Exploring Search Engine Overlap

Think all search engines provide essentially the same results? Think again. A new comparison tool shows the major search engines have surprisingly little overlap, even for popular search terms.

Search engine guru Greg Notess has long studied search engine overlap — the number of pages found by more than one search engine. Notess’ findings have consistently shown there’s very little overlap in the Web page databases of the major search engines. That means you’ll likely get very different results, depending on the engine.

The new Thumbshots Ranking tool allows you to perform your own comparisons. It displays the results visually, making it easy to see both the rankings and comparative positions of pages in search engine results. Small circles represent URLs. A line connects these circles if the page appears in both engines you’re testing. Mouse over a circle, and the full page URL is displayed.

Results for most queries confirm Notess’ research: There was very little overlap in search engine results for most of the tests I ran. In most cases, even the top 10 results varied significantly from engine to engine.

The ranking tool fetches the top 100 results for a given query from AlltheWeb, AltaVista, Google, MSN, Teoma, WiseNut, or Yahoo You can compare the same search term on two engines or two unique search phrases on the same engine to see which ranks better. You can also enter a specific Web site and see if and where the site appears in the top 100 for a particular query.

Though the comparative results are interesting, they aren’t always exactly identical, as you’ll see when conducting a search directly on a particular engine. According to Stephen Lim, executive director of Thumbshots.org (the folks who created the tool), results may vary due to your locality or the tool hitting different datacenters.

Lim says his team has made a number of interesting observations in using the tool. For example, the majority of sites that rank highly in Google rank poorly in Yahoo, and vice versa. This suggests Webmasters optimizing for Google alone may be missing significant traffic from other search engines.

He also notes despite the fact AlltheWeb and AltaVista now use the Yahoo database, results for the same query from the three engines still differ, though that divergence appears to be narrowing.

Lim offered a number tips for searchers, drawn from experiences using the tool. As links from various search engines don’t overlap significantly, it’s important to test different search engines from time to time, for a variety of reasons.

Search ranking and results vary significantly with plurality. For example, using the ranking tool to compare “business” and “businesses” within Google shows very little correlation between results.

This suggests the ranking tool can actually help searchers improve their queries by running simple variations simultaneously and comparing results.

Ranking also varies significantly when keywords are reversed. Using the tool to compare “allergy treatment” and “treatment allergy” within Yahoo returns very different results.

Search ranking and pages returned as results also change significantly with the addition of a single related keyword. For example, comparing “freeware” and “freeware software” within Teoma yields only a 30 percent overlap. Interestingly, many of the top 10 results are highly correlated.

One of the coolest features of the comparison tool is the addition of “thumbshots” for many results. These are miniature screenshots of a site. Mousing over the circles representing URLs brings up a thumbshot for the page, if it’s available.

Thumbshots Ranking was created by volunteers and is free to use. It’s one of those rare tools that’s equally useful for Webmasters and searchers alike.

This column was adopted from ClickZ’s SearchEngineWatch.com. A longer, more detailed version is available to paid Search Engine Watch members.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.

Related reading

hp
penguin-4-0
bbc
click
<