Is Site Traffic a Ranking Factor?

[Altair]

A web report discussing a "site popularity factor" claims that Google and the other major search engines use site popularity (total site human traffic) as a factor in computing rank. Site popularity is described as the web traffic that a domain name receives as measured by Nielsen//Netratings, Alexa, the Google toolbar, or similar measuring scheme. Site traffic does seem to be an obvious choice for including in a ranking algorithm.

If this is true we would expect the following: If a site was banned by one search engine or otherwise had a major reduction in traffic (maybe loss of a major incoming link), site popularity would decline. In following months, traffic coming from other search engines would also decline because of the site popularity factor, reducing traffic more, causing search engines to reduce rank more, and so forth until traffic stabilized at a new lower level.

We have seen some behavior like this. Has anyone else noticed this kind of effect? It may be that it is only significant for more popular sites, say the top 150,000.

A site popularity factor is one explanation for the "Google sandbox effect." If a banned site was completely reincluded, it would still take some time for traffic to rebuild and therefore rank would be lower.

[Frederic1]

I don't think site traffic is a ranking factor. At least I don't hope so, because a lot of sites in my sector trade trafic through popups/adverts.

[sandpetra]

I'm thinking the colours of socks I wear one day is a ranking factor which google can turn off and on whenever it feels like it!

Everything is a ranking factor. We just don't know from day to day / month to month which set of factors is in play.

[tedster]

There can be many types of traffic data. For example, the fact that your site is in a SERP but a large percent of users immediately return to the SERP for a different choice right after trying your link. That kind of traffic data has long been a factor.

Check out this recent patent:
Document Scoring Based on Traffic Associated with a Document [appft1.uspto.gov]

[mattg3]

That kind of traffic data has long been a factor.

The worrying thing about this is that increasing the speed of the server actually would decrease your ranking unless it's factored in.

[Altair]

Tedster: I agree, tracking data on individual users such as backout data could be used as an indication of page quality. But some of the data suggests that raw traffic count, which is actually much easier to do, is also being used as an indicator of site quality. This has some implications. Somebody could add a page about Britney Spears on a site about accounting, therefore increase site traffic and thereby outrank some other accounting site. I suspect Google uses both methods and the others probably depend more on raw traffic because they have less tracking data.

Mattg3: I think having a faster site would generally reduce backouts.

[dataguy]

That kind of traffic data has long been a factor.

The worrying thing about this is that increasing the speed of the server actually would decrease your ranking unless it's factored in.

Funny you should say that.

At Pubcon New Orleans I asked one of the search engineers if the user data gleaned from the toolbar was used in ranking. His response was that it currently wasn't being used directly because of pre-existing patents, but it was being used indirectly.

He then told me that the best advice he could give me was to invest in faster servers and more bandwidth. My impression was that Google was about to put more emphasis on user click trails.

I've always thought that it was peculiar that Google is light-years ahead in their crawling and ranking technology than MSN and Yahoo, but MSN's and Yahoo's search results are often very similar to Googles. My suspicion is that M and Y use user data even more than G, and since G drives the lion's share of the traffic, this makes their SERP's often follow G's SERP's.