Direct vs Natural Last Touch discrepancy

[e30powar]

For some reason we saw a large divergence in our direct vs natural channels on February 22 according to Adobe. We have checked all release notes and tags to see if anything may have been changed to create this divergence and have found nothing. Also when viewing channels using Google analytics we see there are no changes in the trending of direct vs organic. I realize raw numbers will not match between Adobe & Google but trending should if users truly were coming directly vs searching. So to me this points to a setting in Adobe that must have changed.

Last Touch Report Jan 1 2019 - August 21, 2019

vs GA

The other curious thing is the referrer report, you can see Google vs Direct maintains a consistent trend, but this report uses instances (not visits) and I assume referrer=google.com would include paid and organic.

[Andrey_Osadchuk]

Hi e30powar,

If I understood it correctly, the first screenshot is showing Marketing Channels. If it's the case, the issue is likely in the Marketing Channels Processing Rules. You may want to review them and also investigate what was updated in the report suite on 22-Feb (Admin > Logs).

[e30powar]

Yes you are correct, the first is marketing channels. It's actually called Last Touch Report if that matters but it is breaking it down by marketing channels. The admin logs show nothing in February for the report suites log or admin log. The usage and access log show an API call hitting on 2/22

It starts on the 21st with API operation 'WebServices\Classifications.CreateImport'

and then on the 22nd repeats this: API operation 'WebServices\Classifications.CommitImport'

Marketing channel rules did not change and someone else recommended looking at paid search rules. We have them set to the standard "Matches Paid Search Detection Rules" and the paid search settings have the proper tracking codes.

[Urs_Boller · Accepted solution]

what is your timeout for the marketing channel? might it be an issue due to safari and ITP?

if that is the case, the users would loose the cookie earlier (7 days) thus loosing the marketing channel (new visitor). I mean maybe the just came by search once a month, the rest was always direct. the isers just had the "last channel" as "search" even they came directly.

to narrow down if that is the case, try to look at following:

1) any differences for "last touch marketing channel" when comparing safari against all others? mobile vs desktop?

2) do you see higher numbers in unique visitors after that date but visits stayed almost the same?

just a best guess, might be another reason as well (but you already checked the channel setup...)

[Andrey_Osadchuk]

If the cause was with the Paid Search detection, the trend chart would display changes for Paid and Natural, rather than for Paid and Direct.

Could you share (or send a direct message with) a screenshot of Processing Rules and the first chart but with Y axis in absolute numbers (the numbers themselves can be obfuscated if necessary)?

[e30powar]

Thank you for the reply. I'm pretty sure the rules are defaulted so no problem displaying here.

And here is the last touch report in numbers by day

and by week

[e30powar]

This is interesting and hard to tell if there is an issue because the majority of our traffic is from chrome or safari, so you'll see both have a divergence in the last touch channel.

Browser Type = Apple

Browser Type = Google

Also unique visitors vs visits doesn't show a significant divergence

[Urs_Boller]

some more ideas:

- what is your "paid search detection"? (btw: why don't you set the channel details to "campaign" or similar?)

- is "direct" set to "override" or not?

- did you launch a "PWA" (progressive web app) or something similar that made users to have easier access to the website?

[e30powar]

Paid Search Detection is set to Search engine = any with all the following rules. 

direct is not set to override last-touch

asking our developers about the PWA, will edit this once I hear back.

[jantzen_b]

Do any of the answers below answer your initial question? If so, can you select one of them as the correct answer? If none of the answers already provided answer your question, can you provide additional information to better help the community solve your question?