Questions? (877) 643-7244
FutureNow Post
Monday, Feb. 21, 2011 at 9:46 am

Testing: Where to Begin

By Natalie Hart
February 21st, 2011

When you begin testing, it can be an overwhelming experience: where to test, what to test, before you even get to the how to test. If you find yourself in this situation, know that you are not alone, but also know that when done properly, testing can yield a great amount of insight and revenue, so it is worth the investment. We hear a lot of reasons why people are considering one of our OnTarget subscriptions for continuous improvement, but one of the big ones is, “We are having a hard time knowing what to test, or even where to start testing.”

Interestingly enough, you need to start at the end. What I mean by that is you want to begin looking for opportunities to improve at the end of your conversion funnel. For most sites this is the checkout. There are two primary reasons why this is where you should begin:

1. By beginning here you’re focusing on Late Stage visitors who already are convinced you have a solution for their needs. Even though these Late Stage visitors most likely will be the smallest segment of visitors, they are the segment that have the least number of obstacles between them and converting. Unless you already have 100% conversion from your cart to your checkout, chances are there are some low-hanging fruit in your checkout. These tests can often be run to significance quickly and with minimal effort (as long as your platform allows for changes to be made in your checkout.)

2. The second reason why you should start at the end is because leading visitors to a dead-end won’t be to your benefit. By starting at the last step, and working your way backwards, you ensure that each improvement you make to your funnel, lands potential customers on a page that is ready to convert them, and you see bottom line results more quickly. If you start testing with a page at the beginning of your funnel, without having improved the pages behind it, your hard work still doesn’t yield improvements to your bottom line: you’ve just brought visitors deeper in your funnel to abandon. You want to ensure that there is a persuasive path created where the scent of one page leads to the next, and the pages visitors are being funneled into are highly performing.

Your funnel is a key area in your initial foray into testing, and having an effective tool for capturing what’s happening in it is essential. Don’t forget, your greatest opportunities to turn a quick buck live here. If you don’t have Google Analytics (GA) yet, get it! There are many GA features I use on a daily basis when testing with my clients, but I couldn’t live without the Google Goals and Funnel Visualization (see snapshot at right). You will need to set up Goals in GA first, and that will activate the Funnel Visualization. If you’re just starting out, set up a shopping cart to checkout goal. It will provide you with a great amount of insight about where to begin testing.

Want more help? Contact us to find out how we can help you make the most of your website.

Add Your Comments

Comments (23)

  1. This is the first post of this blog that I have read, and I am already hooked. As the owner of a new start-up (not-linked), I know that testing is one of our big questions right now. It’s one of those, “have you ever tested enough” type of questions. Thanks

  2. How can I make my funnel ?

  3. I have set up goals in GA but the problem is I can’t see any Funnel visualization there.Please help.

  4. The second reason for working backwards while implementing a testing strategy won me over. You make a great point about making sure the customers are pointed to a desirable direction from each page.

  5. Can something be tested? There are n degrees of possibilities. Can we test them all. This is why software that I developed was always rolled out on a trail basis. Heck. Even Microsoft do that to the Masses…Updates anyone :-)

  6. @DJH – sure, you could test every possibility, but that wouldn’t be a very efficient use of time, energy and money. You must look to your data to show you problem areas with high potential for wins, and start your testing there. And as for variations, you must come up with some hypothesis, rooted in your data and what you know about your customers. Having experience with a slew of different websites and best practices doesn’t hurt either, although we’ve certainly seen instances where a best practice was not the right choice for a client.

  7. @bipolar homeopathic treatment – here are Google’s instructions for setting up goals and funnels.

  8. @abir_ahmed – see my response to Bipolar Homeopathic Treatment, below…

  9. You can set up funnels by picking the landing pages leading up to your goal page and then setting their urls as steps in the funnel.

  10. I am not sure, that the steps are so important. Why this order?

  11. It’s funny how the topics on your blog so often match what I’m focusing on at work!

    I’ve just started working with our developers with A/B testing and boy am I glad I read your article! Now I can ask some pointed questions about their approach to testing (whether it was started on the front or back end).

    Just a warning to other readers, If you are doing A/B testing consider using a 3rd party testing solution (free is fine, Google’s optimizer is great). Trying to test by hand can lead to some messy situations. As an example, if you’re testing landing pages and you try to differentiate the pages on your hard drive by naming the pages landing1.html and landing2.html, but you also upload the pages with the same file names you can cause some serious SEO issues from crawling errors, to dead backlinks and more. It’s much safer to use a 3rd party tool like site catalyst or Google Optimizer. If you do you won’t have to worry about any of those potential problems.

  12. Great post, and thanks for the link to instructions on how to set up the funnel.

  13. There can be n numbers of permutations and combinations for all test cases but it is impossible to test all.

  14. I think you should mention also about the hamagutchi method. But, the article is great.

  15. I have been skipping the step of working backwards step by step. That makes so much sense but we are always trying to fix some step in the middle. Time to rethink. Thanks for the tip.

  16. What type of testing you are talking about? is it means a test of any type or test related to marketing the business in industry?

  17. @bipolar homeopathic treatment and @abir_ahmed – See today’s post (inspired by your comments). It was too difficult to fit all of the info you need into one post, so there will be a follow-up post sometime later this week or early next week.

  18. @Peter Svenson – the buying process looks like this: early -> middle -> late -> $$. By starting at the end and working backwards, you are optimizing the areas of your site where visitors are closer to giving you money. It means you will generate dollar increases (and ROI on your optimization efforts) more quickly.

  19. @Matt G – see today’s post too! Part 2 is in the works.

  20. It is but true that testing can yield a great amount of insight and revenue, and that it is worth the investment and I completely agree with you here. However, I was completely confused on whether to start testing at the end of every conversion. But then again, thanks for the explanation later on showing the primary reasons why and they are really convincing.

  21. [...] some of our recent posts, we mentioned how valuable and useful goals and goal funnels can be to your optimization and testing. Some readers commented that they’d like to learn more about [...]

  22. This is a very insightful article, the testing methods seem top notch I am going to test these methods.

  23. Geez I didn`t know Google Goals had Tunnel Visualization. Man I`ve been using GA for a long time and didn`t even seen that there …

Add Your Comments

 

Print this Article
Share

Natalie is a Persuasion Analyst with FutureNow.

More articles from Natalie Hart

Marketing Optimization Blog
FREE Newsletter Sign-Up
send it once every: