Major Change in Supplemental Result Handling

[g1smd]

Continued from [webmasterworld.com...]

---

Major Change in Supplemental Result Handling today:

Over the last 18 to 24 months, I have written many times about how a page can appear as a normal result for search terms that are located on the current version of the page, and as a Supplemental Result when you search for words that were on the previous version of the page (but are no longer on the current version of the page).

In the latter case those "old" words also appear in the snippet too. In both cases (old search and new search) the cache is usually just a few weeks old, so it never shows any of the words associated with the "old search".

As of today, the new search is still linking to the new cache, but the "old search" now brings up a cache that is dated just one or days before the date of the last change of content on the page, and therefore the cache DOES now show the old words from the old content.

This is a new thing today, and Google has NOT worked like that at any time in the last two years or more. So, rather than get rid of old supplemental results, Google now gives them more space on their server, now actually keeping the old cache copy for them alive too.

I was hoping that old indexed data with no matching cached page was going to get deleted from Google's index in their currrent tidy up.

However, what they have chosen to do, is not to delete it, but to now keep an older copy of the cache to go with it. This is addition to keeping a new copy of the cache in the normal index.

I have seen this effect on a large number of pages today. It doesn't happen for all sites, may be not all that data is complete yet?

Google begins to look more and more like archive.org every day.

So,if you alter a page, Google will return that page for the current content but it will also return that page if you search for the previous version of the content. Before today, you could only see a modern copy of the cache. Now, you get to see either a new copy or the old copy depending on exactly what you searched for.

[g1smd]

A large website whose domain expired two weeks ago, had 12 000 pages listed, many of them supplemental for the last few years. The root shows a "domain expired" message. All other pages are gone from the site.

Google reindexed the site and overnight the number of listed pages has been reduced to under 100 on the "experimental" DC. It seems like Google is aggressively throwing away old data, whereas before they would have held on to it for years and years...

On the old "normal" DCs, Google still shows 12 000 pages listed.

[bloggy]

One of my sites that had +/- 10,000 indexed pages have been removed from the Google index on May 28/29. Since then I wasn't even able to see my site using this command: "site:mysite.com" in google.

Yesterday Google listed my site again, but only 1 page (the main site / entry page) is listed now in Google. Can anybody here tell me if you think that Google will re-index all of the 10,000 pages that have been indexed before this happened?! Or will I have to start from zero again!?

I appreciate your help and honest opinion.

[g1smd]

Which datacentre is that on?

One from which set... BigDaddy "A", BigDaddy "B", or the "experimental" ones.

[Ellio]

Reseller,

Ellio
"All is OK for keyword searches and rank is uneffected but is seams weird to use header "alt" text rather than useful info in the snipets for site: search. "

You may wish to consider removing that header "alt" text, and force Google to choose something else for the snipets!

I agree this would be a good plan but I am reluctant as our SERPs are very good and I am not convinced that keywords in genuine "alt" tags are not taken into account in scoring.

[bloggy]

@g1smd: how can I find out?! Sorry, but I'm not a Google Pro :-)

[WW_Watcher]

Ellio
I was hesitant to say anything, but IMO, you are correct, it is better to have great placement with a less than perfect discription, that might change from search to search, than not to not show up where it cannot be seen. IMO google does take alt tags into consideration. If the tag is revelant to the search, and not stuffed(as I suspect a properly SEOed page would be) I would leave it.

Number of posts does not always equal quality of posts.

Back to watching
WW_Watcher

Edited to add, I just caught that it was displaying the alt text only during the site: search, I would no-way change that alt tag. If you are the only one seeing the alt tag in the discription.

[suzie250]

Maybe this is not something new, but just noticed it myself today....

When I search for a term that I know is in a supplemental result, the snippet shows capital letters on some words that are not capitalized on the actual page.

That just seems so weird to me. What is changing the case of those letters?

Edit: After looking at it more, some letters that are capitalized on the original page are not in the snippet.

[Ellio]

WW_Watcher,

Ellio
I was hesitant to say anything, but IMO, you are correct, it is better to have great placement with a less than perfect discription, that might change from search to search, than not to not show up where it cannot be seen. IMO google does take alt tags into consideration. If the tag is revelant to the search, and not stuffed(as I suspect a properly SEOed page would be) I would leave it.
Number of posts does not always equal quality of posts.

Back to watching
WW_Watcher

Edited to add, I just caught that it was displaying the alt text only during the site: search, I would no-way change that alt tag. If you are the only one seeing the alt tag in the discription.

Thanks WW_W thats exactly my point. I was simply pointing out a strange change in behaviour. Why would I change the alt tags when it only occurs when doing an informational site: search.

Snipets for real searches are just fine.

On another point as of today the site: search is showing the pages in different order with the homepage first but then it is showing pages from a sub folder rather than the next most important pages.

This is a change fron normal behaviour as we normally see some sort of structure to the site: listings.

[Whitey]

I've got a feeling that Matt might be about to communciate on the instability issues we are facing, coupled with a bit of an appeal by numerous webmasters on his blog.

[mattcutts.com...]

[Whitey]

I'm noticing a change which happened about 4 hours ago

[64.233.185.104...]
[64.233.171.104...]

the meta description is no longer shown. It has been replaced by CSS page elements [ menu items ] used for navigation.

I haven't checked the other DC's

[F_Rose]

I have noticed this morning, on all sets of datacenters our pages dropped out of supplemental results.

We only have 24 pages indexed from our site, which is a small chunk of our site, pretty discusting.

The good news though, no supplemental results..

Now we have to get Google to start indexing all of our pages..

Any suggestions?

[Atomic]

Patience

[ClintFC]

Yup - Patience.

We've been waiting two months so far. Can't be too long now can it? Or...can it?

Patience in this case may well be a complete waste of time. I don't understand where people get the idea that, for Google, this is anything other than all-systems-normal. They've said nothing so far to indicate that they are aware of any problem. What little they have said, via Matt Cutts (a.k.a GoogleGuy(?)), indicates that they firmly believe there is nothing wrong...shoirt of possibly a slight reduction in crawling frequency.

[F_Rose]

"Patience in this case may well be a complete waste of time."

That is my concern. With Google being silent, thier is no indication what G is actually up for..

If Google is up for a major change in algo,maybe we where affected, if so..maybe some changes on our site needs be done..if so..I would like to know..every day passing is a waste of time..

Google please speak up..

[bloggy]

hm.... my sites have been indexed again (for 7 - 14 days). Today I checked Google again (site:mydomain.com) and they are gone, again. But still have their pagerank ... what's going on there ... this is very strange.

[Webdetective]

Is going to the supplemental results a kind of penalty for doing something Google doesn't like?

[glengara]

As some have suspected all along, you're all just a bunch of spammers ;-)

[mattcutts.com...]

[bloggy]

It doesn't make sense. I don't have links to unrelated websites on my site, it's 100% original content, no tricks, nothing ... but still: this site has been removed from the index weeks ago and I can't see any reason for this.

[dgdclynx]

Last Friday all the supplemental entries naming my home page, dozens and dozens of them, disappeared from the listings. The next day a spurious supplemental entry for the home page itself disappeared. Everything has been swept clean. The genuine entry for the home page remains together with a score of mentions of it at other sites.

Is this new by Google or is it part of an ongoing process to tidy up the listings.

[Oliver Henniges]

glengara, I quickly went through the blog-entry you pointed to. If you ask me, a discussion of the issues Matt raises, deserves a thread of its own on the front-page here in WW.

I wouldn't call myself an SEO expert, but if the relevance of outbound links really plays such an important role, this is somewhat of a revolution. Intuitively I pursued just this strategy in the past months (which actually doesn't mean very much, because I hardly bring more that one page every two weeks to the web).

Nevertheless, I absolutely cannot complain about my ranking in google, and I believe this has to do with exactly that linking to my "competitors."

[g1smd]

Google has cleaned out all the Supplemental Results dated prior to 2005 June. If the page has been 404 for a very long time or the domain has expired before that date, then those Supplemental Results are now all gone from the index. If the page has been edited a very long time ago, then Google now only has the "new" version of the page as a normal result, and the old version (previously tagged as Supplemental) is now gone.

For pages that have been edited since about 2005 June, Google now has the page as a normal result when you search for current content, and as a Supplemental Result if you search for content from the previous version of the page from several months ago.

For pages that have gone 404 since about 2005 June, or the domain has expired in the last few months, Google now shows that page as a Supplemental Result with a cache date from just a few months ago.

It looks like Supplemental Results are here to stay, but at least Google now only has stuff going back about nine months, not the almost three years that they did have before.

This is a big improvement and it happened about a week ago. Matt Cutts has also said that the Supplemental Results will be (slowly) refreshed throughout the Summer too.

[g1smd]

The bug that surfaced a week or two back where on a site:domain.com search every snippet showed the same information and so that search quickly hit the "In order to show you the most relevant results, we have omitted some entries very similar to the 154 already displayed. If you like, you can repeat the search with the omitted results included." message has now been fixed.

The snippet has now reverted back to showing the words from the meta description tag like it used to do previously. The site search is now useful in revealing problems with duplicate title tag data and duplicate meta description data again.

< datacenter watching continues here: [webmasterworld.com...] >

[edited by: tedster at 1:17 am (utc) on June 15, 2006]