Data Octopus
Contents
Product identifiers

Identifier consistency across systems

Identifier consistency across systems covers two areas: the identifier level (product or variant) and the identifier type (SKU, id, EAN, MPN and so on).

1. Identifier level: product or variant

Different identification levels across systems make joining data hard, because product sales cannot be assigned to variant sales.

Example: Google Analytics 4 uses item_group_id. We know the Classic T-shirt sold 3 times and generated PLN 450 in revenue, but not which variants sold. Google Ads uses id, so costs are assigned to individual variants:

Google Ads: cost per variant
Variantitem_group_ididSpent
Classic T-shirt, white, size S1a15 zł
Classic T-shirt, white, size M1b2 zł
Classic T-shirt, black, size S1c5 zł

The most precise information sits at the last identification level, the variant. To join GA4 and Google Ads data, both must be reduced to a common identifier, item_group_id, because there's no way to split PLN 450 of revenue across variants. Data quality suffers: we only know the Classic T-shirt (item_group_id: 1) earned PLN 450 and spent PLN 22, not which variant is the most popular.

The same variant level in GA4 and Google Ads
Variantitem_group_ididSpentRevenue
Classic T-shirt, white, size S1a15 zł300 zł
Classic T-shirt, white, size M1b2 zł0 zł
Classic T-shirt, black, size S1c5 zł150 zł

When every system works at the variant level, we know exactly which product is the most popular.

2. Identifier type: SKU, id, EAN, MPN and so on

The second key dimension is the identifier type. Products often carry several identifiers:

  • product card number (often assigned automatically by the shop panel)
  • warehouse number
  • accounting number
  • SKU
  • MPN
  • GTIN

Bad: each system uses a different identifier

  • Google Analytics 4: product card number
  • Google Ads: SKU
  • Meta Ads: warehouse number

Good: one identifier everywhere

  • Google Analytics 4: SKU
  • Google Ads: SKU
  • Meta Ads: SKU

With different identifiers, data from these systems can't be joined simply; extra mappings between systems are needed.

Good practice: give each product one identifier and use it in every system: the shop panel, accounting and warehouse systems, marketplaces and ad platforms.