason: the kind of data parsing, storage, and querying you need to be able to do to answer any kind of question that could be answered with log file data requires some very sophisticated technology – the kind of technology that is traditionally the realm of Business Intelligence and Data Warehousing (if you’re wondering what these terms see).  That’s pretty sophisticated stuff, and it’s not cheap or easy to implement – so don’t expect to get it from your free analytics tool any time soon. 

Yet another reason why free analytics tools won’t be providing this is that – let’s face it – most people don’t use what limited tools are available to them properly or systematically as it is.  Indeed, most people don’t need to – or care to – get more out of their log files; for example, I’m using Google Analytics as my analytics suite for this blog, and I admit that the most I’ll ever do with it is to check my visits, pageviews, and bounce rates every once in while.  So if the demand isn’t there, don’t expect Google or anybody else to provide you with a free tool that barely anybody is going to use.

Ok, so some of you brainiacs out there are starting to object already…yes, you’ll have me know that you’re the web marketing guru for the Central Bank of Rangoon’s high traffic portal, and that you’re not using any free tools, thank you…you’ll tell me that you sold your soul to Mephistopheles in order to be able to pay for a version of Omniture or Deep Metrix that comes with it’s own little data warehouse at the backend, and that you do use it systematically to ask some very complicated questions…that’s great, glad that you’re getting the most out of your log files…except sometimes – no, make that often – your log files don’t have all the data you need to answer your questions…which brings us to the second point of this post:

2.  Sometimes your clickstream data isn’t enough.  Yes, you heard that right – want me to repeat it?  Ok, I will - sometimes your clickstream data isn’t enough.  Don’t believe me?  Ok, well let’s take a simple concept like ROI…ROI, ROI, if you’re any kind of marketing freak, you’re obsessed with ROI…”what kind of ROI am I getting on this ad, or this campaign?”  Well, Einstein, it happens that you can’t calculate your ROI from your PPC ads unless you only sell one thing and your profit margin for each and every purchase is always the same…Why?  Because to calculate ROI, you need to know how much profit you made off of the purchases that came by way of your PPC campaign or ad…[ROI = (total profit from campaign)/(total cost of campaign)], and if you sell many different types of products with different prices and profit margins, you’re just not going to get that from your log files (I know, I know, you can calculate your average profit margin and use that as a proxy, but you’d never be that sloppy, would you?)…after all, different ads appeal to different audiences with different purchasing behaviors…if you’re looking to maximize ROI on your PPC ads, you don’t necessarily want to keep the ads that are generating the most purchases…you want to keep the ads that are consistently generating the highest ratio of profits to ad costs, and these aren’t always or necessarily the ones that are resulting in the most purchases…

ROI is just one example of how log files are not enough to provide you with all the data you may want or need to optimize your web marketing efforts…there are countless other examples, and we’ll cover some of these in articles to follow…the point is that, sometimes, your log file only provides some of the pieces to the puzzle…in order to ask the questions you need to ask to make the right decision, you need to bring in other data sources -  i.e. other pieces of the puzzle, e.g. invoice data from your sales database - and you need to combine them with the data you have in your log files…yup, take the problem we just outlined for ROI and multiply that by hundreds of other missing pieces of data and a gazillion transactions a day, and you’re dealing with a data warehousing and business intelligence problem…unless, of course, the Central Bank of Rangoon only has a handful of customers or transactions a day (far be it from me to even suggest that)……