Most of the times you create and run A/B tests on all your website visitors. But there may be cases when you would like to test your ideas only for a particular segment for visitors. For example, you may want to test if a stripped down layout of website works better for returning visitors. Or you may want to test if a Twitter-specific banner gets more clicks from visitors arriving from Twitter.
New feature: Segmentation
We are proud to announce that we have launched Segmentation feature in Visual Website Optimizer. Using this feature you can choose to run the test only for the visitors who qualify your segmentation conditions. Have a look at the following screenshot:
You can choose and combine any of the following conditions available for segmentation:
Through the seven conditions above, we have covered most of the segmentation scenarios. Is there any other condition that you think could be useful for segmentation?
Behavioral Targeting and Personalization Possibilities
This is the most interesting part of segmentation. Thanks to flexibility of running multiple tests on the same page, you can personalize landing pages according to visitor segment. For example, if you have a generic landing page for all your marketing campaigns, using segmentation you can show campaign specific headlines and images to different types of visitors. In other words, thanks to segmentation feature, you can target your visitors with segment specific landing page and test it against control if it really drives up conversions. (You can also do neat stuff like inserting search keywords into variation or personalizing page according to city or country. See Advanced code mode for details).
We are *VERY* excited about segmentation feature as it allows power users to run targeted tests and personalize experience for visitors using a super-easy WYSIWYG editor. If you have any ideas or suggestions, please leave a comment below.
Do you like the segmentation feature? Would you actually see a use-case for it on your site?
RSS feed for comments on this post. TrackBack URL
I’ve been using this feature since this weekend and it’s been pretty great. I was able to run the test on just people coming from search engines for my money phrase so that other traffic didn’t pollute the results of the test. Kudos!
Comment by Melvin R. — May 4, 2010 @ 8:51 pm
Great feature, writing it up in The Book now. (Hi to any readers who see this via the footnote
A couple of suggestions:
- Geotargeting based on IP to test country-specific messages, alternate marketing copy etc.
- Browser based segmentation, eg run tests of a simplified style sheet for just IE6/7 users, or ignore older IE versions to run an advanced alternate CSS-based design for modern browsers.
- Browser/screen size – maybe detect viewport size to run tests for mobile devices (or to ignore mobile devices), or for people who use big windows on high res monitors etc.
Cheers!
Comment by Luke Stevens — June 21, 2010 @ 12:14 pm
Fantastic suggestions! Though segmenting on browser, OS is something users rarely do. Though geo-segmenting is a great idea.
Comment by Paras Chopra — June 21, 2010 @ 5:19 pm
I want to show a test only for *landing* visitors (ie. only show test if this is their first pageview on my domain).
How do I segment for this?
Comment by J.R.Schriver — July 27, 2010 @ 2:25 am
Hi Jens,
That will require minor amount of customization. What you can do is to drop a cookie at the bottom of your pages and then make a segment which checks the value of that cookie. First time visitors will not have that cookie, so will be distinguished with any visitor who has seen at least 2 pages on your website. Email us at support@wingify.com if you need more help.
-Paras
Comment by Paras Chopra — July 27, 2010 @ 10:32 am
+1 on “Geotargeting based on IP to test country-specific messages”
Comment by Melvin Ram — July 27, 2010 @ 7:24 pm
Also, in addition to select a segment, it would be great if I could exclude a segment. On my website, I get a lot traffic from India, Pakistan & Philippines. 99% of those people are simply looking to get my contact info and spam me with offers for outsourcing work. I would like to exclude them from my tests.
Comment by Melvin Ram — July 27, 2010 @ 7:27 pm
Can we use some other segmentations like new/returning visitors, geo location or operating systems for the users?
Comment by webdesignMTL — November 27, 2010 @ 5:17 am
@Yes, you can use new/returning visitors but not geo-location (yet).
Comment by Paras Chopra — November 27, 2010 @ 12:13 pm
Has geo-location based targeting been implemented yet?
Comment by Melvin Ram — January 8, 2011 @ 1:54 pm
@Melvin: no, not yet. But we recently released other improvements in the product.
Comment by Paras Chopra — January 8, 2011 @ 4:43 pm
Can you please explain me one thing? Once you segmented the traffic let say new vs returning on your home page, and run a test only on a new visitors. A test win and you find a page for new visitors which converts better.
So my question is so what then? How are you going to show this winning page to new visitors only, and to your returning visitors the current home page. It’s not an a/b test anymore, it’s personalization and targeting.
Comment by jan — May 13, 2011 @ 3:43 pm
@Jan: in that case you can disable the control and let the test running. So, all new visitors will see one kind of page (which is variation) while all returning visitors will see your original page.
Comment by Paras Chopra — May 13, 2011 @ 5:22 pm
[...] Since the primary objective was to increase engagement on the landing page, the following segmentation trick was [...]
Pingback by How WriteWork.com increased sales by 50% & doubled conversions by A/B testing a radical new design — May 13, 2011 @ 8:11 pm