Testing for Conversions

The second standout presentation from IMC Vancouver was given by Jared Waxman from Intuit who related some of the company's experiences with testing. Intuit has built a testing framework that enables them to accomplish one day testing for new ideas. Quick testing means that they are able to try plenty of ideas and quickly reject bad ones.

Ten Pieces of Rapid Testing Puzzle

Loosely paraphrased the ten steps for quick testing are:

1. Create a testing strike force that has complete access to the site and the ability to make changes where they are needed.
2. Appoint one "approver" for what goes up. Have a single person approve the tests to expedite the decision making process.
3. Create one straightforward criteria to measure test success.
4. Report at the "go live" point of the test and then again when the results of the test are in. Between these two points, let the test run unhindered.
5. Develop ideas fully before implementing tests. While the testing team can tighten ideas, these ideas should be developed to their fullest potential before reaching the testing team.
6. Base test ideas on a reasonable theory and customer data.
7. Manage deadlines with quality and reduce scope as necessary to facilitate running tests and getting results in one day.
8. Take ideas from anywhere and everywhere, but allow the test team to maintain control of those ideas.
9. Maintain a flexible testing calendar. When one line of testing proves unfruitful, an open testing calendar will allow you to change your strategy and implement a new line of testing.
10. Practice one day test sprints. One day test sprints allow for the testing of many ideas and make testing a part of regular business rather than something only done once in a while.

Faith-Based Business?

Both Jared and Avinash emphasized that much of the work that we do on websites is unnecessarily faith-based. There's no need for this approach. Available tools (like Google Analytics and Google Website Optimizer, which are both free) tell us what actions site users are taking and provide information to help us hypothesize what changes we can make to improve the function of the site. All that's left for us to do is to create the tests to ensure changes are positive ones and use the results of those tests effectively and to our advantage.

Bookmark It

Hide Sites

Don't Suck

The IMC Conference really brought in some great speakers this year. A couple of the talks today were really spot on. The first presentation was by Avinash Kaushik, who spoke about the topic of analytics and understanding your users. His Number 1 Rule? "Don't Suck." Such relevant and a happily irreverant advice set the tone for what was to follow. He exhibited pages related to High Speed Internet PPC Campaigns, which highlighted how many local high speed providers are shooting themselves in the foot by not helping people find what they need.

Hurray for Data

People who market online have a great advantage over offline marketers, who must use expensive means to gather data. Online marketers are just swimming in low-cost data. Anyone who wants to can add analytics tracking to their site and, in no time, have a wealth of data at their disposal. Unfortunately, using the data effectively is still pretty tricky.

Segmentation: Making Sense of Mountains of Data

Segmentation is the act of breaking your analytics down into groups that you can better understand. One of the segments that Avinash discussed were the loyalists (people who visit more than 3 of your pages, i.e., they are really exploring your site). Limiting your analytics results to only this group helps you to see where the interested people are going within your site. In addition to using segmentation, you can examine the top ten pages that are visited on your site and calculate at the bounce rates for those pages. The pages with the highest bounce rates are prime candidates for redesign. People are bouncing for a reason. Maybe that reason is fixable!

You're Probably Wrong

One of the great benefits of working in the online world is the ability to find out quickly when you are wrong. Knowing where you can make improvements is great - it lets you hypothesize what problems those people who come to your site are encountering and how you might be able to fix those problems. Avinash has a great example in point five (cinco) of this behemoth blog post of just how inportant the testing part of this process is. Getting inside the human mind is no easy feat and, try though we might, our plans to divine what kind of page and content will help sites make sales seem almost sure to fail. Testing is the only way to be sure our hypotheses are correct!

Bookmark It

Hide Sites

The Internet Marketing Conference is setting up shop in Vancouver for September 16-18 and we're going to be heading to Vancouver along with the folks from RevenueWire to take in the conference. I'm particularly hoping to connect with Gillian Muessig from SEOMoz and a number of publishers who might be in the market for the written word. The keynote addresses look great, presentations from Google, Facebook and LinkedIn, should really help focus our ideas on where we need to move next to keep the site growing and moving forward.

If you're coming to the conference and are interested to meet up, let me know!

Bookmark It

Hide Sites