Start with a problem statement

In designing a Google Optimize test, it is important to start with a problem statement born from specific qualitative or quantitative research (user testing, survey, heatmap, Google Analytics report data – including data in Analytics from previous Optimize tests) that impacts business objectives.

Brainstorm problem statement ideas and then validate them with data.  Or, observe data patterns and deduce problem statement ideas.  A problem may be more meaningful for a specific segment of users – which research should help identify.

Problem statement50% of paid traffic users that do not submit the primary lead form make entries in multiple fields before exiting the form page without submitting it.”

From that problem statement – form a test hypothesis:

  1. IF – a digital property element is modified, added, or taken away.
    • The proposed solution to the problem statement.
  2. THEN – a predicted outcome will be realized.
    • Impact on business objectives.

Hypothesis“By presenting a lead form for paid traffic that has less form fields, the number of lead form submissions associated with users referred from paid traffic will meaningfully increase.”

Attributes of a good test hypothesis change

When determining the change you want to test in your hypothesis, the following change attributes will increase the return on your investment in the test.

  1. Easily noticeable.
  2. Specifically intended to motivate behavior.
  3. On a high traffic page.
  4. Addresses an issue revealed by research.
  5. Low degree of difficulty to implement.

The “return on your investment” for a test is learning.  Even if the change doesn’t produce the predicted outcome, failed experiments provide valuable information that will shape your decision making going forward.

Start A Positive Testing Cycle

Use Google Marketing Platform product Optimize 360 to test a hypothesis to improve your digital property.

The testing framework below from Craig Sullivan on seems to line up with what is noted above.  A problem is observed, and an “if/then” hypothesis follows.

1. Because we saw (data/feedback)
2. We expect that (change) will cause (impact)
3. We’ll measure this using (data metric)
1. Because we saw (qual & quant data)
2. We expect that (change) for (population) will cause (impact(s))
3. We expect to see (data metric(s) change) over a period of (x business cycles)

Anatomy of a Well Formed Test Hypothesis