Colette Wade, marketing director for EMEA and Australasia at Webtrends, explains why you should always let the data decide when it comes to site development
Too much web development is based on aesthetics. Often, there’s also a misplaced desire to showcase the latest applications, social media technology and Web 2.0 functionality. However, these shouldn’t be the yardsticks by which a site is judged.
Introducing the shiniest new thing on the web for no practical reason will have no impact beyond increasing bounce rates. If you want to make your site more sticky and ensure visitors follow calls to action, all changes should be based on data.
Before optimising any site, you need insight into which pages require adjustment so you can pinpoint exactly where the highest bounce rate is occurring. There’s no point in deciding the homepage needs work just because it seems like a good place to start. It may be a well-worn adage, but it’s true here: if it isn’t broken, don’t fix it.
Test, then test again
Any alterations made should be underpinned by extensive trials so that every button and tab helps the user complete a task – whether that’s buying beans or signing up for a newsletter. Multivariate testing breaks a page down to its core components in order to try out different combinations of how items are placed, what they look like on the page and how they’re received by real-life users.
The behaviour of those users can then be tracked after exposure to variations in content, determining the strengths and weaknesses of each option. These results – combined with data on individual visitors (what time of day they logged on, the browser they were using and so on) flag up the most successful combinations.
Segment-based targeting is another function that site owners should consider if they want to identify and act on specific behaviour. The IP addresses of visitors can be identified, ascertaining location so that you deliver the most relevant content. That’s something that a person’s opinion can’t compete with, so when you’re performing a test, let it run completely in order to obtain the whole picture. Resist the temptation to cut it short and fill in the gaps with guesstimates.
It’s also worth thinking carefully about whether adding something – a live Twitter feed, for example – is going to ease someone’s path through your conversion funnel, or whether it’ll end up drawing people away when they see something else. Web users are flighty things and need no encouragement to find another corner of the internet to explore. If your site’s call to action is to follow a Twitter feed, by all means include one. But without any other reason backed by testing, you’re more likely to distract the user away from your website’s main goal.
Every aspect of developing a great site needs to involve designing, testing and planning, and should be based upon performance analytics, not opinion. To avoid your pages becoming stagnant, the testing process should be continuously implemented. Therefore, content can regularly be refreshed and updated, helping you to ensure that there isn’t too much clutter by way of secondary calls to action, be they affiliate adverts or other areas within your site. Yes, advertising can drive revenue, but it can also push people away from what you want them to do.
Trust the numbers
Many site owners are now using cross-channel optimisation for other areas of their business, ensuring continuity in all contact with customers. For example, once someone has purchased an item, follow-up communication with them should offer products and services that complement the initial purchase. If they buy a printer, don’t send an email saying, “We saw your recent purchase and thought you might be interested in some new printers that we now have in stock.” It won’t get you another sale because they’ve just bought one.
Whatever the future holds (and, let’s face it, in the ever-changing world of the web, no one can say what that is with absolute certainty), individual websites will always be unique and cannot be directly compared to one another. With that in mind, ignore opinion and let the data decide.