Rating drift: the honeymoon ends — but not at the same rate

What we measured

For products with at least four years of reviews, we compared the average rating in the early reviews with the average in the late reviews. The delta is the early-to-late change in stars. It is a measure of how owners’ ratings move over time, not a measured failure time.

The numbers

Across 838 ceramic products, the mean delta is -0.26 stars. For stainless steel, the delta is -0.31 stars; for porcelain, -0.10 stars (n=3,871 vs 195; the stainless drop is 3.1× the porcelain drop).

The direction is the same for every material: ratings drift down. The useful question is how far and how fast. A bigger drop means owners become more disappointed over time — a signal about coating and build quality — but it is not a stopwatch on the product’s life.

Why the comparison is not a material ranking

The porcelain sample is 195 products, against 3,871 for stainless steel. A small sample is more easily moved by a few products, and the product mix differs: a stainless steel category includes different pan types and price points than a porcelain category. The 3.1× ratio is worth noting; it is not proof that stainless steel is worse than porcelain in general.

There is also a survivorship effect. A product only enters a four-year comparison if it has four years of reviews. Products that failed early and stopped selling are not in the late group. The drift we measure is the drift among products that survived long enough to be measured.

What it means for a buyer

• Read the drift with the absolute rating. A product that goes from 4.8 to 4.5 is still well rated; a product that goes from 4.2 to 3.9 is a different story.
• A large negative delta is a warning about expectations, not a failure date. It says owners became less happy, not that the pan broke on a schedule.
• Do not rank materials by one delta. Use the material comparison pages, which state the sample and the comparison group.
• For a specific product, the sample size matters more than the category average. A 0.26-star category drift tells you nothing about a pan with 12 reviews.

What this does not tell you

The delta is computed from review ratings, and reviewers are self-selected. The early and late windows are different lengths for different products, and the product mix changes over time. A rating drop can reflect a change in the buyer population, a change in the product, or a change in expectations. The corpus is a 2023 snapshot; the comparison is an association, not a controlled experiment.