Detect coordination, not souls
A bot detector becomes unreliable when it pretends one weird sentence proves an account is fake. LessBother combines independent signals and reports uncertainty.
REVIEW TRUST · LOCAL ANALYSIS LAB
LessBother looks for manipulation patterns—bursts, duplicated language, reviewer reuse and suspicious rating shapes—then shows the evidence. It does not stamp individual people “bot” from a hunch.
Pasted review datasets never leave this page. If the review-data box is empty, enter up to three product URLs. LessBother makes a separate bounded server-side fetch of each public page and combines only product-scoped structured review records locally. Comparing listings can expose review copy that travels between products or storefronts.
[
{"author":"buyer17", "rating":2,
"timestamp":"2026-08-12T14:30:00Z",
"text":"Zipper split after three uses."}
]Tab-separated exports are also accepted. If you paste full product-page source, LessBother reads only product-scoped schema.org Review data and refuses ambiguous multi-product structured data.
A bot detector becomes unreliable when it pretends one weird sentence proves an account is fake. LessBother combines independent signals and reports uncertainty.
A trust label always comes with the underlying counts and patterns. No unexplained “82% authentic” magic number.
The same Human Trust Engine also has a working Conversation Authenticity lab for Reddit-style reply clusters and manufactured consensus.