Will deleting a processing rule change data 24 hours prior?

[Jennifer_Dungan · Accepted solution]

No.. processing rules run on the data as it's coming into the system. 

So think of what happens when a tracking call is sent.

1. Your website builds the request and sends it to Adobe

2. Adobe does some basic processing - i.e. getting the User Agent and Header information of the call, getting the IP address (and obfuscating it if that is turned on), etc

3. Then comes Processing Rules

4. Now the Vista Rules

5. Marketing Channel Rules

6. Data is added to your suite

Any data that has already been processed (such as tracking that was captured 24 hour prior).

If you need to manipulate "past" data, then you might need to look into classification rules

[abf29]

Thanks Jennifer! I am running a few test in rule builder to see if we can correct the issue.

[Jennifer_Dungan]

Good luck! 

If you are using Regex, I actually like using http://www.regextester.com/ this allows me to test multiple items at once while building the rules, and the hover on each part of the rule allows me to confirm the logic of what I am building.

Don't forget, IF you are creating multiple "overlapping" rules, they are run in order; to make sure the most important rules (the ones that should take precedence) are the last ones in the list.

And luckily, with classification rules, you can keep tweaking them.. as you can reprocess data up to 6 months prior.

[Lin4]

Hi Jennifer, 

Hoping you could assist. 

I forgot to deactivated my testing classification rule builder, which then captured and bucketed the data with 6 months lookback window the next day.

I deactivated the rule, but noticed the 6 months data that has been bucketed in a new classification 'delivery label' didn't revert back to the initial delivery label name, the delivery label still show as the 'unspecified' label that I used when I activated the classification rule builder.

Is there a way to revert back to 'not known' bucket (for the past 6 months) to their initial delivery label, for example delivery label is AAA - BBB - OOO instead of 'not known' (that was set in the classification rule builder before I deactivated it).

Many thanks

Lin

[Jennifer_Dungan]

Hi @Lin4 

I believe that "deactivating" a rule will leave the classification data.. if you put the rules back the way they were and re-process, it should re-process everything back for 6 months.

[Lin4]

Thanks for the quick reply @Jennifer_Dungan , very much appreciated.

I tested it (to reprocess), but I wouldn't be able to get the initial campaign label because they are all different campaign label

So, if I use the Regex of   .* and return $0 it will show 'not known' as 'not known' instead of the initial campaign label AAA - BBB - OOO (because the last recorded label in Adobe Analytics is 'not known').

Any other recommendation please? 

Thanks

Lin

[Jennifer_Dungan]

Ohh wait.. I just realized, did you change a processing rule? Not just a classification rule?

Unfortunately, processing rules actually change the values of the data. I was thinking you only changed a "classification rule" which leaves the original value intact and creates a "sub-dimension" of classification... 

When dealing with classifications, these can be changed and reprocessed... but actual "processing rules" process the data and set it.

However, IF you have access the Raw Data feed.. the original value will show. In raw data there should be two fields for most items

eVarX and post_eVarX (for example)

eVarX is the original value, and anything with post_ is the post-processed value.

In my Original post, when I said "If you need to manipulate "past" data, then you might need to look into classification rules" - that was based on the data actually being collected... if you changed the data collection, you can't really retroactively fix it with classifications.... I was suggesting that as an alternative to Processing Rules.

There is also this possibility  to try and recover data... If these are campaign data (i.e. passed as query strings in your URL) you might be able to look at values captured in a URL eVar if you have one... or use Data Warehouse to export the URL from the (this is one of the raw data fields you can't access in Workspace, but can get through the Warehouse and Raw Data Feeds.. but you won't necessarily have the attribution available to you)

Hopefully you have an eVar that captures the full url and can create a proxy extraction and attribution from that in Workspace.....

[Lin4]

Thank you so much @Jennifer_Dungan , that makes sense, I don't have access to the raw data, I have escalated it to the support to reverse the rules. Such a mess that I made

Many thanks for your explanation.

Cheers

Lin

[Jennifer_Dungan]

If it makes you feel better, I'm sure we've all been there (I know I have)... but here is some advice.. we have both a prod and qa/dev suite set up... I test all processing rules in QA before making any new rules or changes to rules in our production environment. While we can still have issues (rules not copied exactly correctly - i.e. I often make changes to existing rules manually, as I don't want to fully overwrite all rules from QA to Prod in a clean sweep as I may have multiple items in test) it does significantly reduce the chances of having issues.

I know it doesn't help with now, but sometimes we have to take the hits as they come, but the key is to learn from them and grow.

[Lin4]

Thanks heaps @Jennifer_Dungan , for the tips and advise, noted, and very much appreciated. 

Lin