Great post! Please keep sharing your knowledge on DAU/MAU and other metrics relevant to startups.

My contribution to the effort: Can I clean up the shared query?

This looks much cleaner, and should have the same numbers:

DECLARE STARTDATE DATE DEFAULT '2020-03-01';
DECLARE NDAYS INT64 DEFAULT 30;
DECLARE STARTDATE DATE DEFAULT '2020-03-01';
DECLARE NDAYS INT64 DEFAULT 30;
WITH data AS (
  SELECT * FROM (
    SELECT *, PARSE_DATE('%Y%m%d', '2020'||_TABLE_SUFFIX) day, actor.id user_pseudo_id
    FROM `githubarchive.day.2020*` -- `PROJECT.DATABASE.events_*`
  ) WHERE day <= STARTDATE  
), daysN AS (
  SELECT * FROM data 
  WHERE day >= DATE_ADD(STARTDATE, INTERVAL -NDAYS DAY) 
), days60 AS (
  SELECT * FROM data 
  WHERE day >= DATE_ADD(STARTDATE, INTERVAL -60 DAY) 
), stopdays AS (
  SELECT DISTINCT day stopday FROM data 
  WHERE day >= DATE_ADD(STARTDATE, INTERVAL -NDAYS DAY) 
), maus AS (
  SELECT stopday, COUNT(DISTINCT(user_pseudo_id)) mau
  FROM days60 JOIN stopdays
  ON day BETWEEN DATE_ADD(stopday, INTERVAL -NDAYS DAY) AND stopday
  GROUP BY stopday
)
SELECT day, dau, mau, 100*dau/mau daumau
FROM (
  SELECT day, COUNT(DISTINCT user_pseudo_id) dau
    , (SELECT mau FROM maus WHERE a.day=maus.stopday) mau
  FROM daysN a
  GROUP BY day
)
ORDER BY day

DAU/MAU for GitHub:

DAU/MAU for GitHub

The cleaned-up query, compared to the original one:

Thanks, and looking forward to your next post!