Stacked Segments vs Sequential Segments

Hi ​@PaigeQ ,

To me, the segments appear to be different.

The vendor version captures conversions that occur after the Home page and within one subsequent page view. The conversion is also defined based on a specific URL.

For the three internal segments you created (assuming they are being applied as filters on top of the UV metric in the Freeform table), there are some nuances related to sequential segments. Sequential segments provide options such as "Only After Sequence," "Only Before Sequence," or "All Hits." Based on my understanding, the segment would behave as expected only if "Only Before Sequence" or "All Hits" is selected.

The major difference I see is the WITHIN clause in the vendor version. This clause ensures that the user reaches the conversion page within one page view after visiting the Home page. In contrast, the internal version simply uses THEN, which could include users who viewed the Home page, navigated through additional pages in the same session, and then eventually converted.

Another question is whether you are using the same conversion metric across all conversion analyses. In the vendor version, the conversion is tied to a specific URL, whereas the internal version may be using a different conversion definition.

Based on this, I would expect the audience captured by the vendor version to be a subset of the audience captured by the internal segment version.

Thanks,
Nitesh

​@PaigeQ , just to make sure I understand how your segments have been setup, do the following match your segments?

Vendor’s sequential segment

I had to nest the “Page = homepage AND Search Engine - Natural exists” in its own container (because AA doesn’t allow combining AND and THEN together in the same container) and set that nested container to use a “Logic Group” scope. I assume this is how your actual vendor’s segment has been built too.

Internal Team’s segments:

Segment 1

Segment 2

Segment 3

Are all 4 segments shown correctly?

So the first thing to understand is how your first segment is working.

You’re missing a container here – multiple different operators can’t be used at the same level. So your segment must either be

Visit (only after)
   Hit
      Page = homepage
      and
      Search Engine – natural exists
then
user enters the conversion flow

Or your segment is

Visit (only after)
   Logic Group
      Page = homepage
      and
      Search Engine – natural exists
then
user enters the conversion flow

It seems like a small difference, but both have very different results. ​@PaigeQ can you confirm which of the above your segment actually is?

Assuming that it’s a hit container in the visit segment:

In this case, that segment is saying that the homepage view has to come before the conversion flow, and that the homepage view has to happen in the same hit as the natural search engine.

Alternatively, if it’s actually a logic group container in the visit segment:
although both homepage and natural search have to happen before entering the conversion flow, there isn’t anything to say that they have to be in the same hit. For example, if someone came to the site from an email campaign, clicked on homepage, then went to a search engine and clicked on a link that brought them back to your site on a product page – that would get included in the logic group container (just insisting that natural search and homepage both exist, but not specifying that they have to be the same hit).

So depending on which version you’re using, that can explain why you’re seeing different numbers than stacking individual segments.

Hi ​@PaigeQ 

The two reports are producing different numbers because they are measuring different things.

The vendor's sequential segment looks for a specific user journey: a visitor must come from Natural Search, visit the Homepage, and then enter the Conversion Flow in that order within the same visit. If the sequence is broken or the events occur in a different order, the visitor is excluded.

The internal team's approach uses stacked segments, which only checks whether a visitor met all the conditions at some point. It does not enforce the order of events. As a result, a visitor could have come from Natural Search, visited the Homepage, and converted at different times or in a different sequence and still be included.

Because attribution is intended to understand the path that led to a conversion, the sequential segment is the more accurate method. It validates the actual customer journey (Natural Search → Homepage → Conversion Flow), whereas stacked segments only confirm that the required interactions occurred somewhere within the visitor's activity and may therefore overstate the audience.

There’s a lot of discussion, and my apologies for not reading everything…

I did see ​@yuhuisg’s post showing segments, and that is my interpretation as well…

And I think right there is the biggest difference… even with a sub-container at hit level in the one segment, ultimately, the segment is a VISIT scope, and will be returning all HITS from Visits that match all the criteria…. whereas stacking multiple segments, one of those segments is specifically at a HIT level, and thus, even if some of the segments are pulling back Visit data, the one segment is then the major limiting factor to only loading that specific hit.