An Overview of Myna

Myna is a web content optimisation system. What does this mean, and how does it work? Read on!

Content Optimisation

Your website has goals, such as promoting a cause, building a community, or selling a product. Your site’s user experience is designed to support your goals, but there are many choices to make and you can’t hope to get them all right first time. Let’s say you think your sign-up page could be improved. You come up with several variations on your page. Now you want to test which one works best with real users. This is where web content optimisation comes in.

In content optimisation you show different variants of your site to users and measure which best achieves your goals. The traditional method of doing this is A/B or split testing. Myna is a system for content optimisation which offers a better alternative.

Myna’s Workflow

An experiment contains the variants you want to play off against each other. You can run as many simultaneous experiments as you want, and there are no limits to how many variants can be in an experiment.

Every time you want to display a page that running an experiment, you ask Myna for a suggestion. Myna responds with the variant it currently thinks is best, and a token that identifies this choice. If the user then goes on to achieve on of your goals, you reward the experiment, passing back the token so Myna knows which variant led to success.

To Myna the variants are just short (140 character) names. It’s up to you to interpret these names in a meaningful way. We provide a Javascript client to help you do this, or you can program your own server-side code to our REST API.

Rewards are numbers between zero and one, with higher being better. It’s up to you to decide how you allocated rewards though we give some advice in our best practices.

Myna’s Benefits

Myna makes you more money (or whatever is your goal). This is because it starts optimising right away. At every moment Myna suggests the variant that best balances success in the past against Myna’s uncertainty in it’s estimated success rate.

With Myna you can add or remove variants at any time. You aren’t stuck with an experiment once you set it running, meaning you can iterate faster.

Myna adapts to user preferences.

You don’t worry about statistics with Myna, because it takes care of everything. You don’t have to calculate sample sizes, p-values, or anything else.