Looking for Hash collision reporting:
Recently we Increased unique limit value for Tracking Code to 1 million (from 500,000). Is there a way to know if this hash collision happens during processing for Adobe Workspace? Thinking out loud, comparing a report from Data Warehouse to Workspace? Any other ideas?
Solved! Go to Solution.
Hey @ddierking
Great scenario!
I believe there's no such feature to find out whether a Hash Collision occured while processing the data for reporting through Workspace.
Your thought about comparing Workspace vs Warehouse reports, sounds good to me.
However, we may ask Adobe product team to share stats around what % of collisions we can expect with 500K and 1M unique values generally. That stands as a benchmark to identify false-positives, while using Analytics data for major business insights.
Best,
Kishore
Dear ddierking,
Can you explain more about Hash Collision?
Are you trying to get an alert or some notification from Adobe whenever Tracking Code's unique limit value reaches 1 million? If yes, afraid that it is not possible.
Thank You, Pratheep Arun Raj B (Arun) | NextRow Digital | Terryn Winter Analytics
Views
Replies
Total Likes
Hey @ddierking
Great scenario!
I believe there's no such feature to find out whether a Hash Collision occured while processing the data for reporting through Workspace.
Your thought about comparing Workspace vs Warehouse reports, sounds good to me.
However, we may ask Adobe product team to share stats around what % of collisions we can expect with 500K and 1M unique values generally. That stands as a benchmark to identify false-positives, while using Analytics data for major business insights.
Best,
Kishore
Are you referring to the Low Traffic thresholds? https://experienceleague.adobe.com/docs/analytics/technotes/low-traffic.html?lang=en
Comparing with Data Warehouse or Data Feeds might help you discover which values have been bucketed under low traffic. In your case, these should be tracking codes that have very little traffic.
Here is a quick article on Hash collision.
Essentially I want to know how often this happens (or at least an estimate).
I really like Kishore_Reddy idea, perhaps you Adobe could share stats around what % of collisions we can expect with 500K and 1M unique values generally. That stands as a benchmark to identify false-positives, while using Analytics data for major business insights.
@Deleted Account Here are the numbers for you to know when the values will go into "low traffic"
For the first 500K values, we check if new keys have less than 100 instances in a day, if it has less than 100 instances it will go under low traffic and if it is more, it will come out of low traffic
For 1M values, we check if instances are less than 10 on a day.
@VaniBhemarasetty Thank you for this information. Really helpful to understand how we can reduce "low volume" row in workspace reports, with 1M unique values limit.
However, @ddierking was checking about "how do we know if hash collision happened" while Adobe is processing data for eVars and Props for a report.
So, in essence, even after increasing the limit to 1M unique values:
In case there isn't such feature (point 2 and 3) available in Adobe Analytics, then it would be helpful to see a common benchmark that Adobe generally has has seen. Like below:
- It's common to see a hash collision for 3% of the hash values for an Analytics account 500K unique values
- It's expected to see a hash collision for 1% of the hash values for an Analytics account with 1M unique values
This way, a reporting analyst (working on Adobe Analytics reports for a business) can always consider that x% of false-positive data in their reports.
@ddierking Please feel free to correct me, if I have not outlined your query correctly
Best,
Kishore
@Kishore_Reddy Thanks for more inputs on the query. Maybe you can submit this as an idea of how analytics can be notified when a hash collision occurs
You can submit it here
https://experienceleaguecommunities.adobe.com/t5/adobe-analytics-ideas/idb-p/adobe-analytics-ideas
Sure sure. Thank you, @VaniBhemarasetty
Views
Replies
Total Likes
@ddierking Did you manage to find out the percentage of hash collisions?
@Deleted Account no luck here.
Views
Like
Replies