Average Time on Site

[skatofiabah]

Hi,

How do I look at for a given page (for both visits to it and strictly visits where an entry happened to it) how do I look at the Average Time on Site for each page (both visits to them total and separately look at only entries to those pages for the time on site (Whole visit)?

Thanks!

[kayawalton]

Hi @skatofiabah --

If I understand your question correctly, you are looking for the Average Time on Site for any visit that touches a particular page (let's call it "home") and the Average Time on Site for any visit the entered through the same page. 

You will have to create two segments. The first one will be a visit-level segment that includes home. The second one will be another visit-level segment where the entry page is home.

Segment A:

Segment B:

I usually create a freeform table with a time dimension (such as Day, Week, Month; except for Minute and Hour) as rows and Average Time on Site as a column. If you want to compare, you can drop the segments under the Average Time on Site metric.

I do not recommend using these segments as dimensions/rows because that will calculate Average Time on Site as a function of any dimension you use within the segment. (It is how Adobe Analytics calculates time spent.) Applying these segments as dimensions will only calculate for the total time spent on home AND the number of unbroken sequences within that session as opposed to the entire session itself.

[skatofiabah]

Hi @kayawalton,

Wait. So I can't just put a visit and entry visit segment on dimensions/row and average time on site on column? How can I get that?

[skatofiabah]

Better yet, how can I get it so I have each segment as a different page (each row) and column to average time on site (or average session duration) for each page?

[Jennifer_Dungan]

The problem is that you are looking for visits that contain, or visits that start with... there will be overlap between your segments.... if you do add your segments as rows, make sure to turn off the "Totals" on the columns, because those values will be misrepresented by the overlaps.

I am not sure what your last question means... can you try to draw it out, or attach a screenshot showing what visualization you are hoping for?

[skatofiabah]

Hi @Jennifer_Dungan,

Here is a screenshot. Each segment on the row is a different page segment set up like: 

[Visit

               [Hit = Page A]

]

I'm also looking at the last 90 days and using for my metric adobe's OOTB "Average Time on Site".

I understand to ignore the total, but is this the right approach? Can I also replicate the same workspace and change the segments to entry page = "Page" for each segment as a different page/entry page visit?

Thanks!

[skatofiabah]

Also, the first 2 metrics are the same but the latter is different. Again, not sure if this is ok to do and why #3 is different than the first 2 metrics and what I should use/do?

[kayawalton]

I would use Time Spent per Visit instead of Average Time on Site. If you prefer the hh:mm format, you can create a calculated metric that transforms Time Spent per Visit (seconds) into that format (something I learned from @Jennifer_Dungan

[kayawalton]

From what I understand: you want Entry Pages or Pages as rows and Average Time on Site (or any metric representative of the average duration for any session).

This is easier for Entry Pages because you can use any prop for an Entry Page (URL or Title) in the rows and Average Time on Site for columns. That will get you the average session duration for any visits that started with the Page in the corresponding row because the Entry Page props (should) persist across the session.

Getting this data for any visit that included a particular Page is more challenging because most Time Spent metrics serve as a function of the row, and even if you use a prop, that prop does not persist as you change Pages. The output you get by putting Pages as dimension is really the average time spent on the corresponding Page during any given session.

If the challenge is to identify Pages that would be worth evaluating to see if they generate a high Average Time on Site (for the entire session), I would definitely start with the Entry Pages table given that it is your site's first impression. I would also use the Flow Visualization to identify Pages that might be worth analyzing and create visit-level segments for those. I would start with top "Second Pages" (the next page visitors go to after the Entry Page) or consistently present Pages when you look at visitor paths that have large page depth.

[skatofiabah]

Hi @kayawalton,

I have to have them as a segment to get it at the visit level and also because I have multiple old and new urls that count as the same page, so I can't use just the dimension prop or evar item. In that case, which is the one I should use? Both given completely different values and I don't understand the difference. Time Spent per Visit (Seconds) seems incredibly higher compared to Average Time Spent on Site (Seconds).

cc @Jennifer_Dungan 

[kayawalton · Accepted solution]

Got it. So let me walk back a couple of things I've said (I love time metrics in Adobe Analytics, but sometimes they require a bit of thought process):

I would use Time Spent per Visit on your table. It captures most of the session duration because it calculates the difference between the timestamp of the visit's first hit and the timestamp of the visit's last hit (Total seconds spent) AND divides the total time by the number of visits (minus bounces) that apply to that segment. 

The Average Time on Site is a bit of a misnomer because it uses unbroken sequences as the denominator. Depending on your visitor flow, there could be more than one sequences triggered by the dimensions in your segment during one visit, which is why the Average Time on Site is lower than Time Spent per Visit. 

Adobe considers these two metrics similar, but I think their intended use is very different. Time Spent per Visit is meant to look at complete sessions (like in your case), while Average Time on Site is more for when you are looking for more granular time data (such as specific times spent on a particular page or dimension).

Phew, hope that clarifies the difference!