The web analytics industry has come a long way in recent years. What used to be a technical subject for tracking server performance, error pages and hits, has evolved into a dream for both webmasters and marketers. Different conversion metrics and funnel paths are now visible and can be cross-referenced with other data - such as the visitor’s geographic location or which search engine and keyword they used to arrive at a website.

With the correct implementation, modern analytics packages can even get a handle on the success of offline marketing campaigns. Essentially, if a visitor arrives at a website, then you can measure not only how they got there, but also their interaction with the site.

However, despite the growth in features and capabilities of the tools used to measure online marketing activity, there has been slower growth in the understanding and consideration for the power of such tools. A problem I have seen in even the largest of organisations is the fateful lack of time and resource invested in ensuring the right tool is implemented, and that it is implemented correctly.

This is one of the key philosophies behind why we have chosen to make Google Analytics free. Other than making it available to all website owners, regardless of size, we believe customers should invest in gaining insights from the tool, rather than the tool itself. This means hiring skilled data analysts at the outset (not just after a poor implementation has already been made) who are aware of the online business objectives and reporting requirements for all the key stakeholders and whose job it is to ensure that the correct tool is implemented, and is capable of delivering such insights.

Most online businesses are aware of their basic traffic numbers and resulting sales leads. However something I still find quite surprising is the general acceptance in poor online conversion rates, which would probably cause alarm bells to ring in the offline world. This can of course be attributed to the significant difference in maturity between the two sales channels, as online is generally considered to be still in relatively early development. An interesting research finding is that two-thirds of shopping carts are abandoned at the online checkout, something which would be regarded as a catastrophe if it were to occur on the high street.

Using an analytics tool does not have to be daunting or mind-boggling. Whatever you are using, you should be able to answer some fundamental questions regarding the performance of your website. However, before you even begin to look at any data, you need to determine what the objectives of your website are and how they tie into your online business strategy. Does your site exist to sell? Provide a service or information? Generate offline leads? With the objectives in mind, you can use the data to understand whether the site is successful in achieving these and whether the user is coming to the website for the same purpose as what you had intended.

Start by looking at how visitors arrive at your site. Visitors can arrive via a number of channels, be it organic (by clicking on a link under the natural search listings in a search page), paid (clicking on a sponsored link), direct (by manually typing the URL or clicking on a bookmark), or via a referral link on another site or email campaign. There is a lot of information to be gauged whichever method is used.

Campaign tracking functionality and Google’s seamless integration with AdWords allows you to measure the performance of your marketing campaigns. Discover which keywords, affiliates, newsletters, etc are working best. You may find that a particular keyword is great at attracting visitors, but has a high landing page bounce rate. This is essentially a single access page view where the visitor exited the site before clicking further or interacting with the page in any way. What’s gone wrong here? It may be that your site is not meeting visitor expectations when clicking on that keyword. Each bounce is a lost opportunity, and if it is via a ppc campaign, it is also impacting your overall ROI.

Next, investigate what people are doing once they get to your site. Set up goals and measure conversion rates against these. However, bear in mind, goal completions (such as successful purchases) are great ways of measuring success, however not where you have missed opportunities. For this, reports such as funnel path analysis are instrumental in analysing whether people are following your expected conversion path, where people are dropping out of the process, and why. Identifying and rectifying these so-called bottlenecks to conversion can be instrumental in improving your bottom-line.

If your site has an internal search function, you practically have free marketing research on your visitors. You can arrive at some sophisticated insights by looking at the keywords people used to arrive at your site, married with the search refinements made whilst on the site. Does your site actually stock or produce what most people are looking for? If so, then the data should be driving your search marketing strategy by bidding on these keywords in your ppc campaigns.

Equally as important and often neglected, is that if you do not produce these products, then you should most certainly refrain from using them in any campaigns as people will exit your site just as quickly as they found it. However, on the other hand, looking at consumer’s search patterns whilst on your site may give you some food for thought in terms of the products you could or should be selling. Analytics is essentially free market research of products on demand.

Search should also be helping to drive editorial decision making. If you are using half of your homepage to advertise shoes, whereas most people are looking for shirts, then maybe you should consider a change. If you have a content management system that supports personalisation, then you could even vary the content based on a visitor’s location. All this is based on information you obtain from internal search term reporting. And if you do not have an internal search function, you should install one.

Another cool feature is the geo-location report, which most reporting tools offer, showing  which countries and cities visitors are coming from. Segment these with purchase data to help regionalise offline marketing campaigns.

The list of data aspects you can measure and analyse is endless. What’s important is acting on such findings, is something that research has shown to be the most difficult aspect of analytics for a company. This is the testing phase and it is crucial since most design decisions are often made based on what the design team or a key executive believe to be right. However, it should be the consumer driving these decisions and this can only be achieved using testing.

What we often see is online retailers and designers making the critical error of building a site based on what they think works best, without using analytics data to determine what the user thinks is best. Something that simple A/B and multivariate tests can elicit when set up correctly. In a recent article, Avinash Kaushik, Google’s Analytics evangelist,  gave the great example of a site doing things right, Zappos.com which sells a range of shoes, accessories and clothing. Launched in 1999 with virtually no business, it now projects $1 billion in sales this year, becoming the largest footwear retailer on the Web. Aesthetically, its website is unremarkable. But Zappos does two things well: first is top-shelf customer service; second is focusing on data, using it to understand its customers' needs and building its website around that understanding.

This is where tools such as Website Optimiser come into play as this allows you to perform A/B and multivariate tests on pages of your site. You can create different versions of your web pages, and Google then splits your traffic automatically, so that your visitors tell you which version they like best. Website Optimiser enables designers to bring more than their experience to bear in creating a site. It also provides quantitative proof that your site suits your client's needs as well as their tastes.

What I have often seen is that smaller organisations with less complicated development cycles tend to react better to such insight. However, often with larger sites, consumer trends and market developments will probably advance before a change makes it into the site development cycle. What I believe needs to happen is for larger online organisations to make measurement, analysis and test part of the standard review processes, not just a peripheral activity. By doing so, they will become more of a data-driven organisation, acting on insight rather than speculation.

By Dyana Najdi Google Analytics Manager, EMEA