Monday, 15 August 2011

Web Analytics: Bounce rate issues

When I was younger, I used to come home from school and play catch with a friend of mine.  We'd stand a fair distance apart (it got further as we got older, until eventually we went to big school!) and just practise throwing and catching a tennis ball.  We developed some games based on catch, including our own version of dodge-ball (where neither player moves), but they all started from the same basic principle of throwing and catching.  And some of them led to the classic argument, "You dropped it!  You lose!", "No, you didn't throw it properly, that's not fair." "You can't catch!" "You threw it too high!" And so on.

In the online environment (which didn't even exist when we used to play catch), there's a similar argument going on at the moment, in various offices in various countries.  Online marketers are spending large sums of money on driving traffic to a website - on display advertising, on aggregators, affiliates, paid search (PPC) and so on. And website developers are building landing pages (or maybe just standard content pages) which are designed to move traffic towards success events like checkout, form completion and so on.  The marketers are optimising their spend, looking at maximising click-through rate, optimising keyword spend, sites and so on, and are driving as much traffic as possible (and hopefully qualified and relevant traffic, too).  The developers are building pages (and running the rest of the website at the same time, just to keep them occupied) and are looking at what creative is working best across the site, and adapting content and so on, to improve the overall content of the site.
The problem, however, is that these two aims aren't always in alignment.

Let's start with an extreme case, before we look at something more realistic.  The online marketing team decide to offer a free premium widget with every purchase, and the premium widget is worth £100.  The costs of typical widgets are £10 each, so by purchasing one normal widget, customers have the opportunity to get £100 of free merchandise.  However, the content team aren't told about this free offer; or worse still, the marketing team point their advertising links to a typical content page for the £100 widgets (so visitors can see how good they are).
And what happens?  The marketing team declare the campaign to be a resounding success, with huge click-through rates and vast quantities of traffic.  And for sure, the website analyst reports a significant increase in traffic while the campaign is underway, but there's only a moderate increase in add-to-carts, a small increase in visitors starting the checkout process and almost no difference in completed checkouts.

So, what's the problem?  Marketing argue that the content team aren't converting the traffic well, and that the site needs to be looked at - especially all the payment pages.  They point to their success against their metrics, and they claim that they've done their job in bringing the traffic to the site.  And this is a valid argument, until somebody from the content team takes a look at the offer being promoted by marketing.  The analyst does some work, and finds that while the marketing team did direct 10,000 visitors to the site in a two-day period (for example), almost all of them left the site without clicking anything further - in fact, 9,000 of the 10,000 visitors didn't engage with the site any further.

This means that 90% of the traffic that came from online advertising was wasted.  It 'bounced', because it landed on the site and left without doing anything further.  This is called the bounce rate, and is defined as the number of visitors who saw just one page, divided by the total number of visitors who entered the site on that page, expressed as a percentage.

Why is the bounce rate so high?  Well, it depends who you ask - in the same way as my friend and I used to argue when one of us dropped the ball in our game of catch.  Was the ball badly thrown, or was it dropped by someone with butter-fingers?  Eventually, rules had to be developed so that we could determine what was a good throw and what was a poor catch.

Offline stores don't advertise "As seen on TV" on their shelves just because it looks good - it's a useful reminder for those who have seen it on TV that this is the product they're looking for.  Online, it's vital that the message in the marketing matches the message on the website.  Do the offer, message, creative and colouring look the same on the landing page as they do on the advert?  Are all the terms and conditions explained up front?  If it's buy-one-get-one-free, is that for all widgets or just for the platinum-coated ones?  In my extreme case above, the premium widgets were apparently free with any purchase, but this wasn't mentioned on the landing page (because the content team weren't told) and as you progress through the checkout, it turns out that premium widgets are free with any purchase over £70.  This means that the 10% of people who didn't bounce, and who went on to engage with the site had to be interested in expensive widgets, and the free offer is only revealed in full as you approach the end of the checkout process.

Bounce rate, then, is not merely a measure of how successful the landing page is at persuading traffic to engage deeper with the site. It's a measure of the synergy between the traffic source and the landing page.  The consequence of this is that bounce rate cannot be reported at a gross level for a page.  Well, it can, but in order to get any meaningful information from it, it's more useful to look at the bounce rate of individual traffic sources (pages, adverts, sites, keywords - paid vs natural) in conjunction with that particular page.

For example - and this is a real life example from my own Red Arrows website - I've seen a 100% bounce rate for people searching for "typhoon formation" on Google, and landing on my website page about the Red Arrows' Typhoon formation, named after the Typhoon aircraft. This isn't surprising - I've deduced that most people searching for "typhoon formation" are looking for information about how storms develop at sea, and not about an aerobatic shape flown by nine aircraft.  The result?  If I was spending PPC money on "typhoon formation", I'd stop.  As it is, the traffic was all natural, and I was stuck with a source of traffic that had a very high tendency to bounce, until other more appropriate sites started discussing typhoon formation - and the traffic appears to have decreased with Google's Panda algorithm update (but I digress).

So, whose fault is it?  I might be so bold as to say that everybody's to blame - both the marketers and the content developers.  It comes down to the analyst to play the referee and identify exactly what caused a visitor to bounce, and to help both sides work together to stop them bouncing.  If the marketers are producing advertising which is appealing and accurate (i.e. genuine, no-hidden-terms-and-conditions promotions) and the developers are building pages which show a clear connection with the advertising, reiterating the message and reinforcing it (and also providing a clear link deeper into the site towards the success events) then both sides will work together to a common aim, and the success of the campaign should be seen by both teams!

No comments:

Post a Comment