Hi everyone! There seems to be quite a bit of buzz surrounding webkit's ITP feature. We have authored the following KB article to help answer any questions you might have:
If you have any follow-up questions, feel free to ask them here.
Is it possible to identify/trend no. of unique visitors relying on fallback method(s). We're currently using Adobe's third-party cookie as we need to report on uniques across multiple domains and therefore helpful to measure any impact/change.
If changing your implementation to use MCVID is not an option and you absolutely must stay with friendly 3rd party cookies, everyone who is using Safari is going to be your total impact.
Luckily one of the fallback methods is to use a 1st party cookie on the domain visited, so unique visitors per domain will still remain fairly accurate.
We are using omtrdc.net which in theory should not have had an impact as these cookies were already being blocked. However I've noted a sizable increase in Unique Visitors across all iOS devices. Are you aware of this and why this would be the case.
ITP is not necessarily directly correlated with iOS, but rather the safari browser. People browsing via Chrome on Mac won't be affected by ITP, while people browsing via Safari on Windows will.
Are you getting the same trend line by only including people browsing via Safari?
Yes, I'd based on a significant proportion of our audience using Safari being iOS users.
See below ...
Interestingly increase in Uniques only applies if Safari AND iOS, however more importantly (my previous point) it should have not impacted third party omtrdc.net cookie as it was already being blocked by safari
hmm .. only applies to entires to our homepage .. and no referrer
It's across all domains (ips) inc those assigned to larger service providers (ie unlikely to be bots). Whilst it's also possible these may be new device purchases it does seem highly unlikely with uniques on mobile almost double
since the iOS/safari 11 update we have a problem with rising unique visitors on mobile iOS as well. We cannot use first party cookies / MCVID either, since we have to track data from multiple domains. If I understand your "Adobe Analytics and ITP" article correctly, then this problem shouldn't occur at all, since the third party cookies should've already been blocked by safari before. We're using omtrdc.net, too. It sounds like a quite similar scenario as mubarakd5726232's.
The even bigger problem is, that these additional visits/sessions mess up our marketing channels data, since the "new" sessions are recognized as internal visits (HTTP-referrer comes from internal domain on session start).
Is there anything we can do to solve this problem?
seems we're on our own ..
Can you please look to create something similar
segment detail for Bot Review : 1PV/Visit is ..
If you could then duplicate the report but have breakdown by day instead ...
You'll notice a significant proportion in the increase is from new uniques only looking at 1 pv/visit .. ie appears to be an issue with fallback
What I've seen is
1. the weekly uniques increase (with above segments applied) .. is almost 100% .. and with no increase in PVs/Visits
2. the average daily uniques have only increased by 20%
.. so it seems very likely a percentage of users are being seen as uniques on every visit and/or possibly on page view.
If our data is consistent that would confirm an issue at Adobe/Apple end.
I recreated the report on my end but it did not show the same result.
I get "new" Visits/Visitors which do have a referrer and also create more than only one page view each. The problem occurs when a user hops from one domain to the other in our case (can be identified via entry pages). This should be prevented by fallback methods (User Agent + IP) normally, but somehow this doesn't seem to work anymore for iOS 11 users.
Please take a look at the increasing share of Apple users starting in week Sep 25, 2017 running into the Internal URL Filter:
Do you see a similar problem with your internal URL filter maybe?
I've been looking into this further ... the table below shows a substantial increase in uniques and which relates to 'new visits' where visit count is 1.
I was then able to cross reference this with a logged in user ...
The above shows the user doing similar no. of PVs in Oct and Nov vs Sept BUT the uniques increase significantly.
The following shows 2 uniques, 2 visits and 7 PVs were recorded for the user between 10-11
Which then broken down by minute shows a new visit was recorded after the first page view
The above should not be happening as we're using adobe's third party cookie domain which should mean the fallback method is used where a first party cookie is dropped instead.
I've been investigating this further and can state categorically 'we' do have an issue. I've an ongoing ticket and have also asked to be escalated internally. For reference 'we' are using H.25.3 and third party cookie omtrdc.net
In addition to shared insights I've verified by ..
1. Tracking a logged in users' hits/uniques - new visitor mid visit.
2. Comparing to Google Analytics.
I would suggest running the following report to see if your affected by this too. If uniuqes are being inflated then you'll see PV/Visitor decrease overtime specifically for iOS as more users upgrade to iOS11 upwards
Our overall uniques figure is being inflated by upto 25% (December). So this is significant (for us atleast)
Visits are breaking mid session and thus inflating Visitors and Visits metric. As this is happening mid session and the cookies being seen as new .. you should see the inflation if you trend visitors/vists over time and apply the filter apple devices, no referrer and visit no = 1
We were running on older H.code and therefore recommendation was to move over to Appmeasurement library and/or moving to first party cookies. Moving to first party cookies is currently not a viable option therefore we updated to AppMeasurement last week, unfortunately this does not appear to have resolved the issue.
The visitor inflation is affecting our overall monthly uniques by ~16% (in Jan and rising)
This obviously impact other measurements too .. eg PVs/Visitor
Will post here if I have any further updates, but would help if other customers were measuring and reporting this.
Update : Please note Adobe has recently changed it's advice and updated the original article
ITP is inconsistent with Adobe's 3rd-party cookies. Occasionally, cookies are accepted but are frequently removed, even mid-visit.
There's no timelines on a work around solution and therefore my suggestion would be for Adobe customers to move over to 1st party cookies
Yeah we noticed the change on the the article too....
So would you move over to 1st party cookies?
It would be great if someone here can share us a case that moving to 1st party cookies solved their problem.
Coz we need some proof for our management to have confidence that this solution can fix the problem.
Since not many other Adobe users reported the similar issue. We are wondering if this affect more on the high visit frequency sites.
We observed our visit freq.(Visit/Unique Visitor) on iOS 10 or below is 4 and iOS 11 is 2.
How is the visit freq. pattern on your site?
see below ..
Late Friday evening I moved over one of our sites to using Experience Cloud ID Service. For those that aren't aware this whilst dropping a first party cookie also drops a third party demdex cookie and thus allows for cross site de-duplication of visitors (if browser accepts a third party cookie). Based on 2 days worth of data, the results are very encouraging with various metrics showing improvements (and hopefully corrections) in line with issue of cookies being deleted mid visit.
1. No. of Visitors with no referrer has dropped significantly
2. Page Views per visitor has increased
On this basis I plan to roll this out across all our sites.... Not sure why Adobe is not recommending this to customers still using third party cookies.
I'd summarise 'our' steps as follows. Your may differ depending on your implementation.
1. Requirements for the Experience Cloud ID Service
2. If you're updating to AppMeasurement code from H code, check/update your plugins code as some are incompatible or will need updating. AppMeasurement Plug-in Support
Hope that helps.