A/B test
An A/B test splits visitors at random between a control (A) and a variant (B) and compares a metric such as conversion rate. Because the groups differ only in the change, a difference larger than chance can be attributed to it. Decide the sample size before starting, run whole weeks, and judge the result by its p-value and confidence interval rather than by peeking.
Related terms
Activation rate
The share of new sign-ups who reach the moment they first get real value from a product, such as creating a first project.
Churn rate
The share of customers (or revenue) lost in a period. Monthly customer churn of 5% means losing nearly half your customers in a year.
Customer lifetime value (LTV)
What a customer is worth over their whole relationship with you, usually as gross margin: monthly revenue × margin ÷ monthly churn.
Net Promoter Score (NPS)
A loyalty score from −100 to +100: the share of promoters (9–10 on “how likely are you to recommend us?”) minus the share of detractors (0–6).
RICE scoring
A prioritization method from Intercom: Reach × Impact × Confidence ÷ Effort, so ideas can be compared on one scale.
Assumption
A belief a solution depends on. If it turns out to be wrong, the solution fails — so the risky ones are worth testing first.