I got the right answer (and same SQL results) for the exercise even though I counted event action ‘Signup’ instead of DISTINCT pageview_id.
As a rule of thumb, I understand that it’s preferable to count unique identifiers for conversion rates. But are there any risks to using my method below?
My query is below for reference:
SELECT p.device_type, 100.0* COUNT(action) / COUNT(DISTINCT visitor_id) AS cvr
FROM web_analytics.pageviews p
LEFT JOIN
web_analytics.events e
ON p.pageview_id = e.pageview_id
AND e.action = 'Signup'
WHERE url LIKE '%/books/%'
GROUP BY 1
ORDER BY 2 DESC