Guest Post by Dan Siroker, Co-Founder & CEO, Optimizely, Inc.

Data has gone Big. The last year has seen explosive growth in the amount of new data created and has brought into the mainstream the reality of our ever-growing, globally connected digital world.

Last year, an IDC and EMC study projected 1.8 zettabytes of data were expected to be created and replicated in 2011. For some perspective, as Mashable published, this is equivalent to 57.5 billion 32GB iPads packed full of data (which evidently could be stacked to 25 times as high as Mt. Fuji). Needless to say, it’s easy to get overwhelmed.

Couple that with an abundance of data tools and software, and the fog can at times seem too heavy to drive through.

But, the opportunity for organizations to put this data to use is unlike any before – each data source can potentially represent valuable pieces of intelligence – but just how can someone easily make the most of today's data abundant world?

Keeping a simple approach can lend a hand in realizing the valuable benefits of a data driven world and can yield major insights while maintaining a manageable process. By tapping into large amounts of data and using existing tools to help test and analyze, there are an endless number of ways that leveraging data can effectively benefit an organization.
We see it every day in our work.

At Optimizely, we offer a platform for being able to more quickly run A/B website testing – a platform for easily creating and running tests that also collects and reports the data and analytics during and after the test.
Here are a few tips (by no means is this exhaustive) we've learned to help you keep it simple and power through your next data testing and analysis challenges:

• Harvest Ideas - Sometimes the biggest hump to get over is knowing where to start. With so much going on, there is a seemingly infinite pool of options when it comes to getting the most out of data testing and analysis. Whether you have a long list of internal suggestions, customer feedback forms, webinar notes or a wish list straight from your manager, it’s always good to diversify your sources and keep a constant eye out for possible data tests to run. Harvesting information from your colleagues, user base, industry trends, or social media channels can yield some of the easiest and most fruitful areas to consider as a starting point.

• Organize Ideas and Decide on a Data Test - As you collect possible test ideas, be sure to consider how you will prioritize them. Maybe you organize based simply on the quantity of requests or some other quantitative hierarchy. However, consider the possibility that some tests may negate the necessity for others. For example, if you are A/B testing a change to the overall layout of a website after testing a small change to a call to action, the value of the tests may be compromised if conducted in the wrong order or simultaneously.

• Have a Hypothesis - When you are settling/have settled on your test, determine what you think the outcome will be. For example, "asking fewer questions on a sign-up form should increase the number of user sign-ups". Thinking about the possible results that a particular test may achieve makes for a more well thought out test. Forming a hypothesis also tends to enforce (even sub-consciously) drawing from prior testing and knowledge.

• Have a Control - Make sure when testing to have established a good control (A), where the variation test (B), only varies in areas you are actively testing your hypothesis against. If your test is a variation on a sign-up form and the variation also contains changes to the top-level navigation menu, you may not end up with good data to draw conclusions from.

• Measure the results against established Success Criteria - Did it work? How do you know? In order to know whether or not a test is successful, it is crucial to establish how success will be measured in advance. This can be as simple as measuring changes in conversions on a form, or can get far more complex. Establishing success metrics around a test is crucial to understanding whether or not the results can be implemented to achieve a desired improvement.

• Conclude - Assess the results of your test – are there easy adjustments that can be made based on the data you see? Did the test surface any insights worth implementing, or do you need to go back and make changes to the test to achieve desired results?

The world of impossibly large datasets seems to have arrived, but fortunately technology is providing tools for making sense of the mess. With a simple and well organized approach, data can provide powerful insights and fuel progress without overwhelming you first.