Bob Heyman today on Search Engine Land notes that the Google freshness factor may mean big implications for retailers. He notes that the EVP of ice.com, a large Internet retailer, is making proactive changes to their site because of the recent search "options" functionality introduced by Google that allow searchers to select "recency" as a criteria.
They are indeed correct. When you search on Google you will see a "Show Options" link at the top of the SERPs. When you click this link, you will see the "recency modifiers" options of "Any time", "Recent results", "Past 24 hours", "Past week" and "Past year". These allow searchers to refine the search results based on how recently the pages were updated.
If you sell products online then you probably don't need to update your product pages all too often. This will have a negative impact on your traffic levels if many people adopt the usage of Google's recency modifiers, because your pages that haven't been updated in a long time won't get listed in SERPs that require recently modified pages.
The first few things that come into my mind are: "daily changes", "Last-Modified", "checksum" and "page size". If you can keep all of these in mind and know how they relate to each other, then you should be able to engineer yourself into always having fresh content.
Google are looking for pages that are recently modified, so the best way to fit into that criteria is to actually add new content to pages daily. Keep in mind though that they are probably look for pages that exceed some threshold of new content before the page is actually considered changed or updated. So just adding or changing 1 sentence on a page with 100 sentences probably isn't going to cut it. I don't know what the threshold is, but I would be comfortable recommending a guideline minimum of 10-20%. This means 1 or 2 new stories every day for a page that normally features 10 stories.
I know what you're thinking... I'll add some random content and every time a search engine sees the page it will be different. I generally advise against this because if Google find that your content is completely random, then they will be a lot less confident sending traffic to you for a specific keyword, given that the relevant content that was on the page at the time they spidered it will likely be gone when a user goes to see the page. Frequent change = good. Random = bad.
So. Commit to making a few changes throughout the day and you should always be there for a "Past 24 Hours" search.
"Last-Modified" is an HTTP header which a web server sends with the response to a request. The Last-Modified header tells the client (the search engine spider in this case) when the page was last modified. It's very likely that Google and other search engines wanting to determine freshness will look for this header. However they won't completely rely on it because it can be "faked" to whatever date the Webmaster wants. So, search engines will still look for content changes. Always sending the current time is bad.
It's important to note that the Last-Modified header is not always sent by default. It is sent most of the time with static content/pages, but sites that are dynamic generally don't send this header by default due to the complexities in calculating the true last time of modification. If you're selecting a CMS, this may be a worthy consideration. Incidentally, there is also something called the "If-Modified-Since" header, which you should look into.
Finally, a quick and dirty way to check for changes to a page would be to compare the checksum values and the file sizes to previous versions of the document. I won't go into much detail here because I'm not sure that Google are using these methods, but if 2 versions of the same file pulled on different times have exactly the same size, then there is at least a small probability that they are identical.
The checksum method is more accurate, but still not perfect. A checksum comparison will compare the checksum of 2 versions of the same document, and if the checksums are identical then there is a good chance that the documents themselves are identical. This method gives a pretty accurate yes or no answer as to whether the 2 documents will be identical. It does not measure the degree to which the documents' contents differ (the percentage of content that is different).
I hope this helps to at least get you thinking about this important issue. I know that I'm using the recency modifiers quite a bit, but I don't know what the adoption numbers are; hopefully Google tells us at some point. Submit a comment or get in touch if you have something to say!