LookCloser

Review forensics

Do the reviews move like demand, or like intervention?

Authentic reviews accumulate the way sales do. Manipulated ones move differently — and the pattern is often visible in the data.

LookCloser · methodology note · what this check is and how to read it

A star rating is the first thing most shoppers look at and the easiest thing to manufacture. But reviews leave a trail over time, and authentic reviews behave differently from purchased or incentivized ones. The rating isn't the signal — the shape of how it was built is.

Genuine reviews track sales. Bursts that don't track sales, reviews older than the listing, and merged histories are the fingerprints of intervention.

Patterns worth watching

How LookCloser builds this over time

Most of these patterns are only visible historically — you need to compare the same product across dates. LookCloser records a dated observation of every product's rating, review count, and price on each check, building its own longitudinal history. A review count that falls between two observations, or a price that swings before a "discount," becomes a documented change rather than a snapshot.

What this check does not claim

  • A high rating is not evidence of quality; a fast-growing one is not proof of fraud.
  • Patterns are recorded as observations with dates — the reader weighs them.
  • Reviewer-level analysis (verified-purchase clusters, reviewer overlap) needs data sources we name honestly when we don't yet have them.

See this check on a real product.

Every LookCloser record runs this check with named, dated public sources — and shows the gaps too.

Open a record →