I don’t use bounce rate. The reason why pretty interesting, too.
Fundamentally bounce rate is something that you’d really want to understand about your traffic. People go to the site, then leave immediately without doing anything. Campaigns with high bounce rates are usually just generating garbage traffic, and you’d think bounce rate is a great indicator.
On most websites, though. bounce rate in Google Analytics is broken. Let me explain.
Google analytics measures activity by recording “interactions”.
This can be page loads, clicks on specific things like add to carts, scroll depth on a webpage, etc. These interactions are defined as “hit” and “non-hit”.
“Hit interactions” will affect bounce rate. This makes sense for add to carts, new page loads, basically anything that suggests someone is engaging with your site. The user isn’t “bouncing”.
“Non-hit interactions” will not affect bounce rate. This type of event makes for “nice to haves”, like “scroll depth”. Someone can scroll 100%, or view a specific part of your landing page and still “bounce”.
Most people’s sites are so messed up that bounce rate doesn’t even work. Apps, poorly implemented tracking codes, multiple developers. Over time, it’s inevitable that older websites end up with “Non-hit” events that are measuring as “Hit”.
I’ve seen some landing pages record 0% bounce rate because an event carrying custom dimension was configured improperly.
I don’t look at bounce rate, especially for landing page or channel comparison, because it will misguide you. Not “if” it will misguide you, but “when”.
Instead, I measure progress of specific events. Page views, add to carts, initiate checkout, purchase. Where is the drop-off?
This is MUCH more insightful, and much more reliable. For these metrics, whether they’re coded as “hit” or “non-hit” the number of recorded events is always the same.
These events can break in other ways, but they are significantly better than relying on bounce rate as an indicator of performance.
Do you guys use bounce rate? Or something else instead?
Let me know in the comments.
Recent Comments