How can I have my total traffic be lower than the aggregated traffic by separated channels?

[asdf12341z]

So I have a landing page which users can arrive on through a variety of different marketing campaigns in place. I have created separate segments for each of these campaigns, which are tracked by SCIDs. 

Typically, the total traffic from each of these channels falls short of the total traffic to the landing page, and the difference I attribute to 'organic visitors' who may have typed in the URL, or saved it in their bookmarks etc.

However, recently I have noticed that total traffic to the landing page is somehow lower than the aggregated traffic from each marketing campaign, and I have no idea why.

For some additional info, my Landing Page 1 segment is defined at a hit level using page url, whereas the marketing channel segments are defined at the visit level, or visit with a 'hit' container. It's been a while since I made these segments, so I may have to review again why I made some of the segments at a visit level and some at a visit level with a hit container. 

Here is a snippet of the numbers below:

Does anyone have any suggestions on what could be going wrong, and how this is possible? This campaign has been running for a year and the last 2 weeks is the first time this has happened.

[igupta]

Hi @asdf12341z ,

Could you please provide additional info -
1. Is the metric you are using a custom event or unique visitors? Typically Unique Visitors would be de-deduplicated hence will always show lower number than column values sum.

2. If it's a custom event? Can you please specify how it is set?

Are there any values overlap in channels? You can maybe use Segment Comparison to find out why? Can you try to breakdown your table two with Marketing Channels and observe the individual counts?

Best,
Isha

[Jennifer_Dungan]

This is because when you use Dimensions as your breakdown elements the "de-duplication" of your metrics doesn't occur.

Your segments are "Visit" level, meaning a Visit that has both "channel a" and "channel b" will appear in two segments, and the total on that table will count both (inflating your total).

Example:

Visit 1

• Page A - channel set to "x"
• Page B - channel attributed to "x"
• Page C - channel attributed to "x"

Visit 2

• Page A - channel set to "y"
• Page B - channel attributed to "y"
• Page C - channel attributed to "y"

Visit 3

• Page A - channel set to "x"
• Page B - channel attributed to "x"
• Page C - channel attributed to "x"
• (user clicks on another channel while still in an active session)
• Page D - channel set to "y"
• Page E - channel attributed to "y"
• Page F - channel attributed to "y"

Segment A

VISIT

     channel equals x

Segment B

VISIT

     channel equals y

Segment A will return both Visit 1 and Visit 3, Segment B will return both Visit 2 and Visit 3

Therefore, Visit 3 is counted twice.

The lower total, in the second screenshot, is the non-inflated, and "correct" total.

[asdf12341z]

Thanks for you reply Jennifer, that helps!

Above is a brief description of the definitions of each segment (Channel 5 is my 'Organic' segment). If I change these to 'hit' level segments instead, will it likely alleviate the duplication issue I am facing? Is there any way around it? Because currently the reporting I am providing no longer makes sense, these segments are meant to represent the portion of traffic coming onto my page of interest. 

Another thing I realised is that the metric I am using is a custom metric defined as:

this isn't a metric I've created myself, and I've just realised it appears to be at a 'visit' level itself. So I am wondering if changing the definition of the segments to a hit-level will make much of a difference? In that case, would I also need to change this custom metric to be at a hit-level? I have actually never made any custom metric before, I had no idea they could also be at hit/visit/visitor levels. 

[Jennifer_Dungan]

Ah, yes, this is a bit complex.

First off, if you are trying to just get the "entry" channel.. so let's say:

Visit:

• Page A
  • campaign set to X
• Page B
  • campaign maintains value X
• Page C
  • campaign maintains value X
• Page D
  • campaign set to Y
• Page E
  • campaign maintains value Y

Let's say you only want Page A as that has the explicitly set "X", you could add a "Marketing Channel Instance exists", or set your dimension settings to "Instance"

However, are you looking for the "Landing Page by Channel" or the "Clicks on some element on the Landing Page by Channel" (because you won't have an "instance" - i.e. explicitly set value - on the click, if the click is relying on the Visit level attribution)

[asdf12341z]

---

@Jennifer_Dungan wrote:

However, are you looking for the "Landing Page by Channel" or the "Clicks on some element on the Landing Page by Channel" (because you won't have an "instance" - i.e. explicitly set value - on the click, if the click is relying on the Visit level attribution)

---

It's a combination of both, though in the above screenshot I only show the one for 'clicks on some element on the landing page by channel'. 

So the two things I am reporting on are:

1) Of the visits to my landing page, how many originated from Channels 1-7?

2) Of the visits which resulted in a click on a specific element of my landing page, how many originated from Channels 1-7?

---

@Jennifer_Dungan wrote:

First off, if you are trying to just get the "entry" channel.. so let's say:

Visit:

• Page A
  • campaign set to X
• Page B
  • campaign maintains value X
• Page C
  • campaign maintains value X
• Page D
  • campaign set to Y
• Page E
  • campaign maintains value Y

---

Apologies for the silly question, I want to ensure I am understanding you correctly.

Are pages A and D here representing the source page where a user would be interacting with the marketing campaign that leads them to my landing page?

So some element on Page A which takes them to my landing page would be 'Campaign X', and some element on Page D which takes them to my landing page would be 'Campaign Y'?

What I am finding surprising is that if it is the attribution model and visit-level segments which is the reason for my odd reporting numbers of late, that means that users have fundamentally changed their behaviour patterns in how they access the site, that too quite significantly after many months of pretty consistent behaviour (outside of a new marketing campaign).