Branch Quick Links: Handling ECID Integration and Privacy Script Delay

[drashti1712]

Hi Team,
I’m currently working on a new mobile-web feature to incorporate Branch Quick Links into block CTAs.
Jira: https://jira.corp.adobe.com/browse/MWPW-152850

Context: Branch quick links allow us to redirect users to the Product App (if installed) or to the App Store/Play Store. An example of a Branch Quick Link is: https://lightroom.test-app.link/U26PImxk8Jb, and they can be authored similarly to regular links or buttons in Milo. Additionally, we need to integrate the ECID (Experience Cloud ID), a unique user identifier that will be passed to the specific product app.

Challenge: For the ECID to be included in the quick links, specific cookies must be enabled, and user privacy consent must be obtained. However, the privacy script currently loads with a default delay of around 3 seconds after the section load, which causes a lag in verifying privacy settings before appending the ECID. This ultimately delays the redirection to the app once the CTA is clicked.
Original POC link: https://ecid-5--milo--drashti1712.hlx.page/drafts/drashti/milo-banner/quick-links.
(Note: Since the mobile only functionality for this feature will be handled by MEP, kindly review the POC links in mobile view)

To resolve this issue, we are proposing two potential solutions:

1. Implement a frontend solution (e.g., a loading bar) to indicate to users that the redirection is in progress after clicking the quick link CTA.
   POC with a loader with an overlay: https://ecid-loader--milo--drashti1712.hlx.live/drafts/drashti/milo-banner/quick-links

2. Preload the privacy scripts on pages that contain quick links to avoid delays.
   POC with privacy script being preloaded in utils: https://ecid-privacy--milo--drashti1712.hlx.live/drafts/drashti/milo-banner/quick-links

I would appreciate your feedback and suggestions on these approaches, as well as thoughts on the best path forward.

cc: @salonijain3 @narcis-radu @dstrong23 @aishwaryamathuria @mokimo @suhjainadobe @bandana147 @robert-bogos @Ruchika4

[mokimo]

I'd discourage preloading. This makes the issue less visible, but it would still exist. It also might have a LCP impact. It's also unlikely that a user will instantly click a CTA right as he navigated to a page, so this sounds more like something that is flagged by QA/QEs but unlikely to degrade the experience of users on production.

I would suggest trying to go with a loader.

[overmyheadandbody]

I'd also suggest going down the loader path, most users will likely not click on a CTA before those couple of seconds. It might also be that the CTAs are not always in the first section, giving time for privacy to load. When they are in the first section, we certainly don't want to impact our LCP for this feature, since we have no guarantee that the user will click the CTA.

On the UI side of things, I'd suggest you vet this behavior with Consonant as well. Are we certain we want the loading screen covering the whole page? Could we just add an animated loading icon to the CTA itself? What happens should the ECID never become available, for whatever reason? We'd still like to have a usable page that users can interact with and not block any further interaction, right?

[mokimo]

I agree that something lightweight like a CSS loader within the CTA is likely the most un-intrusive while giving the user feedback.

[drashti1712]

Hi @overmyheadandbody and @mokimo

Thank you for your insights.
Regarding the loader approach, we would be getting design from the Consonant team, this is currently in progress. I just created a custom loader for POC purpose.
If ECID is not available or null, it would not be captured by Branch Analytics. The page in that case will be usable and working and the CTA would redirect to the AppStore normally. I will be adding these checks before raising the PR.

[rgclayton]

As you know, my vote is for the loader to let the site visitor know something happening after they click the CTA. As mentioned by Rares and Okan, this approach would help prevent any performance impacts preloading might have.

[Ruchika4]

+1 to using a loader instead of preloading privacy js file.