Shared publicly  - 
Question for the SEO community: Back in June of 2010 Google announced the release of Caffeine. Caffeine changes the way Google indexes the web to make it faster. Since the adoption of Caffeine I have noticed Google indexing pages extremely quickly. Sometimes mere minutes after they are published. So my question is: Why does it then take so long for rankings to change? I can build 10 to 20 very high quality links one week, get them indexed almost instantly, and not see rankings change on target pages for weeks. I have tested this on numerous sites and each time it seems there is a huge wait for links to take effect. Whats going on? #impatientSEO (cc +Matt Cutts +John Mueller)
Melissa Fach's profile photoArnie Kuenn's profile photoJohn Mueller's profile photoAlan Bleiweiss's profile photo
Joe, since the EXACT same thing so interested in the answers you get. :-)
I'd have to say it's based on the difference between indexing and computing the ranking signals. From what I know about pagerank being sort of recursive, it takes a while to compute. I'd imagine things like the panda signals also have a non-linear computation time associated with them.

Then there's the whole issue of propagating the data across all the data centers.

That's just my best guess.
+Ryan Jones Yeah all of that makes sense. I guess I should be asking why index something faster if you aren't computing at the same speed?
QDF! I'm sure there's different algorithms/factors at play when it's determined that the query deserves freshness. Personally, that's where I've noticed the most benefit of caffeine.
Joe. I'm surprised you don't know the answer to this. Dude. Remember - Google has flagged all SEOs and spies on us as an industry. Clearly they're not going to allow our work to be given preferential treatment and in fact, they're probably going to intentionally delay our work from having an impact on results. It's a conspiracy, I tell you! A Conspiracy!

How do I know? Well because I'm best buddies with Matt Cutts. Best buddies! We hang out every friday night shooting pool and he comes over to my house several times a month and asks me to help him suss out various algo issues. Which I do, of course, so I can get preferential treatment and not be targeted like the rest of the SEO community. Although that's now going to need to be our little secret Joe. Okay? Because if anyone finds out I've got such treatment, all hells going to break loose. ;-)
+Alan Bleiweiss Your one of the few folks I know that will take a considerable portion of your busy day to utter complete bullshit! :P
Isn't also partly that while they crawl and index quickly, they assign resources to rebuild the actual index on a priority basis based on the query? Crawling and rebuilding the index are fairly separate issues. I've seen highly seasonal queries for which the index doesn't get updated for weeks or months at a time, then gets rebuilt multiple times per day when in season.

I believe another factor is the age of the links. The full value of the links you are building will not necessarily pass to the page instantly. There is evidence that Googles waits some amount of time before allowing some or all of the link equity to flow to a page from a link to ensure that the link is permanent rather then transitory, or a "rented" link.
+Rob Woods That's a really good point, it does make sense for them to rebuild the index based on query.
Joe, thanks for the compliment. To have the desire, willingness, courage, fortitude, tolerance, capacity, intellect, passion, vision, ambition, patience, open mindedness, guts, and most especially, completely fried concept of rationality, takes a great deal. While most people THINK they're contributing to society, I actually am. #YoureWelcome
A literal LOL > " +Alan Bleiweiss Your one of the few folks I know that will take a considerable portion of your busy day to utter complete bullshit! :P" Just to add a little value... I agree with IncrediBill - they may have indexed the page, but they don't immediately know its value.
The ones getting in quickly are from trusted sites (mostly news/blogs) and IMO all but Rob are wrong about crawl separate from Rank calc cuz with parallal/segmented indexing they are instantly updating a segment of the index... however the Panda (for sure) and possibly PageRank algos are still updated separately from indexing.
Interesting - the words I cut and pasted in my comment (above) from another comment appear as strike through on G+. Never saw that before.
+Terry Van Horne do you have any post/links to support this? I am not disagreeing, I am just curious.
Okay look. Larry and Sergey made me promise not to reveal stuff. Said if I did, they would not invite me back to our semi-monthly "Let's ask Alan what the next thing is we should focus on" sessions. But I will give you a hint on this one. Read Terry's comment.
We use over 200 factors in our algorithms, so while links are also a part of that, there's usually much more involved as well. Sometimes new pages do rank well very quickly, it's not something that's per-se impossible.

Also, I'm sorry to hear that you sometimes see our pipeline taking "minutes" before making content available in our search results :-/. We're working hard on speeding that up.
+John Mueller LOL! I am not complaining about the minutes! You guys are doing an amazing job indexing! I am just trying to understand the process, index first then rank later? Or rank and index at the same time?
Have seen SERPs where there was an item that was seconds old... so has to be some cases (likly QDF sensitive) that it is automatically ranked based on some pre sets, or that content hit a segment of the index that is being indexed. Also have seen fresh content rank for a short time disappear and come back so IMO, is not either or...could even be query specific with other influences like seasonality... ie if is in season is more likely to enter results directly.
Now I know where to go if I want to see Alan A. humor!
Wait. Did John Mu just tell us, in between the lines, that Google has abandoned thousands of factors? "We use over 200 factors". At last count, weren't they up to 10,000 factors.

This is epic. Changes everything!
No, I think it has always been around 200 factors. Never heard a number any where near 10K.
+Joe Hall We continuously update the information that we have about a page (though of course it has to be indexed before it can be shown/ranking). For what it's worth, we're working on some blog posts to help clarify the whole crawling / indexing process, so hopefully you'll get to see some more tidbits :).
And Arnie, I only shared that link with you because I care. Wouldn't want you to be embarrassed out in public thinking there were really never more than 200. And of course, as always, that link is top secret. Please don't share it with anyone. I need my private phone line to Larry and Sergey to remain intact!
Add a comment...