Talk:Social media

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

Re: Twitter engagement questions[edit]

In response to this message on Wikimedia-l regarding engagement rates on Twitter, I would like to share some of the following resources that the Foundation uses to benchmark the metrics of the @Wikipedia account compared to industry standards. There are several resources available, and each varies slightly in its recommendations.

Rival IQ is a social media marketing analytics company that has developed social media industry benchmark reports for several years. According to its 2022 report, the median Twitter engagement rate for brands across all industries is 0.037%; for nonprofits specifically, it is 0.054%. According to Statista, which specializes in market and consumer data, the average engagement rate on Twitter across industries was 0.04%. According to Adobe, "most would consider 0.5% to be a good engagement rate for Twitter, with anything above 1% great." Wikipedia Twitter's engagement rate, according to the dashboard we access when logged-in to the account, over the last 28 day period is 2.7% (see here for a screengrab of these analytics). In March, April, May, and June, it was 2.1%, 2.4%, 2.6%, and 2.2%, respectively. Given the data I shared above, we consider this performance to be above industry standards.

I hope that these are helpful resources as you continue management of the @euwikipedia account. Please note that Twitter defines "engagement rate" as the number of engagements (clicks, retweets, replies, follows, and likes) divided by the total number of impressions (number of times users saw the tweet on Twitter). It is difficult to draw direct comparisons between the @euwikipedia and @wikipedia accounts due to the difference in follower size and our more global focus, as well as the objectives we are prioritizing to support the movement but also build resonance among groups who can help us to push forward our knowledge equity goals. At the same time, a straight comparison—with the understanding that I cannot see the analytics for the @euwikipedia account—reveals more retweets, likes and comments on the Wikipedia account. I'd like to better understand however if we are defining engagement differently. Also, an overall higher engagement rate from Twitter's analytics could be a result of the low base effect (comparing two accounts of different sizes).  

Again, I hope all of this is helpful, and I am happy to continue the conversation here when I am back online on 25 July. Thanks! LDickinson (WMF) (talk) 17:46, 18 July 2022 (UTC)Reply[reply]

Thanks again for the answer. The problem I see with engagement here is that you are using two different measures at the same time for comparison. Rival IQ defines engagement as interactions per follower, and Twitter defines it as interactions per impressions. So we would need to analyze the engagement per followers (630.000) and not by impressions, that I'm sure are way less. Could we know the number of interactions per month so we can calculate the engagement using the Rival IQ standard? Theklan (talk) 18:17, 18 July 2022 (UTC)Reply[reply]

For the calculation using @euwikipedia we had in the 17 days gone in July 1,007 interactions for 89 tweets and 7,600 followers. Is to say, 0,149% of engagement following RivalIQ calculation. Is three times the industry standard for non profit, a good ratio (but July has been a bad month, because of holidays) Theklan (talk) 19:28, 18 July 2022 (UTC)Reply[reply]

I think we can get into a long back and forth on what engagement tools are most telling and accurate and what they say about the respective accounts. Ultimately, though, I'd like to go back to the spirit of your initial inquiry, which was about "centering free knowledge" and our common goal of showcasing the movement in the best way possible. On the Wikipedia account, we aim to do that by highlighting the breadth of the movement's work, values, and identity on a global scale to keep and attract new supporters. We welcome ideas for content from you and other community members so we can do better and strengthen our global focus. Striking the right balance between being inclusive, relevant, and values-aligned is what we strive for, though it's not always easy. Our strategy recognizes that there are people who are not yet part of our movement that must be if we are to achieve our knowledge equity goals. That means that a tweet about American actress Betty White may get more engagement than one about Burundian runner Francine Niyonsaba. And that is okay—because we prioritize inclusion and still send a strong message about the diversity of our content and our movement, so that we can attract new interest, particularly among those who are not yet well represented in our projects. LDickinson (WMF) (talk) 18:09, 25 July 2022 (UTC)Reply[reply]

Thanks again for the answer. We share the same goal, but the way to achieve it seems different. I understand by your answer that you have understood that you are measuring engagement wrongly. A good way to improve it is to have more impressions. We don't know the exact details of the Twitter algorithm, but it seems clear that tweeting more gives more impressions and more engagement. Engagement also rises the number of impressions of the next tweet. We don't know the exact formula, but the ingredients are those. We can work on getting more engagement, but if we obscure the numbers once we know that you are using impressions instead of followers abd selling it as a success, then we can't do anything to improve the numbers. And, trust me, those numbers have a huge space for improvement.

Can we have the number of impressions for the tweets in the last 28 days, so we can all know where we are? Thanks. Theklan (talk) 20:01, 25 July 2022 (UTC)Reply[reply]

The engagement ratio for the Catalan Wikipedia (@Viquipèdia) account, the 4th biggest worldwide which is managed by two community volunteers, with 7-10 tweets a day of 23 weekly content topics:
  • Last 28 days, end June until now: (1900 clicks to articles + 2300 RT + 4500 likes + 129 replies + 841 cited tweets) / (244 tweets · 40710 followers) = 0.973
  • 28 days of February 2022: (1900 clicks + 1800 RT + 5400 likes + 111 replies + 703 cited tweets) / (258 tweets · 40482 followers) = 0.949
At the cost of zero €. More than 65% of tweed bios are women. Weekly DMs and mentorship of new editors that we derive to the Village Pumps with first editing steps. Far from the widely blamable US-bias, we programme 30% Catalan-culture focus instead (not only Barcelona or Catalonia: all territories) by tweeting in all different, academically official accents of the language. Tweets on articles to improve, featured articles, and also articles to be translated. The profile has been featured several times in the press as one of the most fresh, active and interesting Twitter accounts of our linguistical domain. We have lots of data to report and inner style/discourse protocols. One year ago, the WMF Comms office disregarded our request for some institutional help to get the account verified with a 2-month delay and 1-line email reply. Xavier Dengra (MESSAGES) 07:52, 26 July 2022 (UTC)Reply[reply]
I think you're right—we have a different view on the way to achieve our goal. That's why sharing the numbers won't help. As I tried to illustrate with the examples I provided in my last response, we aim to draw in specific audiences, even if that process is methodical and gradual. In that vein, these are the numbers we are most focused on and believe to also be accurate for our needs. As I said previously, we'll be discussing our refreshed digital strategy more in detail with ComCom in the coming months. That will allow us to explore our focus more deeply with a larger group of movement representatives. Based on the tone and circular nature of this thread, I think it would be best to leave things at a respectful agreement to continue to pursue what is best for these different and unique audiences and goals. LDickinson (WMF) (talk) 20:43, 26 July 2022 (UTC)Reply[reply]
Sorry Lauren, but this is not fair. I made a question, you answered with some numbers (because they seemed inportant to you), I showed you that you were using two different measures for the same thing (wrongly) and then it seems that numbers and measurements are no longer important.
If something is important for the credibility of Wikimedia is that we check facts and numbers, and we have [citation needed] templates whenever the data is doubtful. I would like to know how many interactions WE had in the last month, because measuring and sources are relevant. Thanks. Theklan (talk) 21:31, 26 July 2022 (UTC)Reply[reply]
I'm a bit worried about the lack of answer here, especially since we are talking about a communications department. Nevertheless, I have downloaded the engagements for July, as we are entering August. We had a total of 4573 engagements for 141 tweets. This makes 0.425 engagement, is to say, half of the engagement presented by @Xavier Dengra, but 7.87 times higher than the industry standard you proposed as a benchmark. Can we have the numbers for @ wikipedia, please? Theklan (talk) 08:55, 1 August 2022 (UTC)Reply[reply]
We are monitoring this page for new questions and comments with which we can engage constructively and offer new support and information. Regarding your request, we have already established and discussed that there are key differences in our social media strategies and goals, that we seem to use different metrics to measure engagement, and that we will be sharing our refreshed strategy in the coming months with ComCom. Please see my earlier comments for reference: Mainly, "Based on the tone and circular nature of this thread, I think it would be best to leave things at a respectful agreement to continue to pursue what is best for these different and unique audiences and goals." LDickinson (WMF) (talk) 21:02, 1 August 2022 (UTC)Reply[reply]
Archiving a topic that hasn't been answered seems weird to me. I'm still waiting for the metrics. Theklan (talk) 22:42, 4 January 2023 (UTC)Reply[reply]

Mastodon Social Account?[edit]

Mastodon is a free software micro blogging implementation. I think it would be awesome if Wikimedia mirrored their tweets to Mastodon in support of the free software aspect. 18:45, 24 September 2017 (UTC)

Yes, but how? If you're interested, please fix [1] and I'll start updating [2]. --Nemo 14:25, 1 October 2017 (UTC)Reply[reply]
Update: it's very easy now with and similar. Initiatives on our topics, like those by FSFE, tend to reach wider audiences and have more engagement on the fediverse than on the mainstream social media, see for instance the statistics on #iloveFS where it's basically a tie:
We've been running for a while and it regularly happens that some toot produces more engagement than the corresponding tweet. Nemo 11:10, 22 February 2020 (UTC)Reply[reply]
@Nemo bis would you be able to help document and track this in Wikidata or in some other way? I am really interested in following networks and spectrum of different use cases. Please check the request I just made --Zblace (talk) 06:22, 17 July 2022 (UTC)Reply[reply]
Hello, any plans of Communication Team using Mastodon in future? Thanks. --SCP-2000 15:59, 17 December 2022 (UTC)Reply[reply]
Hi @SCP-2000. Thank you for asking. The Digital Communications team has been researching Mastodon and considering our potential involvement with the platform in the future. At this time, we have no plans to create an account for the Foundation or Wikipedia. This is mainly because our observations show us that Mastodon is not yet reaching a large audience, which is one of the key objectives of our communications activity on social media. We will continue to monitor the situation and adjust our recommendations and practices to keep within our objectives. Thank you! LPasqual (WMF) (talk) 22:29, 19 December 2022 (UTC)Reply[reply]
Dear Lucas, this is a disappointing response considering all that has happened so far. I have responded over on the wikimedia-l list in some more detail, and hope Wikimedia Foundation will reconsider its stance on the matter soon.--Eloquence (talk) 22:35, 30 December 2022 (UTC)Reply[reply]