Sampling in struggle analytics is determined by the percentage of sessions, not by a specific number of sessions. For example, if the number of daily sessions is 2 million, the system performs about 20% sampling.

Sampling doesn't mean less accuracy—20% sampling doesn't mean 20% accuracy. When the sample size is more than 30,000, a 20% sample provides 99% accuracy with a 99% confidence level. For organizations with a session count of 1 million, a sample size of 9,000 provides about a 1% error rate.

Sampling is an efficient and cost-effective method for struggle analytics. Tealeaf performs full processing on sessions to accurately collect KPIs, such as total page view, total product view, and total revenue. With struggle analytics, you want to find ways to improve the visitor's experience, so sampling is sufficient to determine the struggle percentage on a specific page (such as median loading time and step counts) or certain UI elements.