Could Google's Page Preview Affect Rankings?

[rango]

If Google is using the appearance of a website as a ranking factor, I wonder if a way they might be doing this is by tapping into data from their preview function. I'm talking about the function where you can scroll over the result and see a screengrab of the page.

I imagine they could have a system that works along these lines:

* Does the user look at the preview?
* Log if user ignores result after looking at preview.
* If multiple users skip over same domain repeatedly = minus points in serps for the domain.

Obviously extrapolated over thousands of users to make it worthwhile. You wouldn't rely on a few users skipping over a result.

What do you think? Would this be plausible?

Not sure how it really changes anything we do, except having a much harder look at what our site looks like in that screengrab :)

[crobb305]

Log if user ignores result after looking at preview.

This sounds perfectly reasonable to me. I have wondered about it myself, based on my own surfing habits. I find that when I am using preview, I ignore websites that generate "preview not available" messages (not 100% sure why those are generated unless it's a nocache thing), I also ignore pages for a variety of other reasons, based on layout.

[deadsea]

I find the previews annoying. They tend to pop up when I don't want them to. I don't think I've ever used them on purpose. Maybe I use them a lot less than other people, but I can't imagine there is a lot of signal in the data if other people ignore them like I do.

[petehall]

I don't think numbers are important for this type of data, only %'s. so even if 50% of people ignored them, the remaining would still reveal a pattern / trend.

[onebuyone]

Especially when looking for solution of some technical problem, previews make it easy to see if some web page is a content farm(a lot of auto generated crap showing up lately).

Low CTR should affect rankings, however I don't see those content farms dropping out of SERPs. So maybe it's just me.

You should also consider Cached problem, how many users don't go to your website, but they visit google cache instead? I do that very often, but again, it may be only me.

[londrum]

the thing that annoys me about previews, is that it encourages webowners to open up their image and photograph directories to google's bot, because they dont want their preview to look rubbish.

i know some of the previews are done on the fly, but some arent, and only use the stuff that the bot has access to (according to googles help pages)

its almost blackmail on google's part, to get your images into their image search

[Samizdata]

I ignore websites that generate "preview not available" messages

Those messages are not generated by the websites, but by Google.

not 100% sure why those are generated

They are likely to be caused by restrictions on Google's bandwidth-eating robots.

I also ignore pages for a variety of other reasons, based on layout

Interesting search strategy - ignore both the ranking and the content.

...

[crobb305]

Interesting search strategy - ignore both the ranking and the content.

Absolutely. If I am doing research, I'd rather not waste my time. The "ranking" you speak of says nothing about quality these days -- but a quick preview will tell me if it has academic merit (for my purposes) or if it is a commercial entity. You only scrutinized my comments. What about the OP's question? Do you have an answer? Back to the OP's point: yes, Google could use preview data to further refine rankings.

[lucy24]

not 100% sure why those are generated unless it's a nocache thing

BrowserMatch Preview keep_out
...
Deny from env=keep_out

It has to be done with a 403 because Preview, like Translate and Wireless Transcoder, is not a robot.

Google's bandwidth-eating robots

Preview loads up the entire page even though the preview is un-scrollable so most images won't even be visible. If the user actually goes to the page, everything has to get loaded all over again.

when I am using preview, I ignore websites that generate "preview not available" messages

If a professional webmaster attributes this message to the site itself, rather than to g###, is it safe to assume ordinary humans do too? So you're damned if you do and damned if you don't.

but a quick preview will tell me if it has academic merit (for my purposes) or if it is a commercial entity

Can you explain that? I never use Preview-- as a human searcher, I mean-- and have no idea what information you're supposed to glean from it.

[crobb305]

a quick preview will tell me if it has academic merit (for my purposes) or if it is a commercial entity

Can you explain that? I never use Preview-- as a human searcher, I mean-- and have no idea what information you're supposed to glean from it.

If you are searching medical symptoms, and you click on a page ranking top 5 with an intriguing description only to discover that it is loaded with ads to buy pills, the preview tool might have revealed that to you before you wasted the time to open the page (especially if the page loads slowly because of the ads). The preview tool is also a path to the page cache, and often times the term that I searched for is no longer on the page (very frustrating) -- but the phrase is still in the cache (a blog, for instance, where the post has moved to an internal page). Also, in competitive industries, we have seen abandoned blogs and member profile pages infiltrate the rankings (discussed in other threads over the past few months and I have witnessed personally). A preview of the page might reveal that before clicking. These are just a few examples of my usage.

[agent_x]

Preview loads up the entire page even though the preview is un-scrollable so most images won't even be visible. If the user actually goes to the page, everything has to get loaded all over again.

The preview may show different parts of the page though, depending on where in the page the snippet of text was found that was relevant to the query.

[Samizdata]

What about the OP's question? Do you have an answer?

In my experience, lack of a preview has no effect whatsoever on ranking or traffic.

The "ranking" you speak of says nothing about quality these days

That would be the "ranking" that the OP spoke of (and is concerned about).

I like it that my sites rank well, and I believe it is related to their quality.

...

[rango]

Hi Samizdata - does this mean that you have blocked the preview functionality on sites and not seen any difference in rankings?

That would indeed point to it not being used as a significant factor.

The ability to block the preview does make it a lot less useful as a factor. I don't really get the preview unavailable message all that often though. I think mainly it's when the page has been removed. Or I get suspicious that the site is trying to hide something, so I don't click those ones.

[Samizdata]

you have blocked the preview functionality on sites and not seen any difference in rankings?

The sites where I block Google Web Preview rank better than ever, and traffic increases.

I attribute that to quality and continued expansion, nothing to do with previews.

That would indeed point to it not being used as a significant factor

I don't think it is a significant factor, though petehall made a fair point above:

I don't think numbers are important for this type of data, only %'s. so even if 50% of people ignored them, the remaining would still reveal a pattern / trend

It is hard to imagine Google not crunching those numbers (or percentages). But I think it would be one tiny signal amongst several hundred others, easily trumped by quality.

I get suspicious that the site is trying to hide something, so I don't click those ones

Obviously I won't get traffic from the kind of person who insists on being provided with a Google screenshot before clicking a search result, but they seem to be a tiny minority and I can happily live without them.

Paranoia strikes deep
Into your life it will creep

...

[rango]

I'm not really *paranoid* about sites without preview. Being suspicious is not the same as being paranoid :)

It's just that I have had a few occasions where the preview wasn't available and after clicking through it led to either a non-existent page or a really awful page that redirected me to another page. So if I notice no preview available now, then yes, I am more suspicious.

In actual fact, I don't really use the previews much at all. It just occurred to me that this could be a fairly reliable (visual) quality signal for Google. I'm sure they have enough users who do use it to get some sort of information out of that.

Then again, by blocking Google from this option, maybe you can avoid it affecting you negatively too :)

[tedster]

I think it's an interesting idea of a potential signal. I think we can assume Google does collect this data. Whether they've found it useful enough to apply it in the algorithm in some way is another story. I'd guess that if they do use it, it's a very minor signal.

[crobb305]

I think it's an interesting idea of a potential signal.

I agree. These data are at their fingertips; they can use anything as a signal. We're talking about CTR, which every advertiser uses as a signal.

Google can create a cumulative frequency distribution for CTR when preview is not used, when the preview is used, and when the preview lead to a click on "cache" (valid entry point to a page), in combination, for a given result. They could calculate a conditional probability that a user will use a result for a key phrase.

[ken_b]

I'd think that would be a pretty weak signal.

I'm pretty sure I'm not the only search user that fumbles around with the mouse and ends up loading the preview without even thinking about it, let alone intending to.

.

[crobb305]

I'm pretty sure I'm not the only search user that fumbles around with the mouse and ends up loading the preview without even thinking about it, let alone intending to.

Good point. As long as Google has the preview tool as a mouseover, rather than an intentional click to a screenshot, it probably is a weak signal. I was just giving a hypothetical on how I think those data might be used, if they are used.

[rango]

I'd say they can detect those users who are loading it accidentally (which I agree happens a lot) and those who load the preview on purpose, just by looking at how consistently that user does use it. They'd then only pay attention to the ones that are frequent users. Still presumably enough to get some good data off of.