A very interesting presentation at September's 2010 Think Visibility by Dr Karl Blanks. Karl is a former rocket scientist who co-founded Conversion Rate Experts applying scientific techniques into internet marketing. He has collaborated with Google, Sony, Vodafone, BT, SEOmoz, SEOBOOK, AVG and other big brands. He has been carrying out split and multi-variant tests since 2004 which made him a pioneer in the conversion world. For SEOMoz alone they generated more than $1m revenue through their conversion analysis and recommendations.
Dr Blanks started his presentation stressing that the main principle is to get more customers without spending a penny more on advertising. You need to suggest changes to web site owners and always ask them to perform split tests, monitor them and then compare what was happening before and after.
Most common causes of conversion death
No Split Tests
You should turn your website into a testing laboratory. That means that there should be ongoing before and after A/B split tests. Launching a new site without measuring the effect it will have on your conversions is a big mistake. New test pages should normally go live on a Sunday night and then you can start comparing conversion, revenue, sales etc on a weekly basis.
The idea is that you develop two or more different versions of each page that you want to improve. User A will see page A, user B page B, user C will also see page A, user D the page B etc. This way your different users get on both versions of the same page(s) and you track all the data associated to their actions.
After a few weeks you will be able to judge and decide which version works better and just keep that one and scrap the one that didn’t work. Not all tests are going to win, sometimes you will need to go back and keep the existing page(s).
Multivariate (or multi-variable) testing is about doing lots of split tests at once on the same page.
Google Website Optimiser is a great free tool to use.
Garbage in -> Garbage out.
You need to think carefully about how long an A/B split test would take so your data is as qualitative as possible. For instance, you would need at least 15-20 days for a conversion test on a site with 10,000 visits per day.
Customer Not in "Shopping Mode"
Some people are just browsing, researching or window-shopping. That doesn’t mean that they should be overlooked as they are also potential clients and as such you can still benefit from them. For instance, try to capture their email address as early as possible so in your next marketing campaign you they can receive an email.
Match.com – This is a good example of a site that captures the email address early, rather than late so they can market their site later, after the user has left their site.
Unclear Value Proposition
This is the situation when it is not obvious what the site does. There’s nothing more frustrating for a user than visiting a site where they can’t figure out what the site is about. You should try asking your visitors for feedback when they leave your site or after they have completed a task. Some of them will perform the task.
Something that is common in Japan and works great in other countries too are cartoons that explain what the site does.
Luck of Trust
There are many different techniques to increase trust.
One way is personalisation so you take advantage of your expertise and present it to the client. A very good example can be found at http://www.mercola.com/ , which is a Natural Health Information Articles site and Dr. Mercola appears in a prominent place in the homepage titled ‘Why Trust Me?”.
Other things you could do to give the site an advantage include:
- Expert reviews
- Customers testimonials
- Media mentions
- Celebrity associations
You should capture as much as possible from the above and incorporate them on your site.
Niche Irrelevant Design
Design which is visually relevant to your niche market is also another way to increase conversion. Think of your audience and design their site for them. Find out who they are, about their age, gender, what they like, dislike, etc.
Being a generalist
The power of niching – How much chance would you stand trying to sell a shampoo called “shampoo”? Specialists are supposed to be better than generalists.
Stick to a few products
Affiliate sites have a big advantage compared to any merchant. This is that they aren’t so bind to sell products of one merchant only. They can be selling anything in a similar manner to what Amazon does.
Get someone to read out loud the content that appears on your site. You won’t believe how many issues can be identified this way.
UserTesting.com - A good question to ask of resource for user testing is - is it a low cost usability testing service which is fast and easy to set up? This way you will manage to find out why users leave your site as you get written summaries from different users describing the problems they encountered.
It is also ideal to find “normal” people to test your site. Having your site tested by geeks only is not a good idea. Ask your neighbours and people who are not online very often.
Silverback – This is another great tool for usability testing, ideal for designers and developers alike. You can download it free for 30 days.
Surveys are another great way to get feedback from your visitors that will help you increase your ROI.
Bad Prior Experience
A previous bad user experience can hurt you in the future. That's not just on the site, but could be bad product delivery experiences too. Returning users with that prior traumatic experience are very likely to leave your site instantly, if they even bother to open your site at all.
Too Much Perceived Risk
Think of how to implement a brave and convincing guarantee.
Prospect Defers Decision
Don’t leave things for later. Get some sort of action and commitment now.