If you are ranked in the top three for your choice keywords, updating your site is a bad idea. Why?
When you update your site Google recalculates your SEO score. This score is made up of over 200 variables, the majority within your website. If you are ranked #1 for your keyword, or #2 or #3, in reality there is no other way to go but down.
When you make changes to the site or add content, you are changing this 'score.' You currently know your site has a higher score that the sites below.
Google's scoring system is not perfect. Within any system of added ratios and metrics the best results are always in the middle ie the mean / average scores.
What tends to happen is that the top (and the bottom) scores are kept artificially high / low. It is similar to any pricing calculator that has a large number of variables; it is impossible to get all of the quantities and prices to be accurate across the board.
This is why the #1 ranked website for a keyword rarely (if ever) falls to second or third.
If you are in a market where your website is bouncing around on page one, you would be better to stop making changes or adding content. Work externally, link building etc. In this way you can only improve your position, there being no such thing as a bad link or a spammy link and hence in theory your score can only improve. However in our recent research we are finding that Google is placing less weight on backlinks that we had previously thought.
If you are ranked #1 don't do anything, as your high score can only reduced and your website can only go down,
The most common method used to be a time stamp. This is time and date each header section of a web page returns when called which indicates when the pages was either created or re-created (ie updated).
However this is old technology. What happened was that websites became DYNAMIC, ie that web pages that call data when the page is visited. This in essence create a new page (or newly updated) page every time it is visited by a user or a bot. Wordpress or any ecommerce site is a good example.
So what happened to a timestamp?
Timestamps are still in play but of little importance. If the web page returns a timestamp then it can only be of account if it shows a past date (not today's date!). As an experiment we left brantacan.co.uk (with a timestamp of Oct 2010) un-updated. It ranked #3 in Google for bridges.
It took two years for the site to fall to #12, even though the site was telling Google that it had not been updated for more than two years! When the site fell to #12 we updated the site, increasing the page sizes, this changed the timestamp to a more recent date. We will now watch it climb back onto the first page.
We know that most web pages do not return Google a timestamp. Pages that do (and have an old creation date) suffer a small penalty and will eventually fall. Timestamps are soon to be confined to the technological bin.
So how then does Google know if a page has been updated?
Possibly by page size. Google can compare the size of web page and compare the KB size to the size of the same page when it last crawled the site. If there has been a change then Google will re-index the page, decide on a score and add that to the websites overall score. This can make the site go up or down! This site is not often updated in terms of page size change, however today (03.04.13) we are going to make changes in the footer which will result in every page changing size. We will then report the changes in traffic flow.
There may be other ways Google can calculate changes, but this would be the most obvious way. In Google's own video (see below) Google actually simplifies it search.
When Matt Cutts (porn cookie king!) say s that "when a user make a search, Google search through billions of pages to see which web page is appropriate" it is not actually true. What Google does is search through billions of pages and then allocates a score for each pages dependent on keyword, It is this database of "scores" that Google is really searching through.
This database is pre loaded; so that when you search with Google's 'Google Instant predictions' turned on and type an 'a', Google has then divided this data alphabetically, so that an "a" can only get the "a" section of this data.
Google is making noise about Semantic Search, SEO gurus are taking notice, online marketing bloggers are filling pages.... good news, bad news or no news?