Hi Flo, It would be really helpful if you let us know in this thread (whenever you're ready for it and if the answer to your questions is not already in the literature) what kind of research or analysis would be helpful for you. Something we would like to start exploring in the Research team is to convert this kind of question to a well-documented open question type and offer it to volunteer researchers who come to us and ask for what questions to work on. I have a few questions on this front which may help you expand from there (or negate and tell me to discard them as not-interesting): * What is the impact of MassMessage on activation (where activation needs to be defined but broadly speaking, it's bringing back the user after a period of inactivity to do at least n edits)? * Is the impact on activation dependent on the topic of the MassMessage? For example, does a MassMessage on a call for elections activate the users the same way as the MassMessage to congratulate Wiki Loves Monuments winners? * Is the impact on activation wiki dependent? For example, are Wikimedia Commons editors more likely to get activated after receiving a MassMessage than Wikidata editors? * Is the impact on activation related to the tenure of the user, where tenure needs to be defined but roughly speaking: it's the number of years since account registration? And finally, we have some learnings from emailing users which may be helpful for you to know as some of them may apply to MassMessage, too, though we can only hypothesize at this point: * On Wikipedia: the users are more likely to respond to a call to action (in this case: create an article) if they have recently edited Wikipedia. Check Figure 10 in https://arxiv.org/pdf/1604.03235.pdf . The more there is time in between the last edit time and the time the users received the email, the less likely they are to respond to the call to action. * On Wikipedia: Personalization matters. The editors who received personalized messages (in this case, it really means personalized recommendations) started to contribute to articles twice more than those who received important articles as part of their recommendations that where not personalized. I want to emphasize that the set-up of the experiment was very specific, but there may be some learnings here for what you have in mind. Best, Leila On Mon, Nov 26, 2018 at 10:02 AM Florence Devouard <fdevouard(a)gmail.com> wrote: