Factor 5 — Trust/review moats (conversion, not just ranking)

Claim: Reviews are the most-evidenced ranking AND conversion lever — and uniquely durable because they are rate-limited by actual customer transactions.

Causal evidence (the strongest in the entire ranking/conversion domain):

• Luca HBS Yelp study — 1-star Yelp increase CAUSES 5-9% revenue increase, driven by INDEPENDENT (non-chain) restaurants. See Luca HBS WP 12-016 — one-star Yelp = 5-9% revenue (regression-discontinuity, the causal anchor).
• Northwestern Spiegel 2017 — 5 reviews → 270% purchase-likelihood lift; 380% for higher-priced products. See Spiegel 2017 (Northwestern) — 5 reviews = 270% purchase likelihood lift; 380% for higher-priced.
• The "too good to be true" effect — purchase likelihood peaks 4.0-4.7 stars and DECLINES toward 5.0. See Spiegel — purchase likelihood PEAKS at 4.0-4.7 stars and DECLINES toward 5.0 (too-good-to-be-true).

Why durable:

• Cannot legitimately be bought. FTC fake-review rule + Google policies prohibit incentivized/fake reviews; manipulation increasingly detectable.
• Rate-limited by actual customer volume. A 500-review incumbent with 4.5 stars + steady inflow is one of the hardest positions for an SMB to attack — no shortcut closes it.

Consumer reliance (vendor, directional): Just 4% of consumers say they "never" read online reviews (BrightLocal). Trust in reviews "as much as personal recommendations" fell 79% (2020) → 42% (2025) — see [[consumer-rating-thresholds-rising]] topic.

Why this matters for Candid: Of all the levers an SMB can pull, sustained review velocity has the most causal evidence behind it. Should be among the strongest factor weights in the widget tier model — both as a difficulty signal (competitor reviews) and an action item (your own review velocity).