Bounce rate: huge permanent change overnight. Why?

General Questions
#1

Anyone here use Google Analytics? I am a rank beginner but the only one in the office who uses it at all.

Our site’s bounce rate went from averaging 5% (always in the 3 to 15% range) to averaging 46% (always in the 42% to 57% range) in one day. It had been that low since forever, and it has stayed that high since the change, three months ago.

Overall number of visits and visitors has not changed much. Phenomenon exists across all types of traffic including search, direct, referral, and email campaigns (though least pronounced in email). Phenomenon exists across all landing and exit pages that I’ve looked at, all demographics, etc. Ecommerce stuff (revenue, conversion rate, etc.) does not seem to have changed significantly. Page views per visit have gone down somewhat, but I think it would pretty much have to with so many more bounces. Day-of-the-week patterns have stayed the same – that is, relatively more bouncing on weekends and holidays, and relatively less on Tuesdays, while count of visits & visitors is down on weekends and up on Tuesdays.

No significant changes to the site or tracking code or content or anything in that time. (The inertia of our site is noteworthy.) Not featured on Reddit or Digg or the New York Times. No major product introduction coincides.

WTF??

Any ideas at all? I don’t know what else to look for!

#2

Update: Our tech guy made some changes to the tracking code at about that time, changes that fixed a false “This Page Is Not Secure!!!” message that was appearing on our shopping cart pages. He can’t figure out how those changes would cause a higher (apparent or actual) bounce rate, but it’s pretty much gotta be that.

The “average length of time spent on each page” metric went up at the same time and at the same rate as the bounce rate. The unique user count has NOT changed. The average number of pageviews, though, dropped inversely to the hike in bounce and time-spent rates.

I feel like Dr. Watson here. Where is my mind palace?

#3

Did your site have any redirects, perhaps associated with the tracking code? Did the site have a design where one page retrieved data or resources from another page? I’d think any of those cases might cause Analytics to think two pages were being hit. So even if the user left after viewing a single page, it would still be recorded as a bounce.

I’m kind of guessing wildly, maybe this will spur you (or someone else) to think of an actual reason.

#4

Thanks, TroutMan, that’s what I was hoping for – wild guesses for spurring thought. I disagree with your interpretation of your own scenario, though: If GA thought two pages were being hit because of a redirect or retrieval, it would have interpreted that as a non-bounce, because the definition of a bounce is a one-page-then-leaving hit, right? But maybe whatever our tech guy did *removed *a redirect or retrieval from the data GA sees. Which seems possible – he says what he did was remove multiple obsolete copies of the tracking snippet.

#5

Just summarized things for my supervisor as follows:

#6

Are you sure your numbers were correct before? 3-15% bounce rate sounds suspiciously low to me.

#7

This was my suspicion - that the recent change removed a redirect or something similar. I agree with Blakeyrat - the 5% bounce rate had to be wrong. I would guess that for the majority of visitors, the first hit to a page was previously counted as two pages.

#8

Not at all sure, and I believe the current numbers are much more normal for online retail / information sites in this sort of business. I didn’t have any reason to suspect them of being fouled up before, and not enough experience to put them in context, but I’m thinking now that they must have been wrong all along. I’m also wondering about what else the tech guy isn’t telling me, because he says the only pages that should have been affected were some of the cart pages, which I have always believed were a relatively small proportion of our traffic – even if they were being counted three times over, it seems unlikely that it would cause such a dramatic jump in the overall bounce rate. And the proportions on just those cart pages are not different from the overall site. So now I get to wonder about the overall site.

This is going to seriously suck when I get to explain it to my boss’s boss, I’m afraid. She’s suspicious of this whole “Internet tech” fad in the first place.

#9

I had a suggestion but I can’t reconcile anything (in that blue text block) with the average number of unique views not going up as well.

The only thing I can think of is that someone (or a few someones) are incessantly checking your page. That would bring the bounce rate up but keep the unique viewers where it currently is.

I don’t play with GA enough with my site to know exactly how powerful it is or isn’t but is there a way for you to determine if it’s one person (or one group of people) just checking your site on a regular basis?
Would there be a reason for that? Without know what your page is about, did do something on it on the ‘magic date’ that would make people want to keep an eye on it?
Take a look a things like what type of devices or what languages etc you can and see if it’s something odd. Who knows, maybe someone set it to their home page.

#10

Also, how long ago was this, GA made some changes a while ago and updated you interacted with the page, it took some getting used to and possibly involved some new code (I put some in anyways, I’m not sure if it was required or not I was changing some stuff around at the same time).
It’s possible you or your IT person is just reading the reports differently.

Another thing, make sure the reports are the same as they used to be. I suddenly had everything drop down to zero. Couldn’t figure it out for the longest time. After a week or so, I realized that GA was only looking at Conversions and since my site doesn’t report conversions back to GA, all the data was 0 for that analytic. I may have had 600 unique visitors in a given time frame but since (as far as GA was concerned) none of them converted, it reported zero. If she accidentally set something funky like that, it’ll skew the results.
Make sure that it doesn’t say anything between “Advanced Segments” and the Explorer tab below that. If it does, for example, like this (where it says “new visitors”), click the little gray X to get rid of that. It’ll cause problems if you don’t realize it’s there.
In that example (that I just did on mine) it’s ONLY showing results for new visitors, which means the ‘returning visitors’ metric drops to zero. That’s fine if you know what’s going on, but if you set it by accident (like I did once, still don’t know how) it’ll cause a real headache. For example, returning visitors are probably a lot less likely to bounce then new visitors (depending on your site I guess…I have a soup of the day on mine, I suppose returning visitors are probably actually more likely to bounce…should go check that…(ETA, yup))

#11

It turns out to have DEFINITELY been due to those changes the tech guy made (the ones that “couldn’t possibly have affected this”). He unmade them yesterday and the results on GA instantly went back to pre-magic-date numbers – you can view hourly results at the Audience or Content Overview levels. Digging into Help docs, I see that Google is clear that “Multiple snippets of the code on the same page is not a supported implementation and may lead to incorrect data.” And it turns out that per my suspicion, the multiple snippets were in an “include” statement drawn on by ALL the pages, not just the cart ones. It’s just that that’s what he changed when the cart pages had an issue, and when he changed it for all the pages, the cart problem went away.

Joey P, I I guess there are ways to see if one person (or bot) is spiking your numbers, using advanced segments, but I guess I don’t have to go down that path. I did scratch around in various other non-advanced categories and saw nothing in devices or languages etc. The reports were all set up the way exactly they used to be – and when I viewed data for any time period spanning the magic date, the change was unmistakeable.

And the upshot is that yes, our real bounce rate is apparently a totally unremarkable 45-ish%, and someone (probably me) should have said at some point "You know, a 5% bounce rate is not just good, it’s literally unbelievably good." Oh well.

#12

Hey, can any of you guys recommend a good primer on web analytics in general or Google Analytics in particular? After all of this kerfluffle I’m realizing that nobody here understands this better than I do, I don’t understand it well at all, and we really really need to use it better. And everything I’m finding online is either dry as old toast and overwhelmingly in-depth (like the GA Help pages) or self-help-hype-booky in tone (sample actual quote: “Five simple things you can do with absolutely no knowledge of Web Analytics! No matter what your site does you can use the ideas below to instantly benefit from your website data.” yuk).

#13

I am actually having the same issue, only that my bounce rate has gone from an average of 45 - 65% to between 2 - 5%, basically over night. It happened about two weeks ago, actually I probably shouldn’t complain, its like a dream come true… but it made me suspicious.
The truth is I had a similar pattern on a differen wordpress site I set a up a couple of years ago, it would constantly show a bounce rate of 5 - 9%, at the time I thought it must be theme related or simply that people were forced to navigate from the landing pages… but now I can’t explain it to myself… looking at the graph at a constant high it suddenly goes from one day 65%, next 22%, then 13% and after that always at 3 - 5% with exception of one Saturday when it jumped to 25% just to go back down to 5. The truth is I have been working on this site a lot, changing content, plugins, site structure… I have no way to tell what it is caused by! I don’t want to screw it up either, but do want to know why it happened!

#14

I don’t trust Google Analytics. They’re probably using the data from it in their search engine.

#15

Of course they are, why else would they give away a product like that unless they can find some way to make money on it? But even so, who cares?

#16

Because it influences your position in the search engine.

Why would they otherwise? Easy, simply to attach people to their brand.