Big News: FounderDating is joining OneVest to build the largest community for entrepreneurs. Details here
Latest Notifications
You have no recent recommendations.
Name
Title
 
MiniBio
FOLLOW
Title
 Followers
FOLLOW TOPIC

Question goes here

1,300 Followers

  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur
  • Name
    Entrepreneur

What is the best way to execute A/B testing on a landing page?

Friends, I am in the process of finalizing a landing page for a new product. My intention is to test various price points, by creating various versions of the page.

Ideally, I would like an equal number of visitors to to see each version of the page. Is there some hi-tech way to do this? Or would I have to run two separate experiments?

11 Replies

Anthony Zeoli
0
0
Anthony Zeoli Entrepreneur • Advisor
Digital Strategy and WordPress Consultant and Trainer
A/B testing is usually done dynamically, so that page A is served to the first user and the page B is served to the second user and it flips back and forth. Both pages should have some marked differentiation of content, style, and offer in order to ascertain which does better.


If using WordPress, there is a plugin called Landing Pages with A/B testing built in. I'm using it myself.
Mikhail Gorelkin
0
0
Mikhail Gorelkin Entrepreneur • Advisor
Algorithm & AI Architect / Developer
I would recommend more sophisticated testing based on the bandit algorithms. Look for example at https://www.chrisstucchio.com/blog/2012/bandit_algorithms_vs_ab.html
Jessica Alter
0
0
Jessica Alter Entrepreneur • Advisor
Entrepreneur & Advisor
Very easy to do with Optimizely or competitors >>
Dirk de Kok
0
0
Dirk de Kok Advisor
Founder and CTO Mobtest
Just create one version in your own HTML/CMS. Then use Optimizely to create variations.
Please note that you need to have a decent amount of traffic to be able to do A/B testings. At an average of lets say 1 % conversion with 20 % difference you want to detect at 95% significance you need over 35,000 views.https://www.optimizely.com/resources/sample-size-calculator/?conversion=1&effect=20&significance=95
Mikhail Gorelkin
0
1
Mikhail Gorelkin Entrepreneur • Advisor
Algorithm & AI Architect / Developer
Assumption for A / B or any kind of multivariate testing: you are working in a * static * environment. It was true for factories, where designs of experiment were developed, but it contradicts with the nature of the Web... So, an easy way doesn't mean a correct way to do things.
Stephen Huson
0
0
Stephen Huson Entrepreneur
Leader in Internet lead generation, SEM / PPC / SEO and analytics
Agree with Jessica...Optimizely or Visual Website Optimizer are good choices.
Haya Levine
0
0
Haya Levine Entrepreneur
Independent Medical Practice Professional
Thanks all. Do you know if I can use Optimizely with a Wix website?
Gil Allouche
0
0
Gil Allouche Entrepreneur
Founder @ Metadata
Hey Haya! Yes you should use a tool like: Optimizely (http://optimize.ly) or Visual Web Optimizer (vwo.com) -- they have a free version and generally speaking are not expensive. You install an embed code on the landing page, then you create a few experiments (same page diffent price, changing location of price etc) decide what % of visitors get that experiment (you can have few versions of the landing page at a time) -- and the rest happens magically. In fact -- it can automatically learn the best price and optimize accordingly. Happy to help further. --Gil | founder at metadata.io
Mikhail Gorelkin
0
0
Mikhail Gorelkin Entrepreneur • Advisor
Algorithm & AI Architect / Developer
Here is a very fresh post: How to Evaluate Machine Learning Models: The Pitfalls of A/B Testing:
http://blog.dato.com/how-to-evaluate-machine-learning-models-the-pitfalls-of-ab-testing
Michael Masello
0
0
Michael Masello Entrepreneur
Digital Marketing Professional
I'm a fan of VWO, but be aware of sample size requirements and measurements of statistical significance when running an AB test. I've seen plenty of tests with low data be declared statistically significant as a "winner" only to end up in a tie or even flip-flop. The smaller the difference in the 2 groups the larger the sample you want to collect.

If you're running this on a small site setup your test to maximize the exposure.
Join FounderDating to participate in the discussion
Nothing gets posted to LinkedIn and your information will not be shared.

Just a few more details please.

DO: Start a discussion, share a resource, or ask a question related to entrepreneurship.
DON'T: Post about prohibited topics such as recruiting, cofounder wanted, check out my product
or feedback on the FD site (you can send this to us directly info@founderdating.com).
See the Community Code of Conduct for more details.

Title

Give your question or discussion topic a great title, make it catchy and succinct.

Details

Make sure what you're about to say is specific and relevant - you'll get better responses.

Topics

Tag your discussion so you get more relevant responses.

Question goes here

1,300 Followers

  • Name
    Details
  • Name
    Details
  • Name
    Details
  • Name
    Details
  • Name
    Details
  • Name
    Details
  • Name
    Details
  • Name
    Details
Know someone who should answer this question? Enter their email below
Stay current and follow these discussion topics?