Key takeaways

In this module, you learned how to:

• The four key report metrics each tell a different part of the story. Unique visitors shows how many distinct people were exposed to each experience. Conversion rate shows what percentage of those visitors completed the goal action. Lift shows how much better or worse each experience is performing compared to the control — expressed as a percentage. Confidence is the probability that the difference observed is a real signal rather than chance, expressed as a range whose confidence interval you want to see tighten over time without the ranges of different experiences crossing over each other.
• 95% is the industry-standard confidence threshold for declaring a statistically significant result. The confidence ranges map to levels of evidence: below 90% suggests no meaningful difference; 90–95% is weak evidence; 95–99% is moderate; 99–99.9% is strong; and 99.9%+ is very strong. If a result holds at 99.9% for a full week, it may be worth considering the winning experience as the permanent default — but always look for stability over time, not a momentary spike.
• Never declare a winner based on the first few days of data. Early test data is inherently noisy — visitors are acclimating, traffic patterns fluctuate, and conversion results can swing widely in the first week or two. In fact, the experience that ultimately wins was often the worst performer during the first days of the activity. Results need time to stabilize into a consistent pattern before the confidence level carries real weight behind it.
• "Unique Visitors" is the recommended counting methodology for most A4T analyses. The three options are: Unique Visitors (increments once when a user first qualifies — recommended), Visits (increments once per session, even across multiple sessions), and Activity Impressions (increments each time activity content is served, potentially multiple times per visitor). Unique Visitors gives the cleanest person-level view of activity participation.
• Remove the "Unspecified" line item from A4T reports in Analysis Workspace before sharing totals. Unspecified represents site visits from people who were not entered into any Target activity during the reporting period — it is simply traffic that Target didn't touch. Leaving it in inflates every total metric in the report. Removing it (via the funnel icon → uncheck "include unspecified") ensures that numbers reflect only the traffic that actually participated in a Target activity, giving an accurate picture of what the tests influenced.