13 replies to this topic

[rolf]

In light of a recent discussion here http://www.highranki...h...c=42753&hl= I found the article at http://www.siteprone...rankings-again/ quite interesting.

To summarise, the author is saying that whilst the meta description still doesn't affect generic rank, all(?) results will be personalised from now and will be affected by CTR, hence a good meta description means a better snippet means a better CTR means more people with your site in their future personalised results.

I can see the logic in what is being said but will continue as I am whilst bearing it in mind and watching how the situation develops - what does everyone else think?

[1dmf]

From which point of view rolf

from G!'s, it will use them when it thinks it is relevant, but it seems G! is producing more and more dynamic snippets, and alot of the time complete garbage.

from a semantic W3C point of view, the description is meant to be there to describe the document content and compliment the title tag.

[rolf]

From which point of view rolf

[copywriter]

I haven't read anything from Google on this, but it is true that they do insert their own description tags some of the time. Regardless of any benefit from the search engines, it makes sense to write a compelling description tag to encourage more clicks to your site whether G plans to use CTR as a metric or not.

Would seem to be a bit unfair if they do plan to use it seeing as how we wouldn't have full control over all the elements that attribute to getting people to click from the SERPs.

[Jill]

First of all, where did Google and Yahoo say they don't use the meta description in their ranking algos? (As it states in the article.) They've said that about Meta keywords, but I don't think I've seen them say it about descriptions.

They do, in fact, use description in their ranking algo since words used within them and nowhere else will cause the page to show up in the SERP for that phrase.

The rest of the article just doesn't make sense either. If I understand it, they're saying that if a particular page gets lots of clicks, it will raise it's rankings in personalized search. And they're saying that your meta descripton being good can help it get clicked.

Both of those statements are misleading. First, as we have already discussed, Google doesn't always show your meta description as the descriptive snippet; it totally depends on the keywords searched. So right off the bat everything else is out the window.

Second, the click throughs are only going to make a difference to ONE person's personalized results--the one who clicked a lot. Personalized results are PERSONALIZED. They don't effect other people.

Silly article, imo, full of speculation based on erroneous information.

[copywriter]

What G says and what G does can be two different things, but this is where they say they don't use KW tag or Descrip. tag in rankings.

http://googlewebmast...s-meta-tag.html

Even though we sometimes use the description meta tag for the snippets we show, we still don't use the description meta tag in our ranking.

Our web search (the well-known search at Google.com that hundreds of millions of people use each day) disregards keyword metatags completely. They simply don't have any effect in our search ranking at present.

This was Sept. 09.

I have no idea what Yahoo does with their tags.

[Jill]

Yep. Another instance where the search engine doesn't even know what they count or don't count! (They don't index meta keywords for rankings, but they do meta descriptions--at least last time I checked a few months ago.) I can dig up my tests if anyone is interested.

[copywriter]

They do that with certain aspects of duplicate content, too.

Sometime - when you're bored and having nothing better to do - I would love to see your tests.

[Jill]

Interesting turn of events. I'm not seeing evidence of Google indexing Meta description words at the moment.

Can anyone with old test pages around corroborate? Most of my old test pages are no longer active, so what I'm looking at now might not be 100% accurate.

[1dmf]

Second, the click throughs are only going to make a difference to ONE person's personalized results--the one who clicked a lot. Personalized results are PERSONALIZED. They don't effect other people.

I like your thinking Jill!

[Randy]

Well, it's a good theory 1dmf. One I could take issue with if I were so inclined and wanted to be a contrarian. Not that I would actually take the contrarian view where meta descriptions->rankings are concerned.

That said, as a for instance we could certainly say personalized results are personalized to the individual. However we could also quite legitimately make the point that when you're working with a large enough sample size of users, some of whom fit one known profile and some of whom fit others, you can in fact start to Predict what action certain users in certain profile sets are more likely than not to take.

Think of it all as a type of large scale conversion testing. Where you're recording real actions and using the data collected to predict what similar users will do in a given situation when presented with certain choices. Eventually it becomes a rather simple issue of Sample Size.

And if anything, Google should be in the best position to obtain a statistically valid sample size.

Jill: I don't know that I have any actual meta description tests still up and running, but I'll hunt around and see if I can't find something somewhere that might be useful.

<Edit to add>

This one may be useful Jill.

The <cough>word</cough> schtuff is used only in the meta description of the home page of my personal site. But the only places Google sees that word on my site is on other pages. (The home page is technically outside of the blog, and only blog pages are returned.)

So this one very limited example still supports the idea that they don't utilize meta description content in the search algorithm. But they do still show my default meta description for the home page if you do a site: type of search and don't include any keyword phrases.

[Jill]

My tests were for words that don't exist on other pages, and in the past, my page where the word was only in the meta description and nowhere else, G would pull up that page when you searched for that word. They don't when it's in the meta keyword tag.

I may still have a test page, I just need to remember which it was and what the secret word was! The ones I found yesterday were meta keyword ones.

(You can see how carefully I document these things )

On edit...Randy, that word appears on your page, if I'm looking at the right one.

[Ron Carnell]

Does it matter?

Either the meta-description doesn't affect rankings at all or the effect is very, very negligible. Like Jill, the last I checked, Google did indeed index the meta-description. If a nonsense word was inserted into the meta-description, something that didn't exist any where else, that page would be returned in a subsequent search for that nonsense word. Ergo, we knew it was being indexed. Unless you're trying to rank for nonsense words, however, that knowledge is of fairly marginal use.

As Jill already observed, click-throughs are only going to make a ranking difference to one individual's personalized results. And even that is largely irrelevant. A click-through sort of means the individual has already been to my site, for crying out loud. If I have to depend on Google for him to find me again, I've already failed. Instead of worrying about ranking better for that one individual I would be far better served to try to figure out why I couldn't sell him something. Or, at the very least, get him to bookmark my site while he was there?

However, that certainly doesn't mean your meta-description isn't important.

In the article, Gilmore says "Google doesn’t always show the Meta description you provide. Sometimes it just compiles a random text snippet from your page that contains the keywords used in the query."

That's crap. There is absolutely nothing random about the snippet. Google uses the first instance of a keyword on your page to build the snippet. If a query uses multiple keywords that are spread all over the page, it can indeed look random, but isn't that a reflection of your copy more than anything else? If the keywords are all over the place either you don't care about those keywords or your content sucks (for those keywords, at least).

A web page will return the same snippet for the same query EVERY SINGLE TIME. That's not random. That's predictable as hell.

You can and should control the snippet Google shows for all search queries you deem important. To not do so would be equivalent to buying advertisement and then letting some yokel off the street write all your ad copy for you. The SE snippet is your ad copy, your enticement, your call-to-action, and it has to be controlled, tested, and honed to a razor's edge. Just as you would if it was an ad you paid $10,000 for in a major magazine or newspaper.

The word optimization, in the SEO acronym, means to make as good as possible. If you build a page for specific keywords and then fail to control the snippet for those same keywords, you've only done half the job. That's not optimum. That's just lazy.

[Randy]

On the main home page (not the others that reference "schtuff" because I have it as part of a category name ) the only place it should appear is in the meta description. That page is outside of the blog structure, so doesn't carry any of the blog category links.