People preach that using testing to tune and optimize Web sites is easy. They say the tools are out there and aren’t terribly difficult to implement. Yet, there are some clear problems that occur if not approached the right way.
Last time, I recapped some trends from the Omniture Summit, including the move by more companies to optimize their Web sites. Tuning and improving site performance through testing (A/B, multivariate, etc) is positive because there’s often a lot of low-hanging fruit.
At the summit, many conversations revolved around testing. A relatively small number of companies say it’s a regular occurrence within their business, but nearly everyone says they’re starting to experiment with it or plan to very soon.
Testing Factors to Consider
Let’s assume a company invests (dollars and resources) in a testing tool, whether it’s Omniture Test & Target, the free Google Optimizer, or other tools, and gets it implemented properly. Let’s go one step further and say these companies set the tests up correctly with the correct measurements of success, statistical significance, segmentation and the like. So far so good, right?
These may seem like basics, but when most companies start experimenting with testing they spend a lot of time struggling through these few basic things. Dedicated companies will figure these things out over time, but it can be painful and time consuming.
But that isn’t where most companies fail. Most companies really struggle and leave money on the table by failing to understand the difference between treating site testing as a project versus a program. Ideally, companies want to “programize” their site testing — this means thinking of it holistically and as an ongoing initiative instead of just a series of one-off tests.
Most companies start by looking at low-hanging fruit — things they believe might be weak areas, that customers have complained about, or what an executive doesn’t like. These are all natural places to start with and often will show some return.
But as time goes on, low-hanging opportunities become fewer and farther between and less fruitful in terms of their impact. Companies begin to struggle. One of two things happens: the desire, focus, and willingness to dedicate resources starts to fizzle out or they get smart and realize they need to look at optimization differently.
Building Testing into a Program
What does it mean to move from a project view to a program view? It means looking at testing holistically and as an ongoing initiative, not just testing something, picking the winner based on the results of one or two measurements of success, and then moving onto the next thing. It means taking lessons learned from each test, understanding the why behind tests, running follow up tests, and many more things.
Here are a few things that make a successful “programized testing” initiative:
- Testing focused on areas of the site with the greatest potential impact to the business — focus on the areas that really matter and can drive the most value.
- Testing focused on business goals for the Web channel — not just things you think you can improve. It’s often easy to identify things that seem easy to change, but they may fall on a portion of the site that only impacts 1/1000th of 1 percent of the site traffic and isn’t related to a key conversion rate.
- Testing paths, not just individual pages. A page is just one piece of the puzzle. To get the most lift out of your testing, consider testing along the entire path of success.
- Understanding the true impact of tests outside of just one or two behavioral measurements. Do you know the attitudinal impact of that test you ran last week on the people that didn’t convert? Do you have any idea how they were impacted? You can often tell when a company truly has programmatic testing when they don’t just measure improvements to single metrics, but also tie it in to overall site satisfaction. That demonstrates that the testing plan is wrapped into larger site vision.
- Documenting and sharing learnings from all tests throughout the organization. Well-structured tests can drive insight throughout the Web team.
- Sharing online learnings with offline channels when applicable. This could be for offline marketing, customer service, merchandising, etc.
- Segmenting your messaging/experience based on test results — don’t just pick one winner for each test. Look for ways when it makes sense to pick a few winners based on different audience, days of week, segments, etc.
- Allocate resources so that you can quickly, constantly, and strategically test new ideas (again based on greatest impact to the business).
- Perform pre-testing as a part of a larger redesign effort. Running a series of tests validating which things are broken and what customers are clamoring for as you plan out and build towards the redesign. And of course, serious companies are launching redesigns with a slew of tests built in.
Again, site optimization is a key trend we’re seeing. Testing is taking off as the economy slows and as businesses must get more out of their sites with fewer resources and less money.
Take the time to really do it right. Low-hanging fruit won’t last forever — and the potential upside is much greater when you programize the process. Shoot me a note and let me know what you have done within your company to address many of these things.
While ad fraud has become part of every marketer’s vocabulary, attribution fraud—the practice of gaming outdated attribution models to justify self-serving means—has ... read more
When you’re just starting out as a business owner it’s easy to become wrapped up in the seemingly endless number of metrics ... read more
Something I’m asked frequently at conferences and from marketers is what metrics they should be striving for from their social media marketing. ... read more