A baseline is a metric’s current or historical performance before changes or interventions. It gives teams a starting point for strategic decisions and goal-setting. section 7.2.2
Establish one by examining historical data and, where available, peer-reviewed research and industry standards. Reviewing distributions, trends, anomalies, segments, time patterns, and correlations helps teams understand what is typical and choose useful targets and time aggregation. section 7.2.2
Monthly customer activity
Customer login records can be converted to the percentage of days active each month, then examined as a distribution or compared across regions and over time. section 7.2.2
logins["n_days"] = logins["month"].dt.days_in_month #1
logins["login_days_pct"] = logins["login_days"]/logins["n_days"] #2
sns.histplot(x=logins["login_days_pct"], bins=32) #3
plt.title("Distribution of the % of Days Active Per Month")
plt.xlabel("% of Days Active")
plt.ylabel("# of Records")
A histogram of monthly active-day percentages illustrates one way to inspect historical metric values. section 7.2.2
A baseline informs a target but does not make it achievable by itself. Use historical patterns and an appropriate level of detail; overly broad or overly granular metrics can obscure useful information. section 7.2.2 section 7.4
