What is a Data Discrepancy/Inconsistency?
A data discrepancy or inconsistency occurs when destination data (the data showing in your involve.ai dashboard) does not match the source data (for example, the data in SalesForce or Gainsight). It happens when your data has been inaccurately mapped, aggregated or segmented/filtered.
Why not to worry?
While it can be disconcerting to discover a discrepancy, there is no need to panic; it does not mean that your data is all wrong and you need to start from scratch! Typically, it will require only a small adjustment on the part of your system administrator or involve.ai’s Data Science Consulting team.
Common Causes
Data isn't unique (parent/child issues)
The source data may have parent/child account hierarchies without reference, duplicate IDs, tables with no reference IDs to tie things together; these are a few challenges where data across various sources isn’t unique.
Pulling data from different fields/tables
This occurs when data is fetching from inaccurate columns; On the user interface (UI), the displayed column names can be different than what they are on the backend. This usually happens in the cases of Formula Fields, Roll Up Fields. Updates made on these fields from your side might not reflect on the destination tables on our end.
Timeframe of Data (We normally grab one year worth of data)
In order to grab the most relevant data without any biases, we prefer to map one year worth of data but we can always work with more or less data when there’s a business context/limitation attached with it.
Interaction data isn't being logged by CSM
This occurs when, for example, CSMs are not logging emails with customers as the respective type.
No common unique identifier across various data sources
When we don’t have common reference IDs across different data sources, we typically leverage something called a Fuzzy Match with a 70% accuracy based on company name/common field.
Decluttering/filtering Data
In case of few fields there would be the need to filter the dataset to make it more accurate and declutter unnecessary values. You can provide us with the respective context and we will apply them respectively whilst mapping your data. Examples we have worked with in the past are: filtering on Opportunity table for stage-names, Account Statuses to include only Active/Churned.
Reporting and fixing data errors
If you identify a data discrepancy, contact support@involve.ai for assistance. They will be able to connect your team with a Data Science Consultant to determine the root cause of the error and work with you to address it.
In the coming weeks, my team and I will deliver step-by-step guides about the above discrepancies, including how to identify and fix them without our involvement if that is your preference (and if not, we are always happy to help!).
