Hey there, I am looking to implement a user retention metric after viewing an experiment. I'm using:

Metric: Retention

Fact table: Events, Row filter: event_name = 'page_view'

However the lowest time I am able to set is: Event must be at least

1 Minutes

after experiment exposure Which effectively is

timestamp >= (exposure_timestamp + '1 minutes')

However, I wish I could use:

timestamp > (exposure_timestamp)

Note: not equal Is this possible with a fact metric?

Hi there! This is a great question... I don't think there is!

For some clarity, we are looking to have a metric that approximates "bounce rate" after being exposed

It's imperfect, but we could consider an option in the Retention metric setting.

In this case, >= is close, but imperfect. We are looking at adding a modal to a certain page. We intend to track close events, but we want to know if someone closed the page (not the modal) without further action. Hence, us looking into a bounce rate metric.

I see. So you need to rule out any concurrent events.

I'd even be happy with

>=

current_timestamp. The real problem is the 1 minute fidelity is not quite low enough for us to feel like the metric is an effective approximation of bounce rate

Ok you can do that just with a proportion metric.

i.e. a metric like:

SELECT
  `user_id` AS user,
  1 AS value
FROM
  `Events`
WHERE
  (event_name = 'page_view') AND
  -- Only after seeing the experiment
  timestamp >= exposure_timestamp
GROUP BY user
HAVING
  COUNT(*) > 1

Correct?

Yep I don't think you need the user filter, since if they have 0 rows they'll be a 0

Ah, I see. I wasn't sure the page_view event in GA would fire before the experiment, but I know it will in our system. Thanks for the help! One last question which is really me rubber ducking to you: Our GA4 session lifetime is 30 minutes (sliding with each page_view). I assume it may make sense to limit to the conversion window? i.e.

-- Only after seeing the experiment
  timestamp >= exposure_timestamp AND
  -- Conversion Metric Window
  timestamp < (exposure_timestamp + '30 minutes')

If a user comes back days later we they would be outside the concept of a "bounce".

Can you explain what you mean by "session lifetime is 30 minutes"

Yes you are right, this is the inverse of a bounce. "session lifetime is 30 minutes" is meaning that our GA setup will only consider a user the same person for 30 minutes. if they leave and come back an hour later GA should treat them as a new session, not returning

Got it.

Thanks this helps clarify a lot. I would recommend thinking about how something like this could be made easier with the retention metric UI. Obviously this is all hyper configurable, but my first gut check was to try the retention metric

I would recommend thinking about how something like this could be made easier with the retention metric UI.

Yep, reasonable.

As someone who is capable to write the sql, I appreciate not having to think too hard and focus on the code changes rather than writing metrics. I appreciate all the work you all have put in!

Yep, thanks for your feedback.