Perfect! The general number of active customers is a good start, but now we'd like to see how many active customers there are in each weekly registration cohort. We can then compare these numbers against the registration campaigns we ran each week and see which marketing activities yielded the greatest number of active customers. Take a look:

SELECT
  DATE_PART('year', registration_date) AS year,
  DATE_PART('week', registration_date) AS week,
  COUNT(*) AS active_customers
FROM customers
WHERE CURRENT_DATE - last_order_date < INTERVAL '30' day
GROUP BY DATE_PART('year', registration_date), DATE_PART('week', registration_date)
ORDER BY DATE_PART('year', registration_date), DATE_PART('week', registration_date);

Compared to the previous report, we added two DATE_PART() invocations (to extract years and weeks) in three places each (in SELECT, GROUP BY, and ORDER BY clauses). This shows us the number of active customers in each cohort.