Office Products: what 12.8M reviews show
The largest category in this research desk. 18.1 reviews per product makes category-level statements more defensible here than in Fashion โ but still not a product rating.
By AMZ Reviews Data ยท October 19, 2026 ยท Data vintage 1996โ2023
What the data shows
Office Products is the largest of the three categories in this research desk: 12,841,907 reviews across 709,449 products, or 18.1 reviews per product. The review-weighted average is 4.21 (ฯ 1.35). 5โ reviews are 67.8% of the sample and 1โ reviews 10.5%. Verified-purchase reviews account for 93.4%, the second-highest of the three categories.
Review volume peaked in 2020 at 1,891,745 and falls to 674,847 in 2023. That is a coverage artifact, not a collapse in demand: the dataset stops in September 2023, so the final year is incomplete. Every yearly series on this site carries the same caveat.
The early/late rating gap (4.32 in 1998โ2000 vs 4.11 in 2021โ2023) is real in the sample, but it should not be read as a quality trend. The early years contain a few thousand reviews, and the mix of products sold on Amazon changed completely over 25 years. We publish the gap because it is in the data; we do not claim it means products got worse.
What it does not tell you
The average is review-weighted, not an average of product ratings. A product with 10,000 reviews and a product with 30 reviews do not contribute equally. That is the right weight for “what do buyers say”, but it is the wrong weight for “how does the typical product perform” โ and the two questions get conflated constantly in category statistics.
The data also cannot separate a product defect from a shipping problem, a compatibility mistake or a review written before the product was used. Category-level counts are a map of where to look, not a verdict on any specific item.
Review volume & average rating by year
Counts and averages for this Office Products sample. Bars are review volume.
| Year | Reviews | Avg rating | Volume |
|---|---|---|---|
| 1998 | 3 | 5 | |
| 1999 | 115 | 4.04 | |
| 2000 | 1245 | 3.93 | |
| 2001 | 1832 | 3.72 | |
| 2002 | 2269 | 3.69 | |
| 2003 | 2703 | 3.62 | |
| 2004 | 3344 | 3.42 | |
| 2005 | 5409 | 3.51 | |
| 2006 | 7314 | 3.59 | |
| 2007 | 15799 | 3.9 | |
| 2008 | 19125 | 3.86 | |
| 2009 | 27503 | 3.85 | |
| 2010 | 45134 | 3.9 | |
| 2011 | 74646 | 3.89 | |
| 2012 | 122829 | 3.96 | |
| 2013 | 304640 | 4.13 | |
| 2014 | 516904 | 4.21 | |
| 2015 | 808738 | 4.26 | |
| 2016 | 1014588 | 4.3 | |
| 2017 | 1061237 | 4.26 | |
| 2018 | 1207700 | 4.24 | |
| 2019 | 1618521 | 4.33 | |
| 2020 | 1891745 | 4.25 | |
| 2021 | 1866756 | 4.15 | |
| 2022 | 1546961 | 4.07 | |
| 2023 | 674847 | 4.12 |
Rating distribution
How the 12,841,907 reviews break down by star rating.
| Rating | Reviews | Share | Distribution |
|---|---|---|---|
| 1โ | 1353052 | 10.54% | |
| 2โ | 588987 | 4.59% | |
| 3โ | 799026 | 6.22% | |
| 4โ | 1400281 | 10.90% | |
| 5โ | 8700561 | 67.75% | |
| Total | 12841907 | 100% | ฯ = 1.352 |
Method & limits
These figures are review-weighted aggregates from the UCSD Amazon Reviews 2023 dataset; they are not a survey and not a product ranking. The average rating is computed over reviews, not over products, so a product with thousands of reviews carries more weight than a product with three. The dataset ends in September 2023, which makes the final year incomplete โ read the last bar as a partial year, not as a decline.
Definitions and thresholds are on the methodology page; scope and known limitations are on data sources & limitations.
Sample size. This page is built from 12,841,907 reviews collected between 1998 and 2023. Every figure on this page is a direct count or share of those reviews โ nothing is modelled or inferred.
Data vintage. Source: UCSD Amazon Reviews 2023 (data through September 2023). Historical statistics are not a statement about a product's current quality, availability or price. Review volume is not sales volume.
Scope & deduplication. One review per reviewer per product is counted; ratings without review text still count toward distributions. Coverage is uneven across years and categories, which is why the sample size is shown with every table. See the methodology for the exact definitions.
No raw reviews. We publish aggregated counts and word frequencies only โ never full review text, reviewer identities or images.
Affiliate disclosure. Outbound links to Amazon are Amazon Associates links; if you buy through them we may earn a commission at no extra cost to you. Links point to a product page or to an Amazon search for the brand / store / category. We never display live prices or stock because those change constantly.