QUARANTINE — "53% of mobile users abandon a page that takes >3 seconds to load" traces to Google "Need for Mobile Speed" 2016–2017; the data was about MOBILE AD LANDING PAGES, not general business websites
Quarantined claim: "53% of mobile users abandon a page that takes >3 seconds to load." (Related: "bounce probability +32% (1–3s), +90% (1–5s).")
Provenance trail: Traces to Google / "The Need for Mobile Speed" / Think with Google (Daniel An), ~2016–2017, built on Google / SOASTA data and a deep-neural-net model trained on mobile ad landing pages. The figure is genuine and primary but ~9–10 years old and not about general business-website visitors. The related "bounce probability +32% (1–3s), +90% (1–5s)" is from the same Google / SOASTA 2017 work.
Reason for exclusion:
- Cannot be re-confirmed as a current dataset.
- Context mismatch — mobile ad landing pages served via Google Ads, not general SMB websites.
- ~9–10 years stale; mobile-web speed has changed materially.
Source: thinkwithgoogle.com (Daniel An, ~2017); Google / SOASTA bounce-probability paper.
Confidence: Verified (the provenance trail; not generalisability of the claim to current SMB websites).
Why this matters for Candid: Do not cite as evidence for general business-website speed thresholds. If page-speed evidence is needed, cite Core Web Vitals documentation and current real-user data, not the 2017 mobile-ads number.
Referenced by (2)
- reference Research brief: the website as a working surface of the business — four capabilities, AI-citation decoupling, freshness as a real signal (June 2026) relates-to
- reference Caveats for the working-surface brief: independent anchors (Pew, peer-reviewed GEO paper, Ahrefs large-N) carry the load; vendor figures are range / corroboration, not independent confirmation relates-to