Were there any design-related changes made to the site over that couple of days period? Seems like too significant of a drop for it to just happen.
It sounds like you are getting a higher percentage of relevant traffic than before. This is speculation only, but someone posted this [webmasterworld.com] (post #4608054) which may be related.
Perhaps your site is part of the area where and if Google is pre-testing their next large algorithm update and you're benefiting from it prematurely.
This can happen when an event is set up in GA. I have a 30 second event to poll the site to see if someone is still there to track real bounces vs. those that read 1 article and were satisfied.
It can also happen if you inadvertently place the analytics code twice on the same page or pages.
Yes, could be events or social button plugins that have Analytics integration.
Check out your keywords & landing pages. You might have gotten more traffic from your good keywords to your good landing pages or less traffic from your bad off-topic keywords.
A site I have changed in 1 day - 29 April 2013 - from *years* with a bounce rate of about 65% to 4%. I made no change(s) to have made it happen. Right now that site is staying between 4% and 8%. Analytics is not particularly trustworthy in my opinion. To me it's kinda like Alexa - The data is OK for reference but that's about it. Definitely not particularly accurate. Anyway - What ever happened was a Google Analytics thing, nothing I did.
I once had a drop from 50% bounce rate to 10% or something like that, without any apparent reason. After looking into it, it turned out that I had installed a new Wordpress plugin on my site, which made the Analytics code load twice per one visit for a large number of my pages. Since the code loaded twice, an actual bounce would have registered as two pages viewed, hence lowering bounce rate.
I'm not saying this is the reason of course, just something to keep in mind. A good way to determine whether this could possibly be the case is to see if you have an abnormally high number of page views (from different visitors) that are even numbers. So if most of your visitors page views are 2's, 4's and 6's (with very few 3's and 5's etc.), this could be a clue.
P.S. yes, I did try to use this fact to my advantage and attempted to "manipulate" my bounce rate to see if Google would increase my website's rankings based on a better Google Analytics bounce rate. No dice :)
What's missing here is information from raw logs. What's physically happening?
Even without this information, the obvious question is: has there been any change in the distribution of landing pages? Something in visitor behavior has to have changed. If not, then it's an error in analytics. You need to find out whether it's a new error (bad)-- or a former error that has now been fixed (good).
Over the years I've seen a gazillion problems with faulty implementation of Google Analytics code, particularly from developers or even flash and interactive content developers.
Are you certain that your Analytics tracks as it should? As Lucy24's post above suggests you need to check with your server logs to discover "discrepancies" between them and Analytics. On pages with traffic check number of "hits" in logs and then validate this number against what Analytics says about them.
Do remember a few years back a campaign that I was assured that the code worked well, only to register things like 88 pages per visit on a 10-page microsite! You just can't be certain on anything if Analytics is not tracking properly...
It coincides with setting up an user action event as #taberstruths suggested.
So, nothing has basically improved with the site?