Every product page on ProductDome carries a scorecard — owner satisfaction, value, demand — and sometimes it disagrees with the raw Amazon stars sitting right next to it. That disagreement is the point. Here’s the math, in plain terms.

The problem with raw star averages

A product with five reviews averaging 4.8 looks better than one with four thousand reviews averaging 4.5. It usually isn’t. Small samples are noisy and easy to seed with friendly reviews; large samples have survived years of real buyers, defective units, shipping damage, and expectation mismatches. Raw averages treat both as equally trustworthy. We don’t.

Bayesian smoothing, in one paragraph

Our owner-satisfaction score starts every product at its category’s average rating and lets its own reviews pull it away from that average as evidence accumulates. Five glowing reviews barely move it; five hundred move it a lot. Statisticians call this Bayesian smoothing — practically, it means a 4.8 with a dozen reviews scores below a 4.5 with thousands, because we’re much more certain the 4.5 is real. As reviews accumulate, the score converges toward the true average.

Percentiles, not absolutes

A 4.3-star rating means something different in a category averaging 4.6 than in one averaging 3.9. So the scorecard grades within the category: cheaper than X% of tracked models, better-rated than Y%, more-reviewed than Z%. Those percentile chips on product pages are computed against every model we track in that category — not a marketing adjective. We only show a scorecard when the category has enough tracked products for percentiles to mean anything (currently a minimum of eight).

The three axes

  • Owner satisfaction — the smoothed rating above, weighted by review volume.
  • Value — price position versus the category, adjusted for how the product rates. Cheap and well-rated scores high; expensive and average scores low. Premium products can still score well when their rating percentile keeps pace with their price percentile.
  • Demand — where the product’s review accumulation and “bought last month” figures sit versus the category. Demand isn’t quality, but a product nobody buys can’t sustain a reliable rating, and dead sellers eventually become unsupported products.

What the scorecard is not

It’s not a lab test — we say this on every page. We don’t own the products; we track their data at scale: specs, prices, ratings, review velocity, brand footprint. AI-written summaries on our pages distill owner feedback and are labelled as such; no review text is copied. And the one thing we never do is put Amazon’s star rating into search-engine schema as if it were our own verdict — visible on the page, labelled as Amazon’s, kept out of our structured claims.

Why this makes rankings feel “wrong” sometimes

If you sort our category pages by rating, a shiny 4.9 sits below a workhorse 4.5 now and then. That’s the smoothing doing its job. The 4.9 might be genuinely excellent — and if it is, its score will rise as its review base grows. Until then, we’d rather under-rank a promising newcomer than over-rank a padded one.