If anyone else has a site that got hit hard around May 23rd by penguin 2.0 I'd be interested to know one very vague metric.
We've spotted a pattern in the small number of hit sites that we have access to. Those hit on that date all had a high (>50%) proportion of page depth 1 visits from organic search traffic.
I'm a little obsessed with the idea of Google using more behavioural data, so I might just be trying to confirm my own hunch here. I'd be interested to know if anyone is seeing the same / seeing anything contradictory. My suspicion is that the metric in itself isn't being used, but it is indicative of something that is. Possibly the proportion of visitors clicking another result from the same search after viewing the "low value" page - maybe using that signal as one indicator of quality alongside things like links.
I don't have any data to help out the theory, but I am concern about this too. I have a retail site that seems to have a very high bounce rate, about 70%. I have the feeling that this is normal for this kind of site as shoppers back out immediately if they do not find what they want. Also, the site is new and not ranking very well in the SERPs so by the time the consumer is down at page 2-3 of results their attention span is low and they are getting impatient. If bounce rate is factoring into the rankings than I hope they are considering current rank and niche specific behavior in the score.
Interesting. We had a commerce site tank around penguin 1 times. One of the changes we made before getting an uplift was to clear out thousands of high bounce discontinued products. That was really one of many changes made though.
If I were Google I'd certainly be using bounce rate as a "signal" of quality.
>I have the feeling that this is normal for this kind of site
I think anything north of 35% would be a very bad signal on a retail site.
I've recently been eyeing "Return Visitors" in analytics as a likely metric.
>eyeing "Return Visitors"
I think it is a great metric and the higher the % the better, your greatest source of revenue should be your existing customers.
Contrast that with a "push" to acquire new customers though and it starts to fall down a little.
A growing and thriving business should have a high % of returning customers.
A growing and thriving business should have a high % of new customers.
I think it is hard to call algorithmically as a quality signal.
Looking at the referring sites specifically, it seems that my high bounce rate is largely due to posting a couple of links on reddit where people look at an image and then back out. So now this makes me wonder if G looks at the source of the traffic when looking at bounce? Is it just from their search engines? Or are they feeding data from Chrome into the algo for rank? This shit could make nuts.
I'd would assume they would be using all the data they have to hand.
Image views for an ecom site are a waste of bandwidth, Pintrest being a small exception.
> the higher the % the better, your greatest source of revenue should be your existing customers.
The reverse is true if you're an adsense whore. Return users are like forum members --never even look at an ad.
Looks like I'm going to have to add some lime and orange and scare those fuckers away.
Its a new site with no existing customer base. It was part of an attempt at brand building.
so many ways bounce rate would be a logical ranking factor, but I'm not sure G s actually using it yet. Also bounce rate as per analytics can easily be skewed by adding an event or similar - for example on one site we allegedly had a bounce rate of less than 1% for ages, which was blatent bollocks, turned out we were triggering something which called another page in nearly all visits - so much like keyword density, you could argue about what actually defines a bounce and change the metric quite drastically with about 10 minutes work.
also sometimes a 'bounce' is a successful visit - if someone wants the address of an offline store, for example, or wants to read an article, even if the overall site is e-commerce.
That said, personally I think if you search and hit back it probably counts for something - ie; if a user repeatedly returns to the same search and clicks another result, the earlier ones probably didn't match the query very well. I don't think it matters if they come to your site, browse, and then go and do a fresh search for the same thing, but I have a very unscientific basis for suspecting the gradual downward trend of one particular page for one specific search (for which it is a theoretical perfect match and has inbounds) is because its a horrible landing page and the organic 'quality score' is dire.
Bounce rate is a very poor metric to use. I think that Cutts has also specifically addressed this question in the past and suggested that it is way too noisy to be of interest to them.
I think that SEOs have a habit of giving too much importance to the things we can easily measure. (Bounce rate, by source for example). Google though has access to much more useful data. Similar to what Gurtie is saying: If my site appears for a particular phrase Google know how many people click through from that SERP. They also know what proportion of those then click another search result because mine didn't satisfy them and how many made another search. Using sampling they don't need access to analytics for that and it is probably more useful than anything that Analytics has.
That said - bounce rate by source / term is still the best way we have od diagnosing such problems.
I'm pretty convinced that the above, or something similar is being used. It ties in with a lot of other stuff (fresh content advantage for instance) too.
The big question I would like to know is whether an exit via adsense would be seen as another search. I would hope not. If it did that would mean that a site switching from another network to adsense/adx would likely receive a drop in rankings over time. That would seem counter-productive.
I was just looking for something totally unrelated and came across this old post from Bill Slawski
http://www.seobythesea.com/2011/04/just-what-user-behavior-data-does-google-use-to-influence-search-rankings/
QuoteIf a document is returned for a certain query and over time, or within a given time window, users spend either more or less time on average on the document given the same or similar query, then this may be used as an indication that the document is fresh or stale, respectively.
The above relates to a patent by Google in 2003. It's a slightly different issue to what is being discussed above, but it does suggest that time on page (or rather between searches) is a factor that they have at least given serious enough consideration to patent versions of it.
Bounce rate on it's own is a crap measurement, but bounce rate as part of pattern matching can help them determine the bigger picture
IMO of course....