---
id: "entity-inertia-field-experiment"
type: "entity"
source_timestamps: ["¶6", "¶8"]
tags: ["research-paper", "field-experiment", "primary-evidence"]
related: ["entity-klaus-miller", "entity-z-john-zhang", "claim-auto-renew-reduces-takeup", "claim-consumers-aware-of-inertia"]
entityType: "other"
canonicalName: "Sophisticated Consumers with Inertia: Long-Term Implications from a Large-Scale Field Experiment"
aliases: ["the newspaper experiment", "the 1.4M-person field experiment"]
url: "https://bfi.uchicago.edu/working-paper/sophisticated-consumers-with-inertia-long-term-implications-from-a-large-scale-field-experiment/"
sources: ["commercial"]
sourceVaultSlug: "hbr-seg-commercial"
originDay: 5
articleStem: "hbr-tier2-08-subscription-auto-renew"
sourceUrl: "https://hbr.org/2026/05/should-your-subscription-business-use-auto-renew"
sourceTitle: "Should Your Subscription Business Use Auto-Renew?"
---
# Sophisticated Consumers with Inertia (field experiment)

The **foundational research paper** behind virtually every empirical claim in this source: a large-scale randomized field experiment run with a leading European newspaper, involving **~1.4 million people**, authored by [[entity-klaus-m-miller]] and [[entity-z-john-zhang]].

**Why it matters:** It is the primary evidence for [[claim-auto-renew-reduces-takeup]], [[claim-auto-cancel-yields-more-subs]], [[claim-consumers-aware-of-inertia]], and [[claim-auto-renew-degrades-quality]]. Its structural model produces the [[framework-consumer-inertia-typology]].

**Key published findings (from enrichment):**
- Strong inertia among takers (a 53–75% monthly chance of not taking a desired cancellation action for auto-renew takers).
- **24–36% of potential subscribers avoid auto-renewal offers** (basis for the article's stylized '35%').
- Auto-renew takers have ~**7× higher** tendency to continue after the promo than auto-cancel takers (real retention), yet auto-renewal **decreases take-up in the short and long run**.
- Auto-renewal and auto-cancel become **revenue-equivalent after ~one year**, with auto-renewal leaving *fewer* subscribers.
- Structural estimates: 35–55% non-inert; of the inert, **83–92% sophisticated** (naïveté rare).

**Caveat:** Single-firm, likely-inertial (news) market — a key limit on generalizing the '23% more subscribers' headline.

**Canonical host:** Becker Friedman Institute working paper (URL in frontmatter).