"Less than five seconds" is highly misleading (?)

Status
Not open for further replies.
#1
Hi everyone,

I'm relatively new to StatCounter, but am so far very impressed by it. It shows me a ton of very useful data. However, one aspect troubles me, namely visit length.

Here's the situation. It has been explained to me that if a visitor loads only one page on my site, then StatCounter has no way of telling how long they were there. The visitor could have checked out that page for 2 seconds and left. Or they could have been there for an hour or more. If they don't load a second page, (apparently) Statcounter says that visit length is "not applicable", at least in one part of the interface.

That would be fine in itself but then there's the data in the Visit Length pie chart area.

When I look at this, I see that 90% of my visitors (apparently) stay for "less than 5 seconds."

What's more, I believe that if a visitor loads just one page, StatCounter puts that visit in the "less than 5 seconds" category in the pie chart.

If that's actually the case, then that creates a pretty major problem: it makes the visit length pie chart data highly unreliable.

Right now, the visit length pie chart for one of my sites says that 90% of visits are "less than five seconds" - there's a huge chunk of the pie devoted to that.

However, if what I've said above is correct, that information could be completely false.

Many people load just one page on the site. They may be on that page for 5 minutes or more. But it seems that Statcounter counts that as "less than 5 seconds" as far as the pie chart is concerned.

Firstly, is that correct?

If so, am I right in thinking that the visit length pie chart and "less than five seconds" category are highly unreliable?

It seems to me that those visits should be marked as "not applicable" or "unknown" and NOT as "less than five seconds." Putting them in "less than five seconds" category may (falsely) inflate the so-called "bounce rate" and give webmasters a misleading representation of visitor activity.

I end up thinking: "Wow, 90% of my visitors leave within 5 seconds" when in reality they could be on the page for 5 minutes before leaving. As I say, this would be very misleading.

However, I'm fully prepared to accept that I've misunderstood StatCounter. As I say, I'm new to the software and so far have been thrilled with (virtually) everything about it.

Could somebody please throw some light on the above, therefore?

Kind regards,

Adam
 
#2
It used to show 0 seconds, people panicked, then Statcounter decided to count it as less than 5 seconds.

I personally interpret it as a "one-hit visit" and leave it at that. Whether it's 1 second or 1 hour it's impossible to measure when there's just one hit involved.
 
#3
Thanks for your response webado.

To be honest, I think "less than 5 seconds" will induce just as much panic (and is just as misleading) as "0 seconds."

If it's impossible to measure (which I accept), then in my opinion it should say "unknown" - not "less than 5 seconds."

It doesn't make sense to say that a visit is "less than 5 seconds" when it could very well have been 5 hours for all Statcounter knowns.

Here are some options which would be less misleading:

(1) Statcounter could have a table showing different visit times, for example:
Unknown = 100 visits
10 seconds or more = 20 visits
etc.
etc.

(2) Statcounter could have a pie chart that shows page loads, not visit length, e.g.
Single page load
Two pages loaded

At the moment it seems Statcounter is (misleadingly) conflating "visit length" data with "page load" data, and this has the potential to give webmasters the wrong impression, I think.

I don't want to be told that 90% of my visitors are on my page for less than 5 seconds when in reality they could be on it for 5 minutes.

Are there any statcounter moderators or staff members who would like to comment on this?

Thanks again for your help. As I say, I think Statcounter is fantastic in so many ways. It's just that this has caused me some issues.















It used to show 0 seconds, people panicked, then Statcounter decided to count it as less than 5 seconds.

I personally interpret it as a "one-hit visit" and leave it at that. Whether it's 1 second or 1 hour it's impossible to measure when there's just one hit involved.
 
#4
The thing is even for visits of 2 pages or more the time spent on the last page of the visit is unknown, so it's not included in that visit length anyway.

This is why I personally don't use the visit length for anything except to note it represents generally 1-page visits.

Maybe one day Statcounter will provide a metric for the number pages visited during a visit rather than or in addition to visit length.

FAIK maybe it's already available and I've not caught up with the most recent developments. The information is available in Recent Visitors activity but would have to be filtered differently and presented sorted by the number of hits from each visitor.
 
#5
I just think it's a mistake to assign visits a definite length (e.g. "less than 5 seconds") when in reality it's unknown.

I was alarmed when I saw that 90% of my visitors were there for "less than 5 seconds." Then, after research, I realized that they could have been there for 5 minutes.

So I recommend that Statcounter either stop saying "less than 5 seconds" and replace it with "unknown" or "single page visit" OR that they include a clear note for users, so as to avoid causing panic.

Like you, I think I'll just starting translating "less than 5 seconds" into "single page visit"
 

Rory_A

Administrator
Staff member
#6
Many thanks to the both of you for your feedback. Hopefully we can get this changed into something that more clearly represents what is actually happening.

In my mind part of the trouble is how does "Unknown" or "One hit" fit into a pie chart where all the other entries are based on time length? Maybe it doesn't belong in the pie chart at all? That'll be up to people smarter than my to decide :)
 

Car Guy

Super Moderator
#7
Well my 2 cents, but I do believe it belongs in the pie chart as it IS a legitimate hit. Blogs will have an abundance of these hits and excluding them will greatly affect stats for them.

I think an "unknown" with a side note, as you suggested Rory would be appropriate.
 
#8
Hi Rory,

Thanks for posting.

Yes, I see the issue. If you call the visits "unknown" then it can't really be in a piechart (how big would the slice be? Nobody knows since you can't express "unknown" as a percentage - there's too much data lacking)

My thought was that it might be better to have a table, listing what IS known, e.g.

Visit length
Unknown: 500
10 seconds or more: 62
30 seconds or more: 25

This way the visits aren't expressed as a percentage/fraction of of a while. They're just listed in a table,

Hopefully this suggestion may help the programmers of the software.

As I keep saying, statcounter is GREAT in so many ways and I really don't mean to suggest that it isn't when I point this out.

I also think customer support is very good here too (which Google Analytics just doesn't offer!)

Regards

A
 
#9
I would have to relate this to Google Analytics "Bounce Rate" which they define as a single page hit. I don't know if Google excludes these single page hits from the "average time length" on my site or not. I don't rely on Google Analytics near as much as I rely on Statcounter's stats so I don't understand Google all that well.
 
#10
Yes, I've heard that Google define "bounce rate" as single page hits. To my mind, that's just as misleading as the current set-up in Statcounter.

If people land on your page, read everything on it and thereby stay there for 15 minutes, that's hardly what you'd think of as a "bounce."

Personally, I think of a "bounce" as someone leaving the page within 5-10 seconds, i.e. someone landing on your page and then hitting the "Back" button asap!

Out of interest, is there any way (within Statcounter or otherwise) to find this out?

As I've said (and as people seem to agree), Statcounter's "less than 5 seconds" metric is not reliable for this and is actually misleading.

So does Statcounter have a REAL way to find out how many people left your landing page within 5 (or 10 seconds).

Or is this something that can't be calculated at all?








I would have to relate this to Google Analytics "Bounce Rate" which they define as a single page hit. I don't know if Google excludes these single page hits from the "average time length" on my site or not. I don't rely on Google Analytics near as much as I rely on Statcounter's stats so I don't understand Google all that well.
 
Status
Not open for further replies.
Top