Challenges of POS Data
POS data doesn't have to be hard to validate.
This can be accomplished via two primary mechanisms:
Validating against a known independent source In this case, retailer POS can be compared to a known third party source
Validating against itself
In this case, data in POS files is evaluated for anomalies
This type of validation is most useful once sufficient POS history for a retailer has been accumulated
This type of validation is most useful in the initial stages of POS usage
Market
Product Items must continuously be harmonized to generate accurate insights at the UPC, Brand, or Category level. The same item in two retailers may be coded differently due to use of Shelf Keeping Units (SKUs) vs. Unique Product Codes (UPCs) / European Article Numbers (EANs), over-stickering, errors at checkout, etc. Period Cutoffs for data processing vary. Differences must be defined to ensure correct interpretations of data. Retailers' Promo Weeks may vary. The period dimension doesn't change frequently, and can be aligned upon initial ingestion. Even if retailers provide information about dayparts, sufficient granularity usually exists for foundational alignments.
Market dimensions are constantly changing. Store-level info provided by a retailer must be aggregated for analyses.
This could be done by:
Creating store clusters around brand definition (e.g. consumer segmentation) Aligning it with definitions (e.g. zip code for Designated Market Area)
Fact
Filed retailer POS facts can differ across retailers. Beyond transaction data, some may include:
Data about the consumer Data about media publishing Data relating to the supply chain
Defined assessment across retailers is needed.
Page 5
© 2023 Fractal Analytics Inc. All rights reserved
Made with FlippingBook - PDF hosting