.rising leaders: 3 tips to optimise your website (and avoid the HIPPO)

The latest .rising leaders post comes from Harry Thuillier, Head of Marketing for Squared Online, the revolutionary digital marketing course developed by Google and powered by Home Learning College.

Here Harry reflects on how to best use data and testing to support key decision making, understand objectives and challenge received opinions within a business. Don’t  forget to get involved in the discussion by commenting below or tweeting @dotrising.

 

What happened in your last meeting at work? Who made the decisions? How were they made? Were they made, God forbid, with just the HIPPO? [*]  If you’re concerned there may be an unconstrained HIPPO rampaging around your business, this might be for you.

With Squared Online we’re learning all the time how to use data and testing to improve user experience and improve the bottom line. Here are three learnings distilled from my last two years of looking after marketing here.

1. Start with the data (but don’t start with the data)

Some people say start with the data. Actually, if you don’t know what you’re looking for I find it easy just to stare blankly at the graphs and charts in Google Analytics. So before you dive in, ask what you want to achieve in general terms, whether that’s generating more sales on the site, increasing number of pages viewed if you’re a publisher, or encouraging more enquiries if you’re doing lead gen.

Only once you have a crystal clear idea of your business aims do you look at the performance indicators you have that enable you to measure against that definition of success, and then see if you can deduce insights from the data by splitting it up and spotting patterns. For a site looking to increase engagement and consumption of content you may be more interested in videos viewed, time on site, or returning visitors – whereas for an e-commerce site revenue and average cost per sale might be key sources.

2. Piece together the story (but remember you’re not the author)

Once you’ve had a dig around, whether that’s looking at bounce rate on a page, traffic sources, or the devices people are using, you come up with a hypothesis. It’s a way of testing your recommendations.

Your hypothesis is a combination of your observations from the data and the insights you are drawing from it, and your idea on what change to the website would bring you closer to your objectives – whether that’s improving the user experience in some way or maximising conversions.

This is where most of us take a wrong turn. Because you’ve probably had a great idea, you think there’s no way it could not be true. So when you construct the test you might accidentally skew the data in favour of proving yourself right. It’s human nature – it feels much better to say to your CEO that the changes you made gave an x% uplift – rather than having to stop the test and start at square one again.

The real point is: you’re not your customer. Even if you are in the same target market – which for Squared Online I am – you can’t account for other people’s preferences and prejudices. You’re so close to your own website that it’s difficult to predict if a change is going to have the intended effect.

3. Test everything (and wreck your own sandcastle)

So not only do you need to test something, but you also need to do it from the attitude of disproving your hypothesis. You’re setting up something fast that you can iterate from so don’t wed yourself to it, and make sure you’re happy to stamp on your own sandcastle!

Be careful to have everything set up correctly so you have a true, unadulterated picture of the data you’re collecting and you’re aware of all the influences on it that might sway your test. And don’t switch off too early – if you’re going to make an important business decision you need to make sure you collect enough data to make it statistically significant and be able to stick by it.

(A great way to do this is by using A/B testing tools, like Google Content Experiments [†] or Visual Website Optimiser [‡].)

The approach we teach at Squared is about jumping in, trying stuff and failing fast so you can learn quickly without wasting time on ideas and projects that don’t have the effect you intended. With every test that doesn’t go the way you expected you learn something more about your customers.

If you’re going to develop a true culture of testing and learning, you need to embrace the so-called ‘failures’ with the same enthusiasm as those that do prove your hypothesis.

If done in the right way, testing can really move organisations forward – especially when carried out fast and scientifically. Without testing, it’s hard to work out if decisions are in the interests of the end users and the company.

So the next time you’re in a meeting with your boss (or if you yourself are the owner of a rampaging HIPPO) and you feel that for a moment decisions are being made on the basis of opinion rather than customer data, here are three words that are hard to disagree with: ‘Let’s test that’.