Today I thought we’d have a nice simple post for any beginners out there who are not sure how to approach CRO. So let’s try to answer a simple question:
“I want to start testing, but I don’t know what to test”
Ok, the first step is to figure out what your key performance indicators are, which is basically just a posh way of saying “when a visitor arrives at your site, what do you want them to do?”
The most obvious answers are:
- Buy something (complete a sale)
- Fill in a contact form (a lead)
- Request a call back (a lead)
- Sign up for email updates (mailing list)
Although of course if your website has a more complicated business model, or perhaps if your website is in fact a blog, your goals could be different and more varied.
It can sometimes make sense to group goals by page types too. This should all be reasonably obvious, but writing it down is a good idea. So for the Think Traffic website our goals might be:
- Fill in contact form
Click on link to services page
- Other key pages
- Fill in contact form
- Share post socially
- Click through to another post
Obviously, when a user arrives on a blog post we love it if they get in touch and ask for our services, but the main point of our blog isn’t to sell, it is to build an audience and also just to help people out – good karma if you will.
So by understanding the goals of your main landing pages you can choose goals accordingly.
Choosing Which Pages To Test
The next step is to choose which pages to test. Eventually you will want to test every page, but it doesn’t always make sense to test them all at once, and it may not even be possible.
Which pages you test first will depend on things like:
- How much traffic do you have?
- How does that traffic flow through your site?
- What are your most important goals?
We for instance would probably optimize for leads generated by homepage before testing for social shares, because that is the metric that impacts our bottom line most directly (we are a business after all!)
Where In The Funnel?
Standard advice is to test further down the sales funnel. So if you have an ecommerce site for instance you might choose to optimize for checkout completions because improving that metric will directly impact your sales…
Whereas testing your home page to optimize traffic flow to category and product pages may not directly increase sales.
Further down the funnel there is less traffic and conversion rates are generally lower. Or in other words:
“If you are only making 3 or 4 sales per week
running a test on your checkout (for completions) will take a long time”
Whereas your home page will be getting much more traffic and running a test to reduce bounce rate is likely to reach statistical significance much sooner.
And if you can reduce your home-page bounce rate, you will be filtering more traffic down your funnel, which makes those higher impact tests more feasible.
Testing Many Pages:
Of course, it is possible to test more than one page at a time. In fact there are two ways to do this:
Such as by testing every product page at the same time. You will of course be testing for general layout changes or structural changes and in doing so you can aggregate the traffic through all of those pages (like this).
The other option is to run more than one test at a time. This could involve running a test on your home page, another on your about us page and another on your product pages.
This is certainly a viable strategy; it won’t speed up your tests (the time taken to reach significance) but it will of course mean that you can run more tests sooner.
You have to be careful though. Running two or more tests within the same conversion funnel can skew your results if you are not careful. To avoid problems you have to consider whether changes on one page affect user behavior on the next:
On an ecommerce site, you run a test on the home page testing promotional messages, including a “free shipping” promotion. You run a similar test on the product pages. Clearly, the effectiveness of a promotion on the product page could affect the effectiveness of another promotion on the home page, particularly if the two conflict.
As a rule of thumb you should be ok if you ensure that simultaneous tests run further down the funnel only test layout, style and usability elements rather than changing the overall marketing message.
Which Elements To Test
Ok, so hopefully by now you have selected the page(s) to test and you know what your goals are for each page. Next you need to figure out what to test. We will cover this in more detail in another post, but some rules of thumb:
- You can test anything, but bigger changes produce bigger results
- You can test changing your marketing message (eg. Copy)
- You can test your usability (eg. Navigation, layout, etc)
- You can tweak and test random things (eg. Call to action colour)
Generally your marketing message and navigation elements are going to produce the biggest results, so focus on those first.
Which things you test will depend on how your traffic is currently acting, so you will want to do some analysis using analytics, heat maps and click maps and perhaps try some session tracking – all of which have been covered thoroughly in those links!
And Finally – Share This!