Dataset Overview

Complete statistical portrait of the Home & Kitchen review data, powered by Amazon Reviews'23

67409944
Total Reviews
4.18
Avg Rating
93.3%
Verified Purchase
8.4%
With Images

Dataset Overview

The Home & Kitchen category is the second-largest in the Amazon Reviews'23 dataset, containing 67,409,944 reviews spanning over 15 years (1996–2023). This page presents a comprehensive statistical portrait of the data powering this site.

Data Source: Amazon Reviews'23 by McAuley Lab, UC San Diego. Raw file: Home_and_Kitchen.jsonl


Rating Distribution

Rating distribution chart

Figure: Distribution of star ratings across all Home & Kitchen reviews.

Amazon reviews skew heavily positive — this is a well-known platform bias.

Rating Count Percentage
5 ★ 44,701,824 66.3%
4 ★ 7,536,382 11.2%
3 ★ 4,600,479 6.8%
2 ★ 3,410,803 5.1%
1 ★ 7,160,456 10.6%

Average Rating: 4.18 / 5.00

Why This Matters

Over 83% of reviews are 4–5 stars. This means:

  • Raw star ratings are a weak signal for product comparison — most products cluster near 4.5
  • Negative reviews (1–2 stars) are disproportionately valuable for identifying product defects
  • Review text is more informative than ratings for making purchase decisions

Review Authenticity

Metric Count Percentage
Verified Purchases 62,921,814 93.3%
Unverified 4,488,130 6.7%

About 6.7% of reviews come from unverified sources (Vine reviewers, review swaps, direct reviews without purchase). The verified subset is generally more trustworthy.


Review Helpfulness

Amazon users can upvote reviews as “helpful.” This signal helps surface the most informative content.

Helpfulness distribution chart

Figure: Log-scale distribution of helpful votes. Note the extreme skew — most reviews receive 0 votes.

Helpful Votes Review Count Percentage
0 votes 51,843,462 76.9%
≥ 1 vote 15,566,473 23.1%
≥ 5 votes 2,446,886 3.6%
≥ 10 votes 1,085,608 1.6%
≥ 50 votes 144,113 0.2%

Average helpful votes per review: 1.02 Most helpful review: 34619 votes

“What can I say about the 571B Banana Slicer that hasn’t already been said about the wheel, penicillin, or the iPhone…. this is one of the greatest inventions of all time. My husband and I would argu…”

Insight

Only 23.1% of reviews receive any helpful vote. This means:

  • Reviews with 5+ helpful votes are rare signals of genuine insight
  • About 144,113 reviews (0.3%) carry 50+ votes — these are gold for content curation

Review Length Distribution

Review length distribution chart

Figure: Review length at key percentiles. The dashed red line marks the average.

Percentile Characters
25th 47
50th (median) 106
75th 216
90th 398

Average length: 179 characters

Half of all reviews are under 106 characters — roughly 1–2 sentences. Only 10% exceed 398 characters. Long, detailed reviews are scarce and valuable.


User-Submitted Images

Metric Count Percentage
Reviews with photos 5,636,288 8.4%
Text-only reviews 61,773,656 91.6%

Only 8.4% of reviews include user photos — but these are the most trusted by shoppers.


Reviews Over Time

Yearly review trends chart

Figure: Review volume (bars) and average rating (line) per year.

Year Reviews 5★ % Verified %
1998 2 100.0% 0.0%
1999 9 33.3% 0.0%
2000 1,001 64.0% 13.3%
2001 2,209 59.5% 20.6%
2002 3,695 55.8% 24.1%
2003 5,938 52.6% 26.2%
2004 7,353 49.8% 25.5%
2005 15,419 49.1% 30.2%
2006 24,972 52.8% 36.7%
2007 58,322 57.8% 45.9%
2008 68,207 56.4% 48.1%
2009 92,012 54.9% 60.1%
2010 159,296 55.2% 78.9%
2011 278,672 55.8% 81.5%
2012 500,445 58.4% 85.6%
2013 1,328,715 61.0% 92.8%
2014 2,222,935 63.6% 89.0%
2015 3,446,051 65.5% 94.1%
2016 4,493,405 66.7% 92.3%
2017 5,059,065 66.6% 93.2%
2018 5,833,876 67.3% 95.3%
2019 8,229,519 71.2% 96.3%
2020 10,540,330 68.6% 96.2%
2021 11,436,430 65.7% 95.6%
2022 9,601,338 63.1% 90.3%
2023 4,000,728 64.1% 86.8%

Statistics computed from the raw dataset. Based on a full scan of all available reviews.

Data source: Amazon Reviews'23 by McAuley Lab, UC San Diego. Full dataset contains ~67.4M reviews.