Define Your Objective and Success Metric

Start by clarifying what you want to achieve and how you'll measure success. For example, if your goal is to drive more Verified Visits, set that as your success metric.

State a Hypothesis

Frame a question, such as, “Will a CTV ad with a ‘Buy Now’ end card increase my conversion rate?” Then, turn it into a hypothesis: “My conversion rate will increase when using a ‘Buy Now’ CTA.” Document this, and after the test, check if your hypothesis was correct to inform future strategies.

Identify the Variable to Test

Focus on one element at a time—like the call-to-action (CTA) or ad length (:15 vs. :30)—to see how it specifically impacts performance.

Note Baseline Performance

Review your past performance data to establish a baseline for your success metric. This benchmark will help you determine the reach needed for statistical significance.

Set Up Control and Exposed Groups

Activate a control group (following existing guidelines) and at least one exposed group (with the variable change). For example, if you’re testing a “Buy Now” CTA, your control group could have the “Available Now” CTA, while the exposed group uses “Buy Now.”

Set an Estimated Timeframe Based on Your Sales Cycle

To gather enough data, consider your sales cycle. If conversions typically take 10 days, aim for a test period of at least 14 days to capture reliable insights. Extend this duration if your sales cycle is longer.

Allocate a Meaningful Budget

Choose a budget that allows you to collect sufficient results. Factor in the cost of achieving your objective and the volume needed to draw clear conclusions.

Analyze Results

Return to your hypothesis: did the “Buy Now” CTA outperform the control ad’s “Available Now” CTA? If so, consider shifting more budget to that creative. If not, it’s a valuable learning to refine future campaigns.

💡 Pro Tip: Revisit your report no sooner than 72 hours after making changes for initial insights. The longer your changes are live, the more accurate your analysis will be.