Skip to main content
If you’ve ever tried to build a unified analytics dashboard, you know the pain.
  • TikTok calls it video_views.
  • Instagram calls it impressions.
  • LinkedIn distinguishes between uniqueImpressions and shareMentions.
  • YouTube has views but also redViews (YouTube Premium).

​
The Engineering Problem

Data Science isn’t dead, but for social media, it’s mostly a Data Engineering problem. You need to:
  1. Fetch data from 10 different APIs with different rate limits.
  2. Map disparate fields to a common schema.
  3. Handle zeroes (Does 0 mean zero views, or “data not available”?).
  4. Store it in a way that allows for time-series comparison.

​
Our Approach

At bundle.social, we spent months building a normalization layer. We decided that Impressions means “The number of times content was displayed,” regardless of what the platform calls it. We do the dirty work of mapping facebook_page_impressions_unique_28_days to impressionsUnique so you don’t have to.