This repository has been archived by the owner on Feb 5, 2019. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 44
Compare with GWO
gregdingle edited this page Sep 13, 2010
·
9 revisions
Head-to-head comparison with Google Website Optimizer
====
As far as I know, Google Website Optimizer is the only other free service for experimenting with different content on a live webpage. It is certainly the most popular.
So, naturally, you may want to know how Genetify compares with GWO. I did myself, so I studied GWO extensively and I talked with people who have run it on real money-making websites.
Here is my comparison, aiming to be as objective as possible.
Two-sentence summaries
- GWO rates your ideas based on rate of conversion. Ideas are expressed in HTML and stored on their server.
- Genetify chooses between your ideas to optimize your goals. Ideas are expressed in HTML, CSS or Javascript and are stored inline, in your code.
Feature | Google Website Optimizer | Genetify |
---|---|---|
Store content in your own files | No. You store the things you want to test on the GWO server. | Yes. You store the things you want to test inline, in your own files. |
DRY | No. You will typically repeat the contents of one of your variants in each <noscript> section of your page. |
Yes. The default, no-javascript option is simply one of the variants you designate. |
Automatic optimization | No. GWO operates an experiment that informs you on the effect of the manipulated variables on a goal. It does not actually optimize. | Yes. Genetify manipulates the given variables in order to increase the occurence of the goal. |
Experiment with CSS | By hack. GWO only injects HTML. So, to experiment with CSS rules, you would need to inject non-standards-compliant <style> blocks into the body of your document. |
Yes. See usage. |
Experiment with Javascript | By hack. To experiment with Javascript, you would need to write it inline in <script> blocks. |
Yes. See usage. |
Modify an experiment at any time | No. You can’t modify an experiment once it is running. | Yes. You can add a new variant, fix a spelling a mistake or otherwise modify an experiment at any time. You accept the risks of altering the meaning of the results. |
Automatic setup of new experiments | No. You need to copy a previous experiment and overwrite its variants, or else you need to enter new control and tracking codes in your page. | Yes. When you write the name of a new gene into your code, you effectively create a new experiment. |
Control panel on your page | No. You must log into the GWO site to see all your variants. | Yes. You can load a control panel onto your page at runtime that lets you cycle through all your variants and do other useful things. |
Seperated content as a single variant | No. GWO organizes variants into contiguous sections. | Yes. Genetify allows you to classify multiple elements anywhere on a page as part of the same variant. For instance, a link repeated in a top and bottom navbar. |
Multiple experiments can share the same goal | By hack. For each additional experiment sharing a goal, you need to duplicate the goal and change the duplicate’s ID. | Yes. |
Weighted goals | By hack. You could record the same goal multiple times to add additional weight to some outcome. | Yes. You can assign any numeric value to a goal. |
Multiple goals per experiment | By hack. Again, you could record the same goal multiple times. | Yes. Goal values are summed to determine optimization. |
Attach goal to any event | By hack. GWO expects you to put the goal in a <script> tag on the conversion page. |
Yes. You can simply call genetify.goal from anywhere. |
Visibility into statistical algorithm | No. Google does not say exactly how the significance of results are determined, although it sounds like they are using some kind of regression based on a normal distribution. | Yes. You can look up the genetify algorithm line-by-line. You can also play with it based on real-time results. |
Reports | Yes. | Minimal. Hopefully more coming soon. |
Real-time results | No. You must wait up to a day for results. | Yes. |
Calculations of statistical significance | Yes. Although, since the algorithm is hidden, the validity of the numbers can’t be checked. | No. Hopefully coming soon. |
Testing of interaction effects | Yes. | No. Hopefully coming soon. |
Segmentation | No. | No. Hopefully Genetify will soon take into account visitor characteristics and other variables when optimizing. |
Integration with Google Web Analytics | By hack. See this article. | Coming soon. As a beta feature, Genetify can record each unique combination in an experiment as a unique pageview in GA, and each goal as an ecommerce transaction. |
Host the necessary JS file yourself | No. GWO requires a dynamically generated JS file. | Yes. Genetify.js can be stored on your server, so you can ensure reliability. |
Host everything yourself | No. GWO is only available as a third-party service. | Yes. You can install the genetify server on your own machine and control everything about it. |