Data feed orders in Data feed don't match the workspace report only for one marketing channel

[GabrielaNa1]

Hi,

The other day I ran into some weird issue with Data Feeds.

I collected in my local database data feeds extracts daily, starting from Jan 1st, 2023 to date. I usually work with data for marketing channel =Email and the formulas I used to calculate different metrics had worked fine so far. However, yesterday I tried to calculate the number of Orders from the entire data set - all marketing channels (va_closer_ID) and I noticed a big difference from what the workspace report shows and the calculation I use on the data feed. I dug a bit deeper and what I discovered was that all channels report the same numbers, except for one. For this channel the differences are in the order of tens of thousands. If there were a few here and there, I would suspect some delays with the data, but such big of a number is kind of suspicious. 

I use this to determine orders in my data feed dataset:

select count(*) as Orders from <my data feed local table>

where (post_event_list like '1,%' or post_event_list like '%,1,%' or post_event_list like '%,1')

and ... <<other criteria>>

Any idea why this might happen?

Thanks.

[Damonwhall]

Not 100% sure on this one, but I personally have found differences in pre and post data when dealing with the AA data-sync into something like snowflake. 

Post processing is more aligned with workspace. 

Where you are off by orders of 10K however, feels like a difference in a hit based metric vs a visit based metric. 

[Nick_Walter]

You mentioning that there is only one channel that is different is what is rough for me. I would look into if there was an update to your attribution model first. Than the standard looking for any segmentation done on both tables.

After that I would need to assess your data set:
Was it the email channel that changed? Is that you biggest converting channel? Are you looking at visitor instead of total orders (basically only counting loyal customers only once)? If you pull total order by week for the same time period does that align with what you have? Were tracking codes left out of an important email blast?

Just some ideas to maybe add a fresh perspective to look into this.

[MandyGeorge]

@GabrielaNa1, looking at your sample code, one thing stands out to me about your 'where' condition. Is it possible that event1 could be the only event firing? You're accounting for instances where it's the first, middle, or last event in a list, but not where it's the only event. Could that be part of the issue?

[GabrielaNa1]

Thanks for your answer. No, is not that, I don't think that post_event_list can simply be ='1' . In any case, not in my data.