YC-Funded Optimizely Makes It Remarkably Easy To Run A/B Tests On Your Website

YC-Funded Optimizely Makes It Remarkably Easy To Run A/B Tests On Your Website

If you’re running a website, you’ve probably heard of A/B testing, which entails running multiple versions of a site at once and tracking which one performs best with users. But there’s also a pretty good chance that you haven’t actually gotten around to running those A/B tests — the prospect can be daunting, and if you aren’t tech savvy, existing solutions can be confusing. Enter Optimizely, a Y Combinator-funded company with a great pedigree that’s launching today in private beta. The service is looking to make all aspects of running an A/B test as easy as possible, and it looks quite impressive. In fact, when I showed it to our lead developer, he responded “This is brilliantly simple. Can we get an invite?”.

If you’d like an invite of your own, head to Optimizely’s homepage, enter your email address, and use the invite code TECHCRUNCH on the survey you’ll receive. Optimizely will be giving out invites to the first 100 readers, with more to follow.
Getting started with Optimizely is surprisingly easy. After logging in and hitting the ‘Create Experiement’ tab, the service asks you to choose which website you want to run an A/B test on. Optimizely will then load that site, but with a key change: as you move your mouse over each part of the page (text, images, etc.) you’ll notice a light blue cursor highlighting each element. Click an element, and you’ll see a menu that lets you modify the element’s position or size, its image or text, or its Javascript.


Optimizely makes it easy for novices to tweak their pages. Want to make an image bigger? Just drag its edges. Want to move a widget that showcases your most popular posts to the other size of the screen? Drag it over. When I tested this I found the widget editing to be a bit finicky at times, but it’s clear that this has the potential to be very powerful. And it also lets experienced developers directly edit the page’s code, so they can adjust multiple items at once, or make other more sophisticated edits. You can create multiple experiments for the same page at once.


optimizelyshot.png



Once you’re ready to run your tests, adding Optimizely to your site is also simple; you just need add a special snippet of code to each page you want to test out. When a visitor hits your site, Optimizely sends them a small package of Javascript that will tweak the site based on how you’ve set up your experiments. If for some reason Optimizely fails to load, users will simply see the ‘normal’ version your site.


Optimizely tracks the performance of each experiment in a control panel, allowing you to compare the engagement rates of each experiment side by side (there’s a rating showing how confident Optimizely is that your changes made a difference in user engagement). By default Optimizely will pay attention to all user actions, like clicking links, but you can also tell it to analyze how often users are clicking through on specific links.


Optimizely’s founders have some pretty impressive credentials, too. Dan Siroker was a product manager for Google Chrome who also served as the Director of Analytics for the Obama campaign (the campaign did quite a bit of A/B testing, which resulted in an extra $60 million in donations). And founder Pete Koomen was the product manager for Google’s App Engine.


Also see VisualWebsiteOptimizer, which also has an editor that lets you edit elements on your site.
 


Demo works really well. Looks like a neat piece of software. I've been developing a multivariate split tester a few months ago and seeing this makes me happy I abandoned the project.
 
Anyone running optimizely?

I have used a few A/B testing platforms including optimizely, google website optimizer and visual website optimizer. Out of the box, optimizely is the definitely the easiest to get going although I only used visual website optimizer for a short time before they updated it last year.

GWO sucks for a couple reasons outlined here A/Bingo vs. Google Website Optimizer | Bingo Card Creator

If you are building anything in ruby, then definitely use ABingo over Seven Minute Abs for the reasons on the link above. patio11 on hackernews is Patrick McKenzie. The guy is G and knows his shit. One of my personal favorites is Strategic SEO for Startups | Kalzumeus Software
 
Last edited:
I liked the idea of it, until I looked at the pricing. $79/mo for 20,000 monthly users.... you have got to be kidding me. Considering I don't trust a test until a minimum of 1,000 impressions for each possibility, how the fuck is 20,000 total impressions even useful to me. I mean, the tests I ran today had 8,701 impressions over four variations, and that was the third page of the funnel...

Besides that, it doesn't look like it handles pages which require logins or posts to get to.
 
I liked the idea of it, until I looked at the pricing. $79/mo for 20,000 monthly users.... you have got to be kidding me. Considering I don't trust a test until a minimum of 1,000 impressions for each possibility, how the fuck is 20,000 total impressions even useful to me. I mean, the tests I ran today had 8,701 impressions over four variations, and that was the third page of the funnel...

Besides that, it doesn't look like it handles pages which require logins or posts to get to.

Agreed ... I used it for some email submit landers that I was working on so it was dead easy .. but for funnels it doesn t really work ... could use some improvement in this area.
 
Absolutely nothing wrong with Googles WebsiteOptimizer which allows you the same shit plus a hell of a lot more with multi-variate testing and I've had awesome results with it (out of all the shit to hate on about G - this isn't one of them for me)

Multivariate for noobs: You have 5 titles, 4 different layouts, 4 different main videos, 3 different header images, 3 different buy now buttons or even price points you want to test all variations - you populate your setup and it pulls the best combo from all of 'em not just A/B - takes much longer but well WELLLLL worth it. AB or ABC-Z is easy as fack with it too and it's free, and it's also just copy/paste snippets on pages you want tested..

This is just a money sweep imo - the 'software' analyzing the data is easy and is based on predetermined objectives [conversions] so why re-invent and charge stupid money for something that's free and much better right now.
 
  • Like
Reactions: guerilla
visualwebsiteoptimizer.com shits on all of these, especially google website optimizer.

I got 60% conversion increases from dropping a single buy button. Similar results split testing sales copy. It uses a gui interface and the results are displayed graphically.