You are a lovely Marketer and all you want is the coveted Marketer of the Year award. Yet your website conversion rate is stuck at 2%… no matter how hard you try. It’s not your fault. There are three forces that are working against you… .
* * * * *
If you do not have an active experimentation and testing program in your company then you are leaving so much money on the table that it is not even funny.
It is not too extreme to say: If you are not doing testing then pack your bags now and go home because there is no way you can win.
You are a lovely Marketer and all you want is the coveted Marketer of the Year award. Yet your website conversion rate is stuck at 2%, no matter how hard you try. Its not your fault. There are three forces that are working against you:
- Your website is being used for purposes it has not been created for. And you don’t even know. Few website owners / Marketers / Analysts understand all of the reasons why people come to their websites (Visitor Primary Purpose).
If you don’t know why they are there, how will you know how to create a great customer experience?
- A monolith does not come to your site, segments do. If I unpack that statement what I mean is that even if 100% of the traffic on your site is there to buy (right now), it still is formed of many different types of customers who need to be communicated to in a unique manner (calls to action, content, promotions, and yes even buttons).
- HiPPO’s rule the world when it comes to creating customer experiences. And that’s a bad thing. No matter what you think the optimal customer experience should be on the website it is quite likely that you walk into a meeting room, or office, and regardless of your competence the HiPPO decides what goes on the site.
HiPPO stands for: the Highest Paid Person’s Opinion. HiPPO’s are your bosses, their bosses, your creative directors etc etc.
In the past you did not have too many options to counter the HiPPO’s and get your ideas to the site or why people comes to your site and what problems they have. Now you can run a simple three question golden survey and find out Visitor Primary Purpose.
You can also let ideas democracy flourish in your company: Captain Experimentation to the rescue!
In its simplest form experimentation and testing is showing two or three or fifty pieces of creative / content / promotions etc to a website visitor. After a statistically significant amount of time let the results of the test help you decide what your customers like / react positively to.
In the last 18 months or so your ability to test has dramatically improved, thanks to evolution of technology, new vendors and integrated reporting.
You can get going with your own website’s capability to do A/B/C testing (if you have a decent website platform then you are set).
Take a page on your site, tell your web server to split traffic to three different version, with your web analytics tool measure success. That’s all it takes to get your idea validated!
The cool babe of the month is Multivariate Testing. It is the ability of simply partitioning up various parts of the page and putting different pieces of content in each partition to see which combination of element works.
Here is your default page (and who on this earth does not need, nay deserve , a hundred roses?):
To test all you do is log into your multivariate testing tool, upload all your wonderful ideas, and turn the experiment. You can pump in different images into the main 100 flowers image above. You can put in customer testimonials in the white space on to top right. You can even take your boss’s wonderful idea of adding Buy Now buttons under each product.
The tool takes care of rending the pages to your website visitors and runs the experiment for you (with the only IT involvement for your company being addition of a few lines of javascript code – you the Marketer is finally in control baby!!). Each visitor to your website sees a unique page:
Now here comes the sweetest part. Analytics are included, you just sit back and wait for the green bars to indicate 95% confidence!
The tool tells you which one works for your customers optimally, and by how much. You don’t have to guess.
Convinced?
Here’s one last thing.
You can start doing multivariate testing tomorrow using the 100% free Google Website Optimizer tool. All you need is a Google AdWords account (you don’t even need to spend money on AdWords!!). You can get the code, implement, upload and start testing in a couple hours.
Once you get really good at this you can get a paid tool where you might get lots more features (initially you might not need them but later perhaps you will).
If you don’t want to go down the free route get one of the nice paid tools, Offermatica, Optimost, SiteSpect etc.
To help you choose the right tool for you (and know the kind of effort involved in your company and the impact it will have on your customers) here is a handy selection matrix:
Ms. / Mr. Super Marketer the shackles are now off! Good luck and God speed!!
Tags: a/b_testing, avinash kaushik, experimentation and testing, google_website_optimizer, multi-variate testing, multivariate_testing, offermatica, optimost, sitespect
Avinash,
Good post full of useful information. I particularly like the three questions. Thank you!
I wonder how easy it would be to test not just the page layout, but the entire user workflow in this manner.
I suppose one could just link to unique pages in each instance, but it might be difficult (and costly) to do the accounting.
I left a comment on your three questions post, too. I can see why you call them “golden.”
Excellent work.
Have you seen the Jupiter Research Report? It stated that of all the multivariate testing players out there, Memetrics is ranked number one for a variety of reasons. Avinash, I encourage you to purchase and read this report from Jupiter Research.
Cam : You can totally test entire user workflows, though it requires a bit of technical creativity. A little while back I had coined a term for it “Experience Testing”.
(It is covered more in the book – and the blog – but I left it out here becuase I think I had already exceeded MP Daily Fix words in a post by 100% already!)
Couple ways to do it:
1) If your ecommerce platform supports it, like say ATG, then you simply create the different experiences and the platform with not a lot of work can take care of the splits.
2) You can have A/B/C versions of a page and each funnels people into a new flow (this is a bit more work but doable).
3) With the multivariate option you can still do #2 and depending on your IT complexity sometimes it might be easier using the tools mentioned in the post (becuase they do drive large amounts of IT independence and “set you free” to be a Marketer!).
In summary: It is possible. And I like the way you think!!
Emerson : I agree that the world would be a better place if we only listened to Analysts.
On a serious note, I encourage Marketers to consider different kinds of inputs. Wisdom from Analysts is helpful. So is the beauty of not having to do a three month RFP to find out what you need. Just install a free tool. Become smart by actually doing. Ditto for calling a couple of industry players and getting them to give you a free two month trial.
Learn by doing.
Hmm… Maybe I am a marketer after all, I can come up with slogans!
Thanks everyone.
-Avinash.
Your description of the HiPPO scenario is right on the mark. We see it so often — either the HiPPO drives decisions or, worse, you have complete organizational paralysis because there’s no way to settle arguments.
But most of our customers would disagree with the characterization that inserting “a few lines of javascript code” is going to instantly remove IT and, for that matter, Web designers and coders from the process. It’s just not that simple.
Obviously, it’s in the interest of the JavaScript-based vendors to underplay the fact that unless your content is in old-fashioned circa 1995 static HTML, the page elements you wish to test are going to require reworking. And if your web content is generated by a back-end database (as is the case with virtually any ecommerce site), substantial work on the programming side is going to be required before your first test can be run.
I’d advise your readers to take a long, skeptical breadth when they hear the words “a few lines of javascript code”. Ask some hard questions, particularly about testing dynamic and database-driven content. Then speak to a reference account and ask the simple question, how long did it take to prepare your first test? The answer will surprise you.
Great article! It really got the wheels spinning on the various ways a tool like that could be used. Thanks!
[...] From Experiment or Go Home [...]