During the past year, I became a little excited at one of Google’s many enhancements to the presentation of search results, because I suspected it could hint at a possible ranking factor they might’ve introduced. The element in question is a date stamp.
You may’ve noticed that in some cases Google will prepend the usual listing snippet text with a date. That change was introduced sometime around late 2008 or early 2009. I noticed the addition of the date with interest, but I became even more interested after I heard Matt Cutts state in a Webmaster Help video that Google considered the date to be helpful to users:
When Google states outright that they consider some element of webpages to be “useful” to searchers, my ears prick up, because Google is so obsessed with Usability that they sometimes use quantifiable elements of user-centered design in their search algorithms, such as their recent introduction of Page Speed as a ranking factor. In this way, Google’s Usability fixation can reveal ranking factors.
I wasn’t alone in twigging to the dates in search snippets — Michael Gray thought their introduction might be significant, Barry Schwartz reported on them, and Performics also shared my suspicion that dates might influence rankings, although their analysis didn’t find any evidence that it did. Edward Lewis (“pageoneresults”) also observed the date snippets and theorized that they might be beneficial for newer content.
On one hand, there’s a compelling reason to think Google might give ranking benefit to pages which have the date — their user testing has indicated that people may prefer to be able to see the date in the search snippets. Therefor, pages which have this data which could be shown in a snippet might be given a small boost in rankings in order to give searchers more of what they want.
Edward Lewis’s early theory also makes sense — for some types of content such as news stories and blog articles about current events (and perhaps even for photos), the freshness could make a difference. Yet, other types of articles are about old subjects or could be timeless — in which case the freshness wouldn’t/shouldn’t affect rankings.
I tested just now while writing this by searching for a major current event from earlier in internet history, and I see pages with 1997/1998 dates in the snippets. See: “worldcom and mci merger“. Since there are even more current articles which mention those keywords, the fact that some of the older stuff ranks higher seems to indicate to me that the actual age is not necessarily a ranking factor so much as perhaps the presence of the date itself.
Another theory could be that while Google might not use the presence of a date as a ranking factor, it could still indirectly affect rank because users might be more inclined to click upon search results where they can see the date value. If CTR from search result listings is a ranking factor that Google uses, then the presence of the date in the snippet could be impactful. (Clickthrough Rate or “CTR” as a ranking signal is a controversial subject, along with bailout rates, which I fully recognize.)
Since I think that there’s a high chance that the mere presence of a date may benefit search rankings either directly or indirectly, I believe it’s better to have them than not, particularly for articles and blog posts. But, this is a point where Michael Gray and I diverge.
He recommends revoking the date after 3-6 months, on the premise that users will tend to be biased against older content in nearly every case. In some cases this is a very valid point. For instance, if I’m trying to research details about some software or hardware, I’ll want current information, and not something six years old. So, for that type of content — Michael’s right — consumers would avoid clicking on the result that appears too stale to answer what they need. The downside of this is that I’m going to bail out of his page ultra-rapidly, if I figure out that it’s old content that can’t help me. If I can’t find any contextual clues to the freshness, then I might waste more of my time there, but ultimately become highly annoyed with the site for failing to post a date, since info about how to do something in software from 8 years back is not going to help me. From this standpoint, suppressing the date field is annoying and just not nice.
But, for date-neutral information, he has a point — adding the date to the SERP snippet for articles which are not date-dependent could introduce a bias against clicking which might not otherwise be there. Assuming that the consumer doesn’t recognize that the date is not a high priority in the context of the particular subject matter.
So, how do you decide what to do about all these variables? Google really hasn’t provided enough information from their usability studies to allow us all to make a decision. In the wake of that, I’ve outlined some possible good reasons to encourage dates in snippets, and also some bad reasons. But, where Michael Gray has decided to eradicate dates in most cases after a few months, I thing there’s enough information to argue that we should include dates in more cases than not.
To illustrate this, I’ve come up with a decision matrix to illustrate why I think there’s more pros than cons in choosing to display dates on one’s site. I created a column for each kind of content — fresh or old and date-dependent or date-independent, and a row for each theory on possible impact of date in SERP on rankings and CTR. The green check marks indicate that the combination would be beneficial, and red ex-outs indicate that a combo would have negative impact:
As you can see, there are relatively fewer possible cases where inclusion of the date would have a negative impact. All of those cases are for older content, but, as I pointed out with the very first screengrab, I see lots of cases where Google provides very good/high ranking preference for crusty old pages which have old dates (perhaps partly due to ranking factors giving preference to older, trustworthy, authoritative content).
Just anecdotally, unless it’s about current events or software details, I know that I click on old-dated content as well as new. So, I’d argue that age doesn’t at all count against rankings, nor even CTR in most cases.
Could Google be using the mere presence of the date as a ranking factor? Maybe. If so, I think that it’s a fairly weak factor compared to many others. But some of these weaker factors are the very hardest to obtain clear research data about — even if you’re an SEOmoz, you have difficulty in isolating the effects of one weak factor versus well over one-hundred other variables.
Due to the matrix I’ve shown above, I think there are more reasons than not to include the dates on your pages. However, I think that Michael Gray does make a number of valid points. You know your own content better than an algorithm, and if you know that your content doesn’t lend itself to display of dates, then avoid doing so. Tread carefully, though, because elements that Google deems important for user-experience may be closely associated with your rankings and CTR. You might ought to actually test whether pages perform better with or without dates before you summarily opt out. Try A/B testing it first.
I’d side with Michael’s implicit stance that Google probably ought to enable webmasters to opt out of display of dates if they choose. I’m generally a fan of giving webmasters rather more control over their own listing display than not — such as what was done with Yahoo’s SearchMonkey and Google’s Rich Snippets.
I’d also strongly encourage Google to publish more of their own internal findings about elements of usability or user-experience. The opacity in this case has resulted in a number of large content sites choosing to suppress dates on pages — a trend that I dislike as an internet user.