Segment to identify refreshes (multiple pageviews on the same page in one visit)?

[florianag]

Hi all,

How would you go at creating a segment to identify cases where a visitor repeatedly refreshed a page during a single visit?

We had a case of non-malicious potential bot traffic, and one of the isolating factors for the traffic (which we could not identify by IP, due to internal anonimization rules) was that the same page was visited and refreshed an unusually large amount of time.

We would like to run the same analysis site-wide, to then reverse engineer potential instances of this type of traffic. How should I build this segment?

Thank you in advance for your help,

Floriana

[Urs_Boller]

did you try to play around with “page views“ and „reloads“?

a) segment where „reloads is greater that 5“

b) calculated metric „visit“ { “page views“ / „reloads“ } and break down by a visitor id (if you have captured one as a prop/evar)

[florianag]

The issue I would have with this is that I would like to segment for reloads on the same page URL, not just reloads during a visit :/

Say someone has bad connection and visits 5 pages, reloading each one every time, I would capture this visit in the segment, while I would be interested only in visits where one page is reloaded more than once...

I guess what I would need is a component that captures for example "number of distinct pages/URLs visited", then create a calculated metric such as "reload/number of distinct pages visited" and filter for visits where this metric is higher then 1, 2, 3 or whatever needed.

I'll try to do a breakdown by URL with a calculated metric reload/pageviews, I think this should give a similar result to what I'm looking for. Thanks for the suggestion!

[florianag]

And... Reload / pageview is a metric that can't be used in a segment

[Gigazelle · Accepted solution]

I think what Urs was referring to was to create a separate report, listing visitorID's by the calculated metric reloads/pageviews. That way, you could immediately identify visitors who have an unusually high ratio of reloads, and investigate them first.

This calculated metric definition worked pretty well for me:

The reload hits segment was simply defined as all hits where reloads exists.