This month I've got some exciting news to share about Web Design from Scratch, plus valuable insights from a busy month of testing & measuring on our own sites.
You'll recall my comments about the value of optimising a web site rather than scrapping it in favour of a new design that might perform less well.
I have to say at this point, obviously that is only the case if the existing web site is OK! If your current site is crud, you’d be better off redesigning it. Polishing a turd is never going to give you a good return on your efforts.
The way I like to think about site optimisation is the leaky bucket analogy. If your bucket has holes, there’s no point spending your energy using it to move water from one place to another. Much better to spend your energy fixing the holes, then your energy will be much more effective. So, if your bucket is more holes and rust than bucket, of course it’s better to throw it away and get a new bucket.
Firstly, though, some NEWS!
Yes, we’ve redesigned Web Design from Scratch. It’s not the first redesign. You can actually see the very first incarnation of the site here. And there have been a couple more in between.
This time, we’re reorganising the content on the site, which has grown over time, into more useful sections, combining some of the smaller sections. It also sports a bolder, stronger design, still clear and spacious (in fact, taking more vertical space), but overall it’s simpler, with less going on in the side column.
We’ve added some visual interest in the form of shiny transparent plastic 3D icons representing each section, created by Dan in an excellent package called Cinema4D.
A lot of people over time have asked for more structure to the site, such as "Read this next" links, or links to other articles in a section. So I’ve added a full menu at the bottom of each article, linking to every other article in the section.
One neat trick I’ve applied is to create a "visited style" for links that greys out the link and also adds a ticked checkbox (as a background image in CSS) to make it really clear that you’ve already read that article. I have a strong suspicion that this little feature will encourage visitors to read many more articles, so I expect the average pageviews per visit to jump up with this new design.
I’ve been busy trying out Google Website Optimizer on a couple of clients’ sites last month, and also on SaveThePixel.org. The results have been fascinating. Let me tell you what I’ve learnt from my tests on our own site.
With GWO, you have the choice to run simple A/B split testing (where you test out 2 different versions of the same page against each other), or to do multivariate tests, which is where you can set up variants of multiple page features, to see which perform best individually, and (crucially), which combination of variants does best. I’ve been running multivariate tests.
To run an experiment, you choose one page to test, and you need a "goal page", which indicates success. For my first test, I’ve been experimenting with variants on the STP home page copy, and the goal was that someone visit the "Buy Now" page.
The way it works is that, after you insert some Javascript code into your pages, Google dynamically switches the content of various sections for different visitors. The clever bit is that they leave a cookie so that the same visitor always gets the same version of the page when they return, to avoid confusion.
I actually ran 3 different tests over the month. Once I got a clear direction with certain combinations doing better, I re-built the test testing the strongest variants against some new alternatives. This way the design can only evolve to be stronger over time.
GWO gives you a really simple, clear report split into 2 sections. One tab shows the performance of the combinations you have (which can be a lot – important to keep the number of variants you’re testing low, especially if your traffic is low). The second tab shows the performance of the individual variants of sections.
You can switch off specific combinations if they’re performing badly (only if you’re sure, as you can’t re-enable them!), to focus your traffic on the more likely combos.
Here’s a screenshot of the results for the specific page sections from Test1.
From this, you can see that my original first sentence was strong, together with the variants I labelled "Best-seller". The other 2 directions I tried were weaker – but not conclusively (otherwise the bar would be coloured red, not yellow).
I also tried 3 variations of the next introductory section, and both variants I tried performed better than the original copy, likewise with the subsequent section of the content. Best of all, my simpler description of section3 (which is the "This book will give you…" section) made it 35% more likely a visitor would visit the "Buy now" page!
What can we learn from this? Well, the point isn’t that any specific style of copy will outperform the other. That’s not my purpose here.
One thing that does strike me is the difference that a section of wording well below the fold has on conversion rates. That tells us that, with this page at least, people are reading, and reading carefully, and they’re not yet convinced whether to buy or not until they’ve probably read the whole page!
It’s certainly evident from this, and other tests I’ve run, that simpler can outperform more complex, but that’s not a universal rule. There are lots of long pages full of text that have been proven to do a very good job of converting the reader to some course of action.
The takeaway I get from all this is that you simply don’t know whether your first guess at the content will be the best. Hey, it almost certainly isn’t! You can only know how good or bad it is by testing it against other informed guesses. So you might try longer, shorter, more direct, more friendly, softer, stronger, bulleted… whatever. You’ll only find out for sure when you put more than one variant in the ring.
That’s why, when we put together a quote for a web site now, we’re offering an on-going "Test & Measure" programme as standard. Along with on-going SEO, monthly cycle of site content optimisation should deliver real value.
In the case of Save the Pixel, here’s the evidence! This is the recent change in conversion rates on the site week by week.
From a typical conversion rate (visitors to buyers) of around 1% prior to the tests, we’ve raised it to 2.15% last week. What happened last week? I simply stopped the last test, and left the best-performing combination as the permanent home page content.
During the tests, because there were several experimental combinations running, conversion was more erratic (so ignore it), but after I found the better one, I’ve now got a solid home page that I know will perform better than it ever has before – about 100% better!
Whether you’re generating sales or leads, or even just spreading a message, the lesson here is that by moving from "best first guess" to "best of a bunch", you can dramatically increase your success rates.
What I’m doing next is running an experiment on the "Buy Now" page to see how the content affects the number of people who proceed to buy or leave… Watch this space!
If you’d like to optimise your web pages, we’ll be pleased to offer a consulting service. We can help you set up a site experiment, or set it up for you, and suggest or create strong alternative content options to help you rapidly evolve your site to best-of-breed. Just reply to me for more details.
What, you remembered? Don’t worry – there is a free ebook on the way for newsletter subscribers. It’s not by me, it’s written by a marketing and advertising guru I know, and it’s not quite ready to send you yet, but when it is you’ll get it. It is well worth the wait!
Do you love our approach to crafting simple & effective web sites that just work for people?
We'd love to hear about your web strategy.
Contact one of our team today!
Sign up for my regular newsletter to keep updated on everything I'm doing and thinking about in web design.
