Open source platform for stress free deployments, measured impact, and smarter decisions.

GrowthBook Users

image.png

Hello, can someone guide me on how the traffic and split is applied? I have my control and variant segments in the same source table in ~50-50 split (total users=4128841, control=2064184, variant1=2064657).  I have traffic set to 100 and split 50-50 however in my result the traffic allocation looks off . Also when I change the split I get sample Ratio mismatch warning.

User count wrt to dimension

```varient=0 and value=1
2062596
varient=0 and value=2
2061
varient=1 and value=1
2062117
varient=0 and value=2
2067```

thats int value with 1 as impression and 2 as user clicked on ad

ah, so those should probably be metrics and not a dimension

you would have 2 metrics, one for impressions and one for "clicked on ad"

yes so in my metric query I am filtering user with where condition (`where value=2`)

I want to measure click on ad wrt to total impression shown, so I defined metric with where `value=2` .

Also if I dont define dimension it keeps on complaining about dimension cannot be resolved

<@U01TCPDB58C> looks like there is bug in the dimension query, by default dimension query uses `substr(0, 10)` however the index position for first character starts from 1.

```__distinctUsers as (
  -- One row per user/dimension
  SELECT
    e.user_id,
    substr(
      to_iso8601(date_trunc('day', e.conversion_start)),
      0,
      10
    ) as dimension,
    (
      CASE
      WHEN count(distinct e.variation) &gt; 1 THEN '__multiple__'
      ELSE max(e.variation) END
    ) as variation,
    MIN(e.conversion_start) as conversion_start,
    MIN(e.conversion_end) as conversion_end
  FROM
    __experiment e
  GROUP BY
    dimension,
    e.user_id
),```

Ah, thanks for catching that. We only use substr for Presto/Athena. I'll fix it to be 1-based.

I fixed the substr bug in the latest Docker build