The Need for People Based Measurement
summary
A billion-plus datapoint study across three companies shows cookie-based measurement fragments identities and undercounts interactions: merging identities yields 1.8x longer paths, mobile/browser/app touchpoints consolidate so mobile credit rises from ~29% to ~52%, and Facebook/Instagram credit rises from ~7% to ~17% (a ~240% uplift for that channel). The work uses media mix modelling, random forests and Hidden Markov Models, plus deterministic and probabilistic identity resolution via a data exchange and big‑query stores to tie IDs together and reattribute credit.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:03How important is peoplebased measurement
- 2:40Whats happening in India
- 3:05How people spend their time
- 3:43Cookies in iOS 11
- 5:16The study
- 6:57Paths
- 7:41Cookies
- 10:32People based attribution
- 11:52Facebook example
- 13:33Hidden Markov models
- 14:50Summary
- 15:30The Purchase Path
- 17:34Identity Resolution Mapping
- 20:00Data Exchange
- 21:49What is the deal
- 23:37How is it being approached
- 24:52How much of this is a legal or privacy policy concern
related talks
23mKeynote: Digital Psychometrics and it's Future Effects on Technology
15mThe Secret Life of Data - Part 2
39mBuilding Causal Measurement
19mSocial Media - Did We Reach the Summit? pt.1
29mUser Motivation: Refining Customer Segments with Location
26m"Reality Mining for Compan
16mThe Secret Life of Data - Part 3
11mData science: Next-gen performance analytics