Initial questions from @Jdlrobson via email thread:

Some questions:

• Do we have open tasks tracking the work that would be needed in CentralNotice?
• Do we see an outcome of this work that QuickSurveys gets swallowed up by CentralNotice or depends on Central Notice (either hard dependency or soft dependency?)
• Last time I checked CentralNotice loaded a large amount of JavaScript (more than the existing mobile site together) - how would this impact us? Is there a small chunk of code we can load given we'd have to load this code on every page view?
• The work suggests integrating with the CentralNotice admin panel but also using CentralNotice for choosing the survey.
  • Are these two things tightly coupled as your write up suggests?
  • If the performance issue in the point above is still a problem could we safely make use of the admin panel but still maintain our own delegation mechanism?
  • This would help us with working around "A user can at max only ever see one banner (survey) from a campaign" too.

Another question

• Should the functionality of QuickSurveys be folded into the CentralNotice extension and Extension:QuickSurveys cease to exist?

In T125125#1979069, @Jdlrobson wrote:

Another question

• Should the functionality of QuickSurveys be folded into the CentralNotice extension and Extension:QuickSurveys cease to exist?

I can talk to that last one. I think almost all of the functionality could, and QuickSurveys could simply exist to reposition the survey on the page. CN banners always show up within the same div AFAIK, so we could use the extension simply for repositioning the survey and hiding the banner div. We could, of course, simply do this with CSS, but perhaps not without breaking page flow for accessibility purposes.

Hi all,

@Seddon is heading up the community product management of CentralNotice and I product manage the fundraising use cases. Seddon can weigh in on community priorities but I wanted to give some context.

I'm going a little high level first off. People approach us from time to time asking if CentralNotice can serve various research and testing purposes, like A/B testing features, controlling content on a page or surveys. All of the current teams and engineering leads around CentralNotice continue to be very much against expanding CN for these use cases. CN is a home grown, slightly fragile tool in desperate need of several overhauls. The current banner logic and targeting need constant attention so as not to impact performance (the performance team is very helpful and patient with us). When we test things like interactive banners and changes to how the content is displayed on wiki pages, we see major performance impacts. Adding any more functionality could really bloat CN and make untangling performance issues very difficult.

My opinions of these use cases:

• I think there should be a survey extension that can run on its own page or be embedded in a page using some A/B testing solution. Also making this an extension would help other mediawiki users.
• There needs to be some A/B testing solution that can test features, adjust wiki pages and serve things like surveys. I think the lack of this tool is why everyone looks at CentralNotice. We have talked to @MMiller_WMF about this and we are available to offer advice on this tool. CN could inspire a solution but we don't think it is the solution.

We conducted some in depth user research with CN admins and potential admins in 2017 [1]. This was a minor topic in interviews. We don't get requests for this that often. There really isn't any engineering capacity for CentralNotice to speak of. Fr-tech has a hard time just keeping it live and making fundraising changes. Seddon and I are making an appeal for a community engineer next year but that hire will likely work on the higher priorities from the research results [2] and any more recent issues Seddon might prioritize.

I (and I'm sure @Jseddon too) would be down to talk about this in greater detail in a meeting.

[1]
https://www.mediawiki.org/wiki/Extension:CentralNotice/Design_Research_2017

[2]
https://www.mediawiki.org/wiki/Extension:CentralNotice/Design_Research_2017/Results

I think a good person to talk to is @jlinehan, who is conducting a series of interviews right now to better understand how Audiences teams conduct instrumentation and A/B tests. CentralNotice and perspectives from Advancement and the communities might be good elements to add to the mix as the Better Use of Data effort thinks about what our future A/B testing frameworks should be like.

I (and I'm sure @Jseddon too) would be down to talk about this in greater detail in a meeting.

I'd be interested to attend. Separately, I'd also like to follow up on what kind of things you've had people ask if CN could do, go into the lessons learned running and maintaining the extension, and your thoughts about a future A/B testing solution. Gory engineering detail could be helpful. I can follow up off-ticket since this is not directly relevant to the task.

Putting a neutral hat on, I completely understand the fundraising concerns. We might also want to think about this in the other direction - improve QuickSurveys so it can eventually run banner campaigns. FWIW I don't have any opinions or recommendations either way.

Possibly, there's an opportunity to generalise the UI CentralNotice uses to setup banner campaigns so that QuickSurveys could use it. Either way we're talking about a lot of effort here and without this being a quarterly goal for either/both team I can't see it happening.