- What are the most popular categories on eBay?
- How many bids do items receive on average?
- How many auctions are won in the last minute?
- How did you compile these statistics?
- Where can I learn more?
Note: The research for these questions was performed in the period from 24 November to 2 December 2003
eBay tells you how many items are currently listed in each category, but that's as far as it goes. To get anything interesting out of this data you have to extract the numbers and perform some analysis - which is what we have done. Below is a snapshot of the most popular top-level categories on eBay, in descending order:
|Category||Number of Items||% of Total Items|
|Clothing, Shoes & Accessories||1,752,844||11%|
|Jewelry & Watches||1,253,376||8%|
|Computers & Electronics||1,117,060||7%|
|Toys & Hobbies||1,085,657||7%|
|Pottery & Glass||425,740||3%|
|Dolls & Bears||270,386||2%|
|Business & Industrial||221,857||1%|
It is no surprise that collectibles are still eBay's strongest suit, being the foundation of their business (interestingly, eBay's code for the Collectibles category is still the number one). A bigger surprise is eBay's strength in clothing and entertainment, each making up 11% of total listings.
Lets drill down into the second level of categories - 285 classifications at the time of writing. The top 20 second-level categories with the most items listed, in descending order, are as follows:
|Category||Number of Items||% of Total Items|
|Clothing, Shoes & Accessories: Women's Clothing||645,710||4%|
|Sports: Sporting Goods||503,224||3%|
|Entertainment: DVDs & Movies||466,826||3%|
|Home: Crafting, Sewing, Art Supplies||419,843||3%|
|Sports: Fan Shop||409,804||3%|
|Collectibles: Decorative Collectibles||305,928||2%|
|Home: Home Decor||277,080||2%|
|Everything Else: Health & Beauty||275,016||2%|
|Pottery & Glass: Pottery & China||264,783||2%|
|Collectibles: Postcards & Paper||229,439||1%|
|Entertainment: Video Games||225,739||1%|
|Clothing, Shoes & Accessories: Men's Clothing: Regular||225,647||1%|
|Computers & Electronics: Cell Phones & Plans||216,213||1%|
|Toys & Hobbies: Diecast, Toy Vehicles||212,841||1%|
|Jewelry & Watches: Charms & Charm Bracelets||197,707||1%|
This paints an interesting picture - the high specialization of collectors knocks their categories down the chart, while women's clothing, craft supplies and home decor make appearances, perhaps reflecting that many keen eBayers are full-time homemakers.
We conducted some research into this question using a sample of 3,500 items across the eBay categories, in proportion to the total number of items listed in each category (see above). This represents approximately one item analyzed for every 4,500 currently listed - not a huge number, but a representative sample. This is what we found:
|Number of Bids||Number of Items||% of Total Items|
More than half of the items received no bids at all, and 23% received only one bid. However, this does include fixed-price ("Buy It Now") listings and auctions with the "Buy It Now" option. But it is interesting that the majority of items listed on eBay do not attract any competitive bidding for one reason or another.
To answer this question we took our sample of 3,500 auctions and stripped out items that received zero bids or were won with Buy It Now, which left 1,413 auctions. Of those, 215 were won in the last minute - this table shows when they were won:
|Seconds Won Before End||Number of Items||% of Total Items|
|5 to 10||53||3.8%|
|11 to 20||53||3.8%|
|21 to 30||27||1.9%|
|31 to 45||31||2.2%|
|46 to 60||11||0.8%|
Overall, 15% of auctions were won in the last minute, and 2.9% less than five seconds from the auction end. Note that it does not automatically follow that sniping software was being used - it is quite possible to manually bid in the last seconds of an auction. The picture of bids placed (but not necessarily winning) in the last minute is slightly different and shows the proxy system at work - sometimes earlier bids were higher or equal to the last-minute bid, and so took precedence:
|Seconds Bid Before End||Number of Items||% of Total Items|
|5 to 10||70||5.0%|
|11 to 20||63||4.5%|
|21 to 30||33||2.3%|
|31 to 45||38||2.7%|
|46 to 60||11||0.8%|
The total difference of 50 auctions represents last minute bids that were foiled by the proxy bidding system - in other words one in five last-minute bids did not win the auction (50 out of 265).
The raw data for these statistics was taken from eBay web pages - the same pages you see when navigating the site with your web browser. The results were fed into a database where they could be queried and aggregated.
ASP (Active Server Pages) and VBScript (Visual Basic Script) were used to fetch web pages from eBay, and convert the HTML data into a format that could be analyzed. Specifically, MSXML was used to request the web pages, and regular expressions (using the RegExp object) were used to locate and extract the data required. Details of the process for specific questions are below:
What are the most popular categories on eBay?
A single web page contains all the data needed to answer this question. It was fetched from eBay and regular expressions were used to extract the names of top-level and second-level categories and the number of items listed in each category. The data was entered into a database and queried to answer the question. To see exactly how this was done view the ASP, HTML and SQL code.
How many bids do items receive on average?
This question was more difficult to answer because eBay's completed items database must be consulted to find the final number of bids items receive. A number of factors make this difficult:
- A large number of items, covering every major category, must be analyzed to get a representative sample of auctions.
- A keyword must be entered to search the database - you can't just see everything.
- To match the widest sample of auctions a wildcard search is required (using an asterisk symbol to match any letters), but this requires at least two letters to be entered before the asterisk.
- eBay tries to prevent automated searching of the completed items database.
- The results from the "most popular categories" question were used to calculate how many items should be sampled from each category. Every category with 1% or more of the total items was included, with 35 items analyzed for each percentage point - from 595 items from the Collectibles category to 35 items from Stamps and other 1% categories. This makes a total of 3,500 items.
- Common digraphs were used as search keywords. A digraph is simply a pair of letters - the most frequent at the start of words are "th", "be", "wh", "an", and "no". By searching for as many digraph-wildcard combinations (e.g. "th*") as possible in one search, a very wide sample of items can be obtained. There is a limit to the number of results that can be returned by the completed items search, and we found a maximum of between 3 and 5 digraphs could be used with the "Any of these words" option, depending on the category selected.
- eBay would not return results to computer-generated searches, so we took the approach of searching manually then pasting the underlying code into a "parser" web page which would quickly extract the required data.
The data returned was entered into a database and queried to answer the question. To see exactly how this was done view the ASP, HTML and SQL code.
How many auctions are won in the last minute?
The sample of 3,500 items used in the previous answer were taken as a starting point. The list was refined to include successful full-duration auctions only - first the 1,891 items with zero bids were removed, then auctions that ended with Buy It Now.
The next step was to collect the bid data from eBay using an ASP script which requested the relevant pages one by one, extracted the bid information using regular expressions, and entered the information in a database. Finally database queries and Excel were used to identify the winning bid for each auction, and determine how many seconds before the auction end each bid was placed. For more information view the ASP, HTML and SQL code.