“Web analytics data alone cannot accurately judge satisfaction”
IN-DEPTH: Evaluating a website’s performance
Published: 11 Nov 2009
IN-DEPTH: Evaluating a website’s performance
It is said that if one is only acting on numbers, it is impossible to accurately judge whether a user has had a satisfactory experience with a particular website.
At the end of the day, a page view is just a page view. Numbers need context to become meaningful and websites have more than one goal, be it online sales, lead generation, customer relations, etc. By segmenting the site visitors into meaningful groups first, for example, the search engine they used, keyword typed, country of origin to name just a few, and then monitoring how those segments perform across the site and convert, it is then possible to gain a good overview on how the site is performing.
Web analytics data
Web analytics data is primarily concerned with the “what” – it is an extremely effective way of collecting clickstream data relating to what people are doing on a site and how long it is taking it, says Vicky Brock, co-founder, Highland Business Research. To effectively judge why, and thereby whether someone was happy, frustrated satisfied, dissatisfied requires using a wider toolset.
“Web analytics data alone cannot accurately judge satisfaction,” says Brock. Does a longer time spent on site mean a visitor is engaged or lost? Does the fact they keep coming back means they are loyal or that the site is broken and they are being kicked out of their booking.
Brock added that context is everything and part of that context comes from the visitor themselves. To understand satisfaction, rather than click sequences, conversion rates, say ROI or revenue by traffic source, then specialist tools such as Foresee Results are required. Foresee Results uses specific elements of satisfaction, based on survey data collected from the site, to predict which elements are the site are impacting visitor experience the most.
“Web analytics is aspect of an integrated analytical tool set, use it for “what?” and “how many?” – my advice is don’t try to use it for “why?” There are other methodologies that do “why” better,” says Brock.
Website performance reporting and analysis
Every website is different and as such will have different goals. For example, an e-commerce manager would monitor how many visitors book hotels, how many engage with hotels (i.e. either sign up to receive offers or submit a “contact us” form), read press releases, sign up online to the loyalty programme, activate loyalty programme online, drop out during the booking process, return more than a couple of times to the site but do not book, book more than twice etc. These would be just a few of performance indicators for e-commerce, but they might not be relevant to a different website, for example, a non e-commerce one or a blog. One should use the tools to understand how customers are interacting with the site now and why they are not converting, whichever the goals might be, even if just a simple fact sheet download, if that is important to the business in question.
The same principle applies to choosing the right web analytics tool. If goals and key performance indicators have not been established in advance, one cannot asses if a basic or more advanced analytics tool might be required. However, tools alone are not sufficient. Even the most sophisticated and expensive web analytics system is pretty useless, without the right people to use it and make those numbers “talk”.
Brock agrees and says it is a cliché, but one has to measure what matters to his/ her business. What does online success look like to you? Is it more revenue, money saved compared to offline servicing, higher customer satisfaction? Its probably a combination of all three. You have to look at the points of interaction on your site that represent those success outcomes and focus your analysis and reporting on those.
“Frankly page views, unique visitors, time on site etc are not things that management should give a damn about in isolation. An analyst needs to understand them intimately but the business question an analyst should be answering with every report is “how is the web contributing to overall business goals today,” said Brock.
“And on the whole, it’s not about the tools. Tools are generally great. People may not pick the best solution for them, but on the whole the tools/offerings/solutions are often way ahead of most organisations abilities to use them effectively. When organisations go through analytics tool after analytics tool, its often a symptom that they are expecting the answer to come out of a box, as opposed to investing the time and energy required to use the data being generated to answer their own specific business questions. You can get more actionable insight from a basic tool used by a great analyst, than you’ll ever get from a high end tool if there are not the people to actually generate analysis (as opposed to reporting),” explained Brock.
Role of an analyst
Tools, according to specialists, alone are not going to do much. Rather, one should also budget for a proficient analyst, who is then going to be able to interpret the data and provide actionable conclusions to improve the site and conversion.
Referring to an established model, Brock said for every £1 spend on the tool itself, spend £9 on the people who are going to use it. That ratio alone should help people figure out the real level of analytics investment they are able to make.
In terms of web analytics tools, options include Google Analytics and Yahoo at the free end, through to Webtrends, Unica, Coremetrics and Omniture at the higher investment end. “Correctly implemented and integrated with your booking engine, you can measure the visitor from the search term they used to find the site, right through to purchase and even repeat purchase,” said Brock.
“You cannot optimise web without the tools that measure onsite activity, so getting the tracking right is the first step in any website optimisation process.”
Remember that even the free tools require implementation, people to use them and developers time. Paid for tools typically are priced by traffic, and may have additional usage costs beyond the initial purchase fee.
In all cases, one needs to understand the total cost of ownership.