This message was deleted.
# ask-questions
w
This message was deleted.
c
As datasource we are using a redshift db, that currently only stores data for this test experiment we have 2 tables: • sessions - this table stores the user sessions that have seen the page we want to track, and also stores which variant was served • conversions - this table stores successful conversions, together with a session_id which is the same one we have in the sessions table sessions table create statement:
Copy code
CREATE TABLE sessions (
    session_id INT PRIMARY KEY,
    user_id INT,
    timestamp TIMESTAMP,
    page_name VARCHAR(255),
    experiment_variant VARCHAR(50)
);
user_id
is an optional field only populated with logged in users, but we also support conversion for “guest” users in which case it would be null conversions table create statement:
Copy code
CREATE TABLE conversions (
    conversion_id INT PRIMARY KEY,
    user_id INT,
    session_id INT,
    timestamp TIMESTAMP
);
based on these 2 tables we have also setup 2 key metrics: _*page_sessions*_ which uses the following query:
Copy code
SELECT
  session_id as user_id,
  timestamp as timestamp
FROM
  sessions
conversions which uses the following query:
Copy code
SELECT
  session_id as user_id,
  timestamp as timestamp
FROM
  conversions
f
Hi Davide, What have you set for your trackingCallback in java?
oh, looks like its tracking correctly
(if you're only testing it with 7 users)
it seems like you're really close
I would suggest you view the query used and see if there is something off there
I wonder if its the conversion window
c
hi @fresh-football-47124 thanks for the response, we are actually now starting to see conversions, we did a couple changes and used timestamps over multiple days and now it’s working… not sure exactly what fixed it but it’s looking good now
we saw in a previous thread some other users also add variant_id to their queries so we did that too, it’s not a required column when setting up the metric but maybe that helped too
this is how it currently looks like, also showing a breakdown by dimension…. we will keep playing with it for a bit and add more questions if they come up
f
👍
130 Views