BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//pydata-amsterdam2026//talk//Q7CB9J
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-pydata-amsterdam2026-Q7CB9J@pretalx.com
DTSTART;TZID=CET:20260911T110500
DTEND;TZID=CET:20260911T113500
DESCRIPTION:Across **digital platforms**\, the most powerful features often
  require **voluntary user adoption**. Whether it is a customer claiming a 
 discount code\, a user opting into a new loyalty program\, an e-commerce s
 eller adopting a smart-pricing tool\, or a vacation host enabling an 'inst
 ant booking' feature\, the opt-in barrier is a ubiquitous industry challen
 ge across both demand and supply.\n\nFor Product Data Science\, measuring 
 these tools introduces a severe "Trilemma": **Voluntary Adoption**\, **Ext
 reme Heterogeneity**\, and **Finite Sample Sizes**. Standard A/B tests can
 not overcome this\; they severely dilute the feature's true impact across 
 everyone who failed to opt in. Consequently\, businesses risk abandoning h
 ighly lucrative product innovations simply because a high-friction opt-in 
 process or poor discoverability led to low adoption—**masking the featur
 e's true value**.\n\nThrough a code-driven simulation\, this talk demonstr
 ates why standard approaches fail. We will visualize how observational met
 hods—such as **Propensity Score Matching** or **naive ML**—fall into t
 he trap of **collider bias** by comparing intrinsically motivated adopters
  to unmotivated non-adopters\, yielding structurally invalid estimates.\n\
 nTo solve this Trilemma\, we introduce a rigorous causal framework. First\
 , we use **Randomized Encouragement Design (RED)** to bypass collider bias
  and create a valid instrument. Then\, we walk through the Python implemen
 tation of Double Machine Learning showing how to utilize the Interactive I
 nstrumental Variable Model (`DoubleMLIIVM`) to handle extreme user heterog
 eneity and highly skewed platform data using tailored models like Tweedie 
 regressors.\n\nAnchored by a real-world case study\, attendees will leave 
 with a practical blueprint for extracting precise causal estimates. Ultima
 tely\, you will learn how to move beyond diluted "average" metrics to **un
 cover the hidden value of voluntary features**—empowering your organizat
 ion to stop prematurely killing high-potential products and start isolatin
 g true product value from distribution bottlenecks.
DTSTAMP:20260710T141221Z
LOCATION:Main stage
SUMMARY:The A/B Testing Blind Spot: Solving the Opt-In Paradox with Randomi
 zed Encouragement and DoubleML - Lin Jia\, Kexin Fei
URL:https://pretalx.com/pydata-amsterdam2026/talk/Q7CB9J/
END:VEVENT
END:VCALENDAR