©Inkspot Antiques and Collectibles
To score the merchant and search page sites, we evaluated both generic searches such as Civil War buttons and specific searches such as "Civil War" Cavalry button. For the counts or hits for a specific items listed, we generally relied on the results page, but this number is not valid for Google Shopping as explained below.
It seems most search engines default to fuzzy logic in their search results. For most people conducting general web searches, this can be advantageous; however, if you are looking for a specific product, this default can be annoying, because the search may be providing you pages of irrelevant information. On ebay, if you do a title search, except for certain misspellings, and certain categories, you will generally get listings that exactly match what you entered in the search bar. Thus if you entered the words American Flyer keystone boxcar, in search of that train set's specific car, you will (generally) only get listings with those four words. When we did our search on ebay, we got exactly one listing for that search. However, when we did the same search on Bing Shopping, not only did we get a keystone boxcar listing (at a non ebay site), but we got more than 4,000 other listings including American Flyer Luggage sets (a existing company that sells luggage), as well as every other American Flyer boxcar and train sets imaginable. The one AF keystone boxcar listed on Bing Shopping was fortunately on the first page, but it was like finding a needle in a haystack. If there were any other listings on subsequent Bing Shopping results pages, we would miss them altogether.
To get around this problem, we used the current default search engine logic of putting "must include words" in quotation marks. This means that by entering even single words in quotation marks you get different results. We were surprised that this technique even works with ebay. In other words, a search for "Trifarri" "Jelly" "Belly" is not the same as a search for Trifarri Jelly Belly (these were lucite gems used in jewelry brooches and earrings. Some search engines have quirks, if not outright flaws on how they handle certain compound words or apparent misspellings. For example, on Google Shopping, a search for Hot Wheels Redline will get you 9,800 items, however a search for Hot Wheels "Redline" will only get you exactly 1 listing, despite the fact that "Redline" (one word) is the correct usage, and that is how the vast majority of products are listed. This is a classic example of a search engine presuming that you really don't know what you area searching for. At least ebay, as it evolved in handling search, finally allowed completed searches for "apparent" misspellings if they yielded results, with the additional tease "Did you mean to search for..." Bing, The Finding and Google Shopping could learn some lessons from ebay in product search logic, and how not to frustrate potential buyers.
As noted below, in some cases, we just visually rejected certain listings if they did not result in a product that matched our search. Such products might appear at such sites if the search found the word in the description of in the keywords submitted with the product.
In the table below are the specific items we searched for, and the number of results we found at each site. For each search the sites were ranked, with ties having equal score. The average score of all searches was calculated, and these average scores were then ranked to calculate the final ranking of each site. This table is representative and does not show all searches evaluated.
As noted above, fuzzy search had to be tightened using the tips used above, and the counts show only what were deemed to be valid matches. In most cases, we forced a strict search by requiring all terms to be present (called "no relax" in some search engines). This is achieved by placing words in quotes. In some cases, we had to modify the search at some sites with NOT exclusion of terms. Even then, counts were adjusted by visually inspecting the results and subtracting out those that did not match the search. In some sites, this occurred because of hidden keywords associated with the listing, or certain text for in the copy add (some sites search only titles, others search descriptions as well. In some sites, particularly The Find, Bing Shopping, and Google Shopping, the search engine defies control because of apparent word substitutions or ignored words in the search engine logic. The worst example of this was the search for antique type trays. These are standard fodder at antique shops and are the trays used at newspapers and magazines to hold lead type during print setting days. These are now hung on walls as curio displays. In any case, ebay lists precisely those objects in a search. However, the Find, Bing Shopping, and Google Shopping apparently ignored the word type and showed all manner of antique trays. Fuzzy search is a useful tool when it comes to looking for information or ideas on line, but the search engines have not caught on to the fact that when you are looking for a specific item, it is a hindrance.
Another peculiarity with respect to Google Shopping is the first page shows an in correct count of items found. For example, in the case of the search "Waterford" "crystal" "champagne" "flute", Google says it found more than 1,300 items. However if you go to page 2 or subsequent pages, it indicates only 227 items were found. The 1,300 items is consistent with Bing and The Find results, and that number probably relates to the number of merchant feeds it received, but only 227 paid Google to advertise since they made the transition to paid only listings on Google Shopping. Please note that Google organic search was inserted as number 2 on our search ranking, even though it could not be ranked in the same way as other other sites.
|ebay||google shopping||bing shopping||Replacements||Etsy||The Find||Ruby Lane||Go Antiques||Bonanza||TIAS|
|"Hot" "Wheels" "Redline"||4,542||2,200||321||0||42||1,386||2||2||45||1|
|Hot Wheels Redline Dump Truck||23||2||4||0||0||4||2||0||2||0|
|1891 ms63 Morgan dollar||86||6||1||0||0||9||0||1||0||0|
|Lionel "Train" "union pacific"||138||318||62||0||0||298||1||3||11||0|
|American Flyer Keystone Boxcar||1||0||1||0||0||2||0||0||0||0|
|Akro Agate Popeye Marble||30||6||3||0||3||7||0||0||0||0|
|antique federal andirons||32||0||3||0||2||6||0||0||0||0|
|"glass" "flask" "pontil"||8||6||7||0||5||19||16||3||0||4|
|"flexible" "flyer" "airline"||32||4||1||0||2||2||0||1||0||0|
|"Cambridge" "glass" "flower" "frog"||43||27||10||0||13||33||17||7||6||5|
|"Tiara" "Sandwich" "Glass" "Plate"||103||47||20||0||28||43||8||8||10||12|
|"aqua" "ink" "bottle"||36||33||3||0||64||45||2||0||11||3|
|Arcade "Coffee" "Grinder"||76||8||8||0||4||15||8||5||2||4|
|Belleek "shamrock" "cup" "saucer"||28||28||33||33||4||68||9||5||3||2|
|"yellow" "fiesta" shaker"||10||23||7||20||12||30||5||2||0||3|
|"Stickley" "mission" "Chair"||56||30||6||0||37||53||5||0||1||1|
|"Fenton" "cake" "plate"||75||67||8||20||79||70||6||1||5||4|
|"1942-D" "Walking Liberty"||191||14||8||0||0||135||0||0||2||0|
|"figural" "meerschaum" "pipe"||48||4||6||0||9||27||10||0||3||1|
|"antique" "type" "tray"||0||0||0||0||0||0||0||0||0||0|
|"Schrade" "pocket" "knife"||1577||71||1060||0||32||2540||2||2||9||0|
|"Waterford" "crystal" "champagne" "flute"||227||69||862||611||3||1778||9||5||3||2|
|"Journey Into Mystery" "Thor"||1219||73||55||0||1||49||1||36||9||0|