Chapter 6: Easily avoidable pitfalls

The good thing about starting your optimization and personalization program now is that people who have been doing it for a while have already uncovered many of the mistakes that are easy to make. By knowing about these pitfalls, you can easily avoid them or remedy them.


See Ten common A/B testing pitfalls and how to avoid them for detailed information. In addition to the ideas presented in that article, keep this short list handy to avoid or fix these common testing and personalization pitfalls.

Pitfall Solution
Not having enough traffic to reach statistically significant results. Use the Adobe Target sample size calculator in advance to understand how long the test must run, and then run the test to completion.
Making a change that is too small or unnoticeable. Make the change substantial enough so that it can be visible when you stand a few feet away from the screen.
Failing to align activities with business objectives. Increase focus on the prioritization methodology and communicate that methodology to internal stakeholders.
Having little or no testing backlog. Engage with internal stakeholders and allow them to submit tests that are aligned with key business goals.
Extending activity launches past deadlines Improve documentation for activities and include specific details about the test setup.
Selecting activity success metrics data that don’t help with decision making. Document all primary test metrics and include additional metrics to gain additional insights.
Making changes during the test that could impact results. Maintain a calendar of upcoming promotions and site changes. Clearly communicate Target activity launches to avoid conflicts.
Changing test experiences or criteria close to an activity launch. Request sign-off from key stakeholders early in the test activity process.
Concluding an activity before the results are statistically significant. Verify test duration using the Adobe sample size calculator.
Receiving negative or flat results from a Target activity. Confirm that the activity is worth testing.
Not hard coding the winning experience. Use activity data to show value to get buy-in for hard-coding.
Not sharing results beyond the testing team. Expand communications to include lunch and learns, develop a program dashboard, and schedule stakeholder meetings.

On this page