4 replies to this topic

[iamlost]

Note: this started as a full fledged rant but then I edited it, myself edited it, and finally me edited it and the raging irritation was pretty well dampened down. I think.

I wish folks writing articles would define 'bounce rate' beyond the generic 'single page visit percentage'. Why? Because (1) the purpose of the page and (2) how people leave the page are critical. Ye olde 'bounce rate' catchall statistic is near worthless.

Appropriate segmentation changes data to information:
1. Page Purpose:
* page with an implicit/explicit call to action within page.

* page with content that explicitly/implicitly answers query with/without links to additional information.

* page for a specific visitor intent, i.e. navigational, transactional, informational (to use one segmentation model).

* page irrelevant to query.

* etc.

2. Exit Method:
* click the browser 'back' button - this is considered the 'worst' signal to send the referring SE. BUT... what is critical is differentiating how long they wait before clicking 'back'; the damage time threshold for a SE bounce is probably quite short (~5-seconds).

* typing in a new url - while a technically correct bounce cause and certainly not a 'good' signal you are never likely to know if it occurs.

* close the open browser window/tab - this is neutral simply because you have no idea why it was done; there being both good, bad and indifferent possibilities.
Note: this is an identifiable signal but not one for which most webdevs script so practicably also a technically correct but you'll never know...

* session timeout (according to your analytics application setting) - also neutral for reasons above.

* clicking an onpage external link - this is the 'best' bounce and while technically a bounce I consider it a conversion.

A third bounce signal worth investigating is that of the referer.:
* is it a slashdot type site that typically sends looky-loo lemming type traffic?

* if not, what call to action, i.e.anchor text, surrounding content, are they using that is accentuating such traffic behaviour?

Whether a particular bounce or bounce rate is good or bad or indifferent is a great big 'it depends'. On the page, on the query - and results relevance, on the referer...on the visitor...

Regardless, bounce rate as typically portrayed is near meaningless and even less actionable.

[EGOL]

I wish folks writing articles would define 'bounce rate'

I think that you have thought about this subject a lot more and a lot deeper than these article writers.

My rant is that the folks who develop tools don't use common methods of determining bounce rate (and lots of other metrics).

I think that an awful lot of these webmaster tools are designed by "programmer guys" who have not thought deeply about how their tool should actually work - they might not even own a website.

(I am tempted to rant about shopping carts made by programmer guys who have never run an ecommerce site... but will resist.)

Edited by EGOL, 25 January 2012 - 11:44 PM.

[A.N.Onym]

That's one of the reasons why I've stopped on relying on the bounce rate metric too much.

If I get a lot of targeted traffic to the page and I get low conversions, then yeah, that's the bad variant of bounce rate.

If I get a lot of traffic from Slashdot, then by no means, having a bounce rate above 80% for this type of traffic would've been a miracle and a solid achievement (in reality, around 85-95%).

It all depends and all that usually falls back to other metrics, so I see bounce rate only as a signal that there might be a fire, but it's definitely not the fire itself.

Edited by A.N.Onym, 26 January 2012 - 01:52 PM.

[jonbey]

My bounce rate bounces at such a rate. Sometimes it can fluctuate hugely without me making any changes at all. One page that is popular on StumbleUpon use to record at about 90% bounce rate, but now records at around 20%. Not changed anything on that page in the meantime, as far as I recall!

One really odd stat, my Brazilian readers seem to produce an unbelievably low bounce rate, many pages recording as 0%. Maybe a cultural thing, maybe a web thing? I dunno. So long as some people are happy, I am happy. Never been one to optimise though, just make do!

[Michael_Martinez]

Every entrance path/channel/funnel has its own unique bounce rate and that bounce rate changes over time.

If you're looking at search referrals, you have to compute bounce rates for each keyword + page combination to really get any value from bounce rate analysis. It's not worth the time and effort, in my opinion, on a large site. Very small sites with only a small number of targeted expressions *might* benefit from bounce rate analysis.

As for whether the search engines use "bounce rate" in their algorithms, since we don't have access to their user performance data there is no value in trying to reverse engineer their processes. Even if we knew what they were measuring and what they were doing in response to what they were measuring, without access to their data we are dead in the water.