Say you wanted to determine how many errors a typical visitor would encounter when browsing a site, what would be the calculated metric you would use? So, for example a typical visitor is likely to see 'X' error upon every 10th visit but then also 'Y' error upon every 15th visit. How would you be best show the likelihood of a visitor to see that type of error?
Solved! Go to Solution.
Views
Replies
Total Likes
The formula would look like:
(errors / visits) * 1000
Just like "Percent" meaning "per 100" is:
(errors / visits) * 100
Now, normally percent is shown as a "percent" notation (adding a % to the end, this indicates the per 100).. but if you want to show just "number or errors", you can show as decimal and say "Average Errors per 1000 Visits is approximately: 3.46"
Meaning that out of every 1000 visits, there are likely to be about 3-4 errors.
Views
Replies
Total Likes
How are you tracking your errors? Does error X and error Y have separate events? Are you using a single event with a dimension to distinguish them (either the standard custom link value, or maybe a prop or eVar when the errors trigger - depending on if this is tracked as an action on on the page view)
I would think the best place to start would be:
Error X / Visits
or
Error Y / Visits
but you may need to apply segmentation if these are a shared event (or maybe it's just a dimension and no event at all - we can also deal with that).
But Error / Visit should get you an average "error per visit" data (leave as decimal for a "per visit" - or you could do % for an "error rate")....
If you want to scale this to something like "errors per 10 visits" or "errors per 20 visits" you could always scale your denominator....
Views
Replies
Total Likes
Hi @Jennifer_Dungan , yes each error is a separate event. Can you please give me an example of scaling the denominator. Are you meaning, "errors per 30 visits" as the next scale, etc.? Thanks
Views
Replies
Total Likes
Errors / Visits is "Errors per Visit"... meaning you get approximately 0.something errors in a visit (unless you have a lot of errors, then you may be over 100% error rate (when using visits, since you have the potential to encounter multiple errors within the same visit, over 100% is a valid calculation).
But if you say wanted to know approximate errors in 30 visits or 40 visits, you can just take your "errors per visit" (i.e. per a single visit) and multiply by 30 or 40 (this will get you the average errors in 30 visits or in 40 visits)
Depending on number of errors, this should still be a decimal value....
And if you have high traffic, and low errors, you might prefer instead to look at errors per 1000 visits, or for a really high traffic site, errors per 10K or 1M, etc....
So maybe you end up with a "3.46 errors per 1000 Visits" or something like that...
Views
Replies
Total Likes
Thanks @Jennifer_Dungan What would that custom metric look like for 3.46 errors per 1,000 visits? Just errors divided by visits?
Views
Replies
Total Likes
The formula would look like:
(errors / visits) * 1000
Just like "Percent" meaning "per 100" is:
(errors / visits) * 100
Now, normally percent is shown as a "percent" notation (adding a % to the end, this indicates the per 100).. but if you want to show just "number or errors", you can show as decimal and say "Average Errors per 1000 Visits is approximately: 3.46"
Meaning that out of every 1000 visits, there are likely to be about 3-4 errors.
Views
Replies
Total Likes
Ok great. Thanks @Jennifer_Dungan . This is quite helpful.