MetaFilter Usage Statistics August 17, 2022 7:46 AM   Subscribe

In the Steering Committee voting thread (please vote!), I posted some data analysis how much people are using MetaFilter, based on the data in the infodump. That became somewhat of a derail, so here is a thread to discuss what this data means, and how it should inform our decisions going forward.

The charts I made are cohort analyses by month and year. The X axis is time, the Y axis is number of active users (where "active" means they posted post, posted a comment, or favorited a post or comment), and the colors represent groups of users that joined during the same month or year.

Additionally, here is a table showing relative number of users active each year. The Y axis is time, and the X axis is the cohort year, so, for instance, the bottom right cell being "58" means that 58% of users that joined in 2021 were active in 2022. (Here is a version that is shifted so that the y axis is relative to join date) You can see that it used to be that ~70% of people stayed for a year, then ~80%-85% of the people who remained stayed for a second year, but these days it's more like 60% staying for the first year, then 70%-80% for a second year. I personally think that figuring out why so many more people than before are bouncing off the site is important — to me, it seems easier and more impactful to figure out why people who are interested enough to join MetaFilter end up bouncing off of it than to figure out why people aren't hearing about MetaFilter.

However, we'll definitely need to do both, since the total number of registered users has been quite stagnant, compared to what it once was.
(click to expand)

Here is the number of users who joined each year, for reference
   2695 2000
   9886 2001
   3157 2002
     34 2003
   2895 2004
   4591 2005
   5588 2006
   5515 2007
   5965 2008
   5536 2009
   5173 2010
   4694 2011
   4180 2012
   2800 2013
   3140 2014
   2337 2015
   1860 2016
   1485 2017
   1180 2018
    899 2019
    762 2020
    587 2021
    393 2022


Additionally, here are some more charts from a couple years ago, and some data on AskMe posts per day.

I hope that this data will allow us to take better stock of where we are, and better understand how the userbase is changing.

I'll close this post by saying that it's a mistake to think that just because something is easy to measure, it's also important, and I would rather see MetaFilter be high-quality and low-volume than low-quality and high-volume. However, I think right now the size and (lack of) diversity in the userbase is a significant problem for both the quality of discussion and the financial stability of the site, and it's thus important to find fresh faces and convince them to stick around.
posted by wesleyac to MetaFilter-Related at 7:46 AM (67 comments total) 18 users marked this as a favorite

In the other thread Miko asked for a chart excluding favoriteshere's that chart. I can do a annual version as well if people want, but it's a little more work and I think this one gets the idea across.

Interestingly, the chart without favorites shows pretty clearly that there hasn't been a decline in users posting or commenting for the past ~year, which seems like good news :)
posted by wesleyac at 8:23 AM on August 17, 2022 [10 favorites]


I really appreciate this data, wesleyac!
posted by It's Raining Florence Henderson at 8:26 AM on August 17, 2022 [4 favorites]


Interestingly, the chart without favorites shows pretty clearly that there hasn't been a decline in users posting or commenting for the past ~year, which seems like good news :)

That is actually pretty interesting, yeah. Thanks for doing all this, wesleyac. Next time we get out on the canals of Amsterdam (there will be a next time, right???) you get all the snacks.
posted by lazaruslong at 8:32 AM on August 17, 2022 [2 favorites]


And now, for something slightly different — the number of comments per user during July 2022, and the same thing on a log scale. Each bar has width 1, so there were ~1100 users who made one comment, ~500 who made two comments, etc.
posted by wesleyac at 9:36 AM on August 17, 2022 [2 favorites]


I was initially confused by the retention graphs because I was expecting them to be absolute percentages of the users from, say, 2000, who were still active in 2022, rather than YoY changes in number of active users who'd joined in that era. The numbers over 100% is what tipped me off.

But wait I just noticed that 110% of the people who joined in 2004 were active in 2005. I take it back, I still don't understand those charts.
posted by aubilenon at 9:56 AM on August 17, 2022


Oh I guess that might be people who joined in 2004 but only lurked until 2005. Okay, I'm fine.
posted by aubilenon at 9:57 AM on August 17, 2022


(Reminder that people open and close accounts and someone like me, who joined in 2001, closed that account and opened this one in modern times after a very short break, aren't clearly represented here - probably a small percentage but this is a little bit blurry of a snapshot, is all).
posted by tiny frying pan at 10:02 AM on August 17, 2022 [3 favorites]


aubilenon, ah I didn't explain it well. Each cell is the number of users in that cohort year who were active that year, divided by the number that were active the previous year. This means that people who create a account but never use it aren't counted. You're correct about numbers over 100%, for instance, of the 2895 users who joined in 2004, 1934 were active that year, and 2244 were active in 2005. Presumably that's because signups were re-opened in November of 2004, so a lot of people joined and didn't do anything for just a couple months, and then got counted as being inactive that entire year.

If you're interested, here is one where each cell is the number of users in each cohort who were active in a given year divided by the number of users in their cohort year, which is I think what you were expecting it to be at first. (and a normalized version of that)
posted by wesleyac at 10:25 AM on August 17, 2022 [1 favorite]


Interestingly, the chart without favorites shows pretty clearly that there hasn't been a decline in users posting or commenting for the past ~year, which seems like good news :)

Admittedly I'm eyeballing the numbers from your graphs, but if we look at just the last two years or so, aren't registered users increasing while active users are decreasing? If so, doesn't that mean that the percentage of registered users who are active is decreasing?
posted by NotMyselfRightNow at 10:25 AM on August 17, 2022


Good question, here's a chart. I'd say active users (not counting faving, just posts and comments) are staying about the same (looking just at the past ~year), and while registered users in increasing, it's not by a large absolute value (the number of total users grew by ~0.7% last year).

But zooming out, I don't think it's a problem if this percentage goes down. As MetaFilter gets older, people will drift away, and the percentage of users who are active will trend towards zero. That's normal and expected — we should care more about whether, on the balance, we're growing, shrinking, or staying the same size. Right now it seems we're staying around the same size, if you just look at the number of users who are posting and commenting regularly.
posted by wesleyac at 10:53 AM on August 17, 2022 [1 favorite]


"Registered users" is not ACTIVE registered users? So I, for example, am on my third account: that's two registered users who are not active.

Active users flame out, give up, or get booted. Then they miss Metafilter so set up a new account. So the usage can be stable even as the number of registered users slowly climbs.
posted by one more day at 2:15 PM on August 17, 2022 [1 favorite]


one more day, yes that's correct — I made the chart since NotMyselfRightNow asked for it and since I think it's somewhat interesting in and of itself, but I agree that it's not a particularly useful metric, hence my second paragraph :)
posted by wesleyac at 2:32 PM on August 17, 2022


Probably not for this thread (and wesleyac has already done a ton of work for free here), but I'd be really interested to see a deeper dive into the patterns behind retention. Is there some commonality to people who stay vs. churn (which subsites they first participated in, which subsites they eventually participated in, how much they comment vs. post, frequency of engagement, etc.). Are people who get more favorites on their comments more likely to stay (kinda joking about that one, kinda not)?
posted by primethyme at 2:51 PM on August 17, 2022 [2 favorites]


Is their last comment on MetaTalk? heh.
posted by Rumple at 3:10 PM on August 17, 2022 [13 favorites]


(I did think of that but chose to not mention it!)
posted by primethyme at 3:14 PM on August 17, 2022 [3 favorites]


Those numbers are depressing. Do we have Google analytics or equivalent numbers on page views/unique visitors from the same time period?
posted by geoff. at 3:30 PM on August 17, 2022 [1 favorite]


Thanks for all this work, wesleyac. Really interesting!
posted by Miko at 3:37 PM on August 17, 2022 [1 favorite]


geoff, my understanding is that we have Matomo, but I don't know who has access to the data. Presumably it will be available to SC members, at least?

However, my adblocker blocks it and probably so does a lot of other people's so it's unclear to me how useful the data will actually be. It is possible to configure it to be significantly less likely to get blocked, if we're interested in the data and that is something that users do not find morally objectionable.
posted by wesleyac at 3:40 PM on August 17, 2022 [1 favorite]


I doubt they have it going back that far, but IIS/w3svc logs don't lie. Hopefully there was a dump to an HDFS or equivalent or maybe an aggregate.
posted by geoff. at 3:42 PM on August 17, 2022


Much appreciated!
posted by snuffleupagus at 4:07 PM on August 17, 2022

we should care more about whether, on the balance, we're growing, shrinking, or staying the same size
I would say yes, kind of, but these conversations seem always to start with the assumption that growth in numbers is positive, and shrinking undesirable, which I've always hesitated at. When I look at a lot of the curves in wesleyac's charts which have either a peak or an inflection point around 2010--2012, I interpret it as a broader shift of internet use (in the USA and worldwide) toward social media, and away from websites/blogs.

Twitter has grown colossally in the years between 2012 and 2022. Whether that's been good for Twitter, for its users, or society at large, is an open and unanswerable question.
posted by Fiasco da Gama at 6:32 PM on August 17, 2022 [8 favorites]


Via basic math, if the site needs to earn more money* (to pay mods, cover hosting costs, pay accountants, whatever), it's going to either need to get more users or it's going to need to get the existing users to pay more (and/or more of the existing users to pay). So if we decide that growth is not the path, what is the suggestion for bringing in more money per user?

* If this is an incorrect assumption on my part, then the question is moot, of course.
posted by primethyme at 6:39 PM on August 17, 2022 [8 favorites]


Presumably it will be available to SC members, at least?

I don't personally know what we have for logs, but I know frimble does. While there are some open questions about the remit of the SC, I don't think it's a stretch to say they will likely have access to site metric types of data.
posted by jessamyn (staff) at 8:06 PM on August 17, 2022


When I look at a lot of the curves in wesleyac's charts which have either a peak or an inflection point around 2010--2012, I interpret it as a broader shift of internet use (in the USA and worldwide) toward social media, and away from websites/blogs. posted by Fiasco da Gama at 6:32 PM

I completely agree with this. To me, there are two sides: internal "Metafilter" factors, and external "Internet" trends.
Re the internal factors, i think awareness is high of issues, and investment to address them is highly visible in all the current metatalk threads.
Re external factors, i feel they are at times underrepresented in this discourse, especially that the Internet (for lack of a better word) has become highly commercialised in the last ten years, driven by data harvesting and fuelling and fuelled by a global industry of social media.

Personally i would be sad to see a change towards a more commercial approach, eg. using user analysis tools, quantifying activity, retainment, etc. To me, one of the most attractive features of metafilter was and is that no one is tracking my involvement and linking it up to my demographic in order to grow. Personally i would be willing to pay more per month instead.

I worry this current drive to quantify usage, analyse demographics, etc will change the site very much. but then this is a very sentimental approach and perhaps not feasable.

My own use and enjoyment of metafilter i feel is not reflected by statistics of how often i post, comment or favourite. My activity waxes and wanes, through factors that are often not at all related to metafilter as such but circumstances of my life. And If related to metafilter, Not for reasons easily explained and easily misunderstood.
posted by 15L06 at 2:01 AM on August 18, 2022 [10 favorites]


My own use and enjoyment of metafilter i feel is not reflected by statistics of how often i post, comment or favourite.

Right? We have a number of active users and then we have a number of lurkers who are here (hi!) who engage primarily if not exclusively by reading. I have a friend who buttoned years ago and still knows more about the site goings-on than I do!

From the survey, we know a lot of these lurkers (hi!) say they lurk because they're just lurkey lurkers by nature and that's just their preference, nothing personal. And then we had this other huge group of lurkers (hello!), who say that they comment infrequently or never at all out of genuine concern of saying something wrong, or feeling intimidated, or feeling like the conversation has already passed them by, or even just keeping what they feel is a healthy distance due to past interactions. So there's this whole... metafilter ghost contingent... (hellooooo!).... that's here with us but maybe not showing up in the stats. I wonder if there's a way to accurately capture that.

Anyway, I'm fascinated by these stats and everything you've put together for us, wesleyac. Thank you so much for doing this!
posted by mochapickle at 3:19 AM on August 18, 2022 [18 favorites]


The survey was a fairly small representation and self selecting. We really need visitor data in the last 4-5 years to see if people are really lurking, or the long-tail of AskMe brings in a lot of non-recurring users. Without that data we can't really know anything other than we won't get any new users sometime mid-2023 if the trendline continues.

Hopefully, visitor data is encouraging which means we just need to convert visitors to users which is much more straightforward of a task.
posted by geoff. at 4:09 AM on August 18, 2022


Without that data we can't really know anything other than we won't get any new users sometime mid-2023 if the trendline continues.

What?
posted by mochapickle at 4:12 AM on August 18, 2022 [7 favorites]


If you take the data, "Here is the number of users who joined each year, for reference" for sake of convenience call it Table 1, and then chart a simple trendline it goes to 0 in 2023. Or is there a better way of doing that or is my method just wrong?
posted by geoff. at 4:15 AM on August 18, 2022


Your method is just wrong. :)

Thanks!
posted by mochapickle at 4:18 AM on August 18, 2022 [3 favorites]


How would you calculate it?
posted by geoff. at 4:23 AM on August 18, 2022


I don't think trendlines are very useful, and I probably wouldn't use a linear trendline for this data if I was using one, but one thing that might be tripping you up is that 2022 is only partway over. A reasonable projected number of signups for 2022 would be ~551 (since that data is from ~August 7th IIRC, it still has ~40% of the year remaining). Probably I shouldn't have included the 2022 number to avoid that confusion.

Given that, it seems to me like new user signups are fairly correlated with site activity — declining from ~2012-2021, but relatively stable for the past ~year.

(It's possible that signups peak at the beginning or end of the year, haven't looked, but I don't think it's super important)
posted by wesleyac at 4:34 AM on August 18, 2022 [3 favorites]


How would you calculate it?

Well, for starters, I wouldn't throw out findings from the 900+ survey responses of people telling us about their lived experiences here on Metafilter for being self-selecting.

But geoff., I don't mean anything personal by this and I sincerely wish you well because I think you have a lot of good ideas, but it seems like too many things I've commented on or said to you recently have ended up on reddit and that's just... weird. So I'm going to limit my engagement with you.
posted by mochapickle at 4:35 AM on August 18, 2022 [12 favorites]


That's fair, I respect that and I apologize if I offended you.
posted by geoff. at 4:37 AM on August 18, 2022 [1 favorite]


Maybe I missed this but what do the colors in the graphs represent?
posted by mon_petit_ordinateur at 2:09 PM on August 18, 2022


Each color represents the group of users that joined that month (or year, as the case may be). So, for instance, the thick grey line at the bottom of this chart are all the users who joined the month that signups reopened in 2004.
posted by wesleyac at 3:55 PM on August 18, 2022 [1 favorite]


Does anyone know off-hand when the catastrophic Google algorithm change was? Was that 2012 (surely not...)?
posted by hoyland at 5:03 PM on August 18, 2022


10/12

'Why has Google forsaken MetaFilter?'' Slate, 2014.
posted by clavdivs at 5:14 PM on August 18, 2022


Yeah, when wesleyac posted the chart in the other MeTa, I was promoted to go back and search for the thread Matt had posted about the state of the site. He posted it in 2014, but referenced the change happening about a year and a half prior. I had forgotten until re-reading that thread and these charts just how dramatic the hit was—Matt stated in 2014 that traffic dropped 40% due to Google’s shenanigans. That’s a lot, and I think a large part of what we’re seeing is just due to just not getting that continual influx of eyeballs on the site (particularly AskMe). I know the charts feel kinda dire, but I was actually surprised the drop in activity wasn’t more precipitous. It’s just been a long, slow decline.
posted by DiscourseMarker at 5:55 PM on August 18, 2022 [5 favorites]


It’s amusing how the $5 N00bs stick out like a giant strata of reef limestone across that entire first “by month” chart. We are a persistent cohort if nothing else.

Also, looking at the “without favorites” version, the taper appears to have stopped in 2021 and entered more of a steady state phase. This is maybe good news
posted by Devils Rancher at 8:35 PM on August 18, 2022


I feel like my personal involvement in the site really dropped off a cliff with the death of the megathreads and I assume that I'm not the only one. I realize that they're not coming back but that was a choice made by management that probably drove a lot of us away.
posted by octothorpe at 5:22 AM on August 19, 2022 [1 favorite]


It’s amusing how the $5 N00bs stick out like a giant strata of reef limestone

My people were waiting a long time to get in, and it was a big deal for them to become part of the community!
posted by Meatbomb at 5:39 AM on August 19, 2022 [3 favorites]


We were hitting that "Sign Up" button every day for years hoping that new memberships would open up.
posted by octothorpe at 6:07 AM on August 19, 2022 [3 favorites]


Building on what DiscourseMarker said about not getting the influx of eyeballs from Google that we used to, I think there are three interrelated but distinct shifts from the halcyon "build it and the revenue will come" days of the 2000's web that threaten Metafilter's long-term stability:

Firstly, ad rates have dropped precipitously across the board, and show no sign of recovering. It's just no longer true that you can build a serious organization on the back of a somewhat-popular website. Short of setting up our own ad network, there's nothing Metafilter can do about this.

Secondly, search engines generally and Google in specific no longer value Ask the way they used to, partly because of changes to the ineffable algorithm, and partly just because there are a lot more sources of information on the web today than there were 10 or 15 years ago. Metafilter could probably do some SEO to goose these numbers a little bit (and it would be great for someone to run some numbers to see if this is the sort of thing that might pay for itself through a modest bump in ad revenue) but getting back to the sort of search-engine driven random visitor traffic that Metafilter had ten years ago seems wildly unlikely to happen.

Thirdly, some percentage of those search engine randos decided to stick around and become posters, commenters and/or supporters. As Metafilter becomes more and more user-supported, this seems like the real existential threat to the site. People who button after big policy changes or massive arguments in MetaTalk are the most visible leavers, but it's actually hard to see any of those events in the data. The real threat to Metafilter is the general entropy of any community, where every month some percentage of active users just drift away, and I'm not seeing any evidence in the data that that's being driven by anything in particular that Metafilter is doing or not doing. A community the size of Metafilter just needs to attract a certain number (1000? 2000? 500?) of new users every year to replace those who leave, and for the last 8 years or so, Metafilter hasn't done that.

I guess if there's anything that I'm taking away from this very useful data set, it's that any strategy for ensuring long-term stability that focuses on keeping existing users over and above attracting new users or changing the model of what Metafilter is seems highly likely to commit us to a long, slow decline as the best case scenario.
posted by firechicago at 6:24 AM on August 19, 2022 [4 favorites]


firechicago — thanks for that insightful comment — I agree with much of it. However, I'm curious if you saw this comment of mine in the other thread.

I decided to dig a bit deeper into the models I was making to try to make them more accurate. First, two charts: this one is the percentage of users who joined in given year that were active N years after they joined — a chart version of the second table in this comment. Other than 2000-2004, the curves there are remarkably consistent, it's mostly just the percentage of users who are active on the year that they join that changes. This is clearer if you look at this chart, which is just the previous one but with all the lines shifted up to start at 100%, just to give a better picture of the curve.

With that, I went and made a more detailed model, based on a fixed number of users joining per year and falling off according to the average of those curves from 2005-2021 (linearly extrapolating for users that are around more than 18 years). It gives a predicted steady-state number of users for the 2021 number of new users and rate of newly signed up users participating as ~1350.

If we moved the rate of users who are active after signing up for an account from the 2021 rate of 55% up by 10% to 65% (where it was from ~2003-2013) while keeping the number of new signups steady, that would result in a steady-state of ~2378 users. Getting that same increase from attracting new users would require getting 1340 new users per year, 228% of the current number of signups. It seems much much easier to me to convince 10% of people who have already signed up for an account to be active than it does to convince 753 more people than are already joining to join, every single year.

I do want to note that there are legitimate reasons to lurk, and if that's what people want to do, I absolutely support that. Lurking is great, and it's what I do in a lot of places on the internet. However, I suspect that many of the 287 people who signed up for an account last year and then didn't post, comment, or favorite anything are doing so not because they want to lurk, but because there are social or technical barriers in the way of them participating.

Before doing this analysis, I didn't realize how huge the cohort of people who register for a account but don't use it are. Especially given how annoying registering for an account is on Metafilter compared to the rest of the internet, I think it should be very high priority to reach out to these users and figure out why they aren't participating.
posted by wesleyac at 7:35 AM on August 19, 2022 [6 favorites]


I realize that they're not coming back but that was a choice made by management that probably drove a lot of us away.

Question: do you really feel like it drove people away to close the megathreads? I'm not nitpicking for the heck of it. :) Genuinely curious how you or others experienced the policy change. I would've guessed that a word like "disengaged" or "drifted" or something would be more accurate, but recognize that's colored by how I experienced those threads, and that's not universal.
posted by curious nu at 7:40 AM on August 19, 2022 [3 favorites]


My people were waiting a long time to get in,

Meatbomb, site lore is you are the very first 5$ noob? Is this true?
posted by Rumple at 8:24 AM on August 19, 2022 [1 favorite]


Wesleyac -

Thanks for pointing me to that, I think I had seen it but didn't have it in mind when I was writing my comment. I hadn't realized that the drop-off in the second year had changed so much over time. I guess if I had looked more carefully at the data and had your analysis more clearly in mind I would have rephrased "any strategy ... that focuses on keeping existing users" to "any strategy ... that focuses on keeping long-time users". I agree with you that there does seem to be some room for improvement and potentially a big impact from doing a better job of convincing a larger percentage of people who sign up and start participating to get comfy and stay a while.
posted by firechicago at 8:56 AM on August 19, 2022 [1 favorite]


Wesleyac, I bet you've already accounted for this, but just to double-check: the numbers you're using to determine new account sign-ups, does that include everyone who started the process (and thus reserved a user number), or just those that made it all the way through to the end (i.e., paid $5 and generated a new user page)?
posted by nobody at 9:07 AM on August 19, 2022


The user data comes from the list of users in the infodump, which the wiki tells me are users that have completed the signup process. My understanding is that "completing the signup process" entails paying the $5 or going through the fee waiver process, but I don't know what the precise definition is.
posted by wesleyac at 9:15 AM on August 19, 2022


Ah, perfect, thanks!
posted by nobody at 9:35 AM on August 19, 2022


Before doing this analysis, I didn't realize how huge the cohort of people who register for a account but don't use it are.
To a certain extent though this is true of all online communities, and always has been. It’s also true of other parts of the internet, like free online education (i.e. MOOCs). Vastly more people sign up for things than ever end up doing anything with the thing they sign up for. Similarly, churn is also a regular and recurring pattern for online communities generally.

None of this is to say that we shouldn’t still try to engage more of the existing user base, but I agree with firechicago that any strategy also has to account for attracting new users. And of course the tricky bit is that strategies to engage existing users don’t always neatly overlap with ones that bring in new eyeballs.
posted by DiscourseMarker at 9:50 AM on August 19, 2022 [1 favorite]


Meatbomb, site lore is you are the very first 5$ noob? Is this true?

Please see full accounting on my profile page. (short answer: not exactly, but pretty close)
posted by Meatbomb at 10:19 AM on August 19, 2022 [2 favorites]


Google's making a lot of fuss about their upcoming ranking algorithm change boosting "helpful content made by, and for, people". I've done the dance trying to divine their intent from their search-rater instruction guide before, and there's always a *lot* of "maybe they mean this? But also that?" in there, but "helpful content made by, and for, people" could practically be a slogan for AskMe.

So on the above topic of great SEO Spurning of Metafilter, here's hoping this brings fresh winds to us.
posted by CrystalDave at 1:31 PM on August 19, 2022 [3 favorites]


Google being both the source of traffic and the seller of advertising, I wouldn’t be surprised if metafilter would do better in search if there were more adsense ads on the pages. But I also don’t think it would ever come back as the same magnitude of revenue.

It seems much much easier to me to convince 10% of people who have already signed up for an account to be active than it does to convince 753 more people than are already joining to join, every single year.

I would normally agree with this statement, but the site also has a $5 signup fee (that maybe can be waived now, though I’m not sure of the process). I think it is hard to guess how many people might sign up if the process was free and as easy as possible. No matter what, I don’t think it is an either-or situation. Sustaining the site will require both things.
posted by snofoam at 6:28 PM on August 19, 2022


My limited understanding is that with many activities, more people sign up for things than actually do them. For example, consider unused gym memberships and exercise equipment. It's not just online communities. I've done it myself.
posted by NotLost at 9:29 PM on August 19, 2022 [1 favorite]


DiscourseMarker and NotLost bring up that some number of people signing up and not using their accounts is to be expected. I agree with that. My concern stems from the fact that the rate of people doing that has been consistently increasing — here's a chart of that. If you don't think the 2021 rate of 49% of people who sign up not using their accounts is a problem — what does that number need to get to for it to start being a problem?

I agree that we do need to be thinking about attracting new users as well, I just think we've historically done very little to try improve the retention rate of those new users, so there's a lot of low-hanging fruit there that will be much more impactful than the things we can do in order to attract new users, at first.
posted by wesleyac at 6:38 AM on August 20, 2022 [1 favorite]


So again my first reaction to that new chart (and really, thank you for all of this work as the data really are fascinating), is that 49% really isn’t that bad. If anything, it makes me wonder if MeFi’s earlier years weren’t the statistical outliers, compared to other parts of the web. But we’d need data from other sites to compare that to, and like I don’t think big commercial entities are going to be chomping at the bit to publicize how many of their user accounts are dormant. That’s why I was comparing to free online education earlier, because there are lots of published articles.

But anyway…what this is making me think though is what do we even mean by “use” their accounts? We are using posts, comments, favorites, as proxies for engagement because those are the only things we can see and count. Maybe there are some % of those dormant users who actually do read the site frequently, maybe even logged in, but they don’t click stuff. Maybe they just don’t like posting but they like not seeing ads? Or maybe they all did just never come back. We do know from the TT survey that there are some true lurkers, but it’s not possible to say what the proportion is. But that is also another recurring pattern with online communities—the proportion of the total user base that is “active” in a visible way is always smaller, sometimes much smaller. Like much of the internet, it’s a long tail. That’s why sites like Twitter etc target growth and then do all sorts of shenanigans to promote “engagement.”

I think we have to be very clear on what the problem is that we are trying to solve. We know that MeFi has a revenue problem, and we know one major cause of that: Google ad revenue dropping off a cliff. Would having more of the current user base be active in some manner fix that? I certainly don’t have an answer to that question though I can’t really see a connection at this time. So what would increase revenue? More users? Would more single $5 signups really make a dent? Seems unlikely but again I don’t know.

And maybe the problem we want to solve is just engagement, and it’s not tied to revenue at all. What is it that people would like to get out of a more active user base?
posted by DiscourseMarker at 8:01 AM on August 20, 2022 [6 favorites]

But anyway…what this is making me think though is what do we even mean by “use” their accounts? We are using posts, comments, favorites, as proxies for engagement because those are the only things we can see and count. Maybe there are some % of those dormant users who actually do read the site frequently, maybe even logged in, but they don’t click stuff.
Yeah, I agree with this. I'd love to do some analysis on data from server logs or Matomo (although as mentioned upthread, I'm a little dubious of the Matomo data), to see how many of these people are just lurking.
I think we have to be very clear on what the problem is that we are trying to solve. We know that MeFi has a revenue problem, and we know one major cause of that: Google ad revenue dropping off a cliff. Would having more of the current user base be active in some manner fix that? I certainly don’t have an answer to that question though I can’t really see a connection at this time. So what would increase revenue? More users? Would more single $5 signups really make a dent? Seems unlikely but again I don’t know.
My understanding of the situation (which admittedly comes from reading between the lines in the site update MeTas) is that the most significant revenue source (and definitely the one we have most control over) is user donations. I think it is reasonable to expect that having more (active) users will increase the number of donations — that assumption is why I am focused on active users. My impression is that $5 signups are a tiny slice of the pie — at most, $2,935 last year (not sure what percentage of signups were fee waiver signups), which is probably around 1% of the annual revenue, based on the numbers I've seen in previous site finance MeTas.

I think planning on ad revenue to save the site financially is likely a tactical and strategic error (although that opinion definitely could change if the percentage of revenue that comes from ads is a lot higher than I think it is).
posted by wesleyac at 8:24 AM on August 20, 2022 [3 favorites]


The site is so ugly when not logged in possibly some committed lurkers pay for an account just to not see ads and/or to be able to tweak the look.
posted by Rumple at 10:12 AM on August 20, 2022 [4 favorites]


I think probably everybody here agrees that the ad revenue ship has long since sailed. It’s certainly a plausible hypothesis that more active users would increase financial support, and maybe one option MeFi could consider is a more of a true subscription model, like other sites have gone to. (I personally feel like the word “subscription” is more appropriate than “donation” bc the site is not and will not be a non-profit and the word donation muddies the waters.)

I would find it fascinating and useful to know more about what has induced current financial supporters to support the site. For me it’s not really tied to my own personal activity, which has waxed and waned over the years, but I might be a weird statistical outlier, who knows. And maybe a targeted campaign for lurkers would help; it seems unlikely to hurt at any rate.
posted by DiscourseMarker at 10:31 AM on August 20, 2022 [1 favorite]


If Metafilter went to subscription only then I likely would not be able to participate. I can understand that the site needs revenue but I have extremely low income and have to watch every outlay.
posted by JHarris at 3:24 PM on August 20, 2022 [4 favorites]


I don't make many official pronouncements but if there was a subscription model planned (currently, no plans, but Steering Committee might decide to go that way, I have no idea and it's pointless to speculate at this stage) I feel that there will always be a way for lower-income people to continue to participate. There is already a fairly robust fee-waiver system for signups.
posted by jessamyn (staff) at 4:44 PM on August 20, 2022 [10 favorites]


>My concern stems from the fact that the rate of people [signing up and not using their accounts] has been consistently increasing

It doesn't line up with the data quite as well as I'd have expected, but "Add to Activity" was added in 2015. Prior to this, you had to publicly engage with the site (favorite or comment) to follow a post in Recent Activity. Now it can be done in a way that is fully private (and therefore not in the public data). As someone who reads far more than I comment, I find this feature useful; I can only assume that would also be true for someone who never comments. It would require an analysis by the site staff to tease out if those accounts are truly unused, or if they are used only in non-public ways.

I think it would also be interesting to see if the never-used accounts cluster around certain dates. The Our Father question has a steady trail of users who joined to post specifically in that thread, and answers were being duplicated to the point where several answers were deleted. With over 100 new users in the first three days and ongoing media coverage after that, I'd bet that a significant number of the "never used" accounts registered in 2014 were used to post an answer that was later deleted from that thread, or were registered to post an answer, but someone else posted it while they were registering.

Also, would be worth considering how spammers show up in the data -- whether it excludes closed/banned accounts, or accounts for deleted posts/comments. Otherwise, after a spammer has their spam post/comment deleted and account banned, they look like a never used account. Also, you used to have to post three comments to the Blue before you were allowed to make a FPP. I'm not sure when, but there was a policy change that removed that. That may have also lead to the appearance of an increase never used accounts, as even when banning a spammer's account, the mods usually didn't go back and delete the three comments if they were innocuous or actually contributing to the discussion.

This is not to say that it isn't interesting data, and it is still worthwhile looking into the reasons why people register then never use the account. We may well find that there are real technical, cultural, or other reasons why people register and then never become active, and that the cases I listed above account for a tiny minority of the number. I'd bet we didn't retain most of the new users from the Our Father question, and maybe there's something that could be improved.
posted by yuwtze at 7:45 PM on August 20, 2022 [2 favorites]


Yeah, the two things being studied here are the 'new user funnel' and longer-term user retention.

For the new-user funnel growth means one of two things: You can either identify leaks in the funnel and find ways to seal them, or grow the number of inputs at the top of the funnel. Plugging leaks might have a couple easy wins available; occasionally there will be some undreamt-of issue where a small intervention goes a long way. Perhaps there's one or two things along this line to do, and then you'll be trying to squeeze blood from a stone. As others have mentioned, lurkers gonna lurks and spammers gonna get ban-hammered.

The numbers at the top of the funnel are small enough, though, that it'll be difficult to meaningfully A/B test anything. So, if you were to put me to it, I would bet on trying to increase the number of user signups somehow. Unless miracles, I would expect 'engineered' sign-ups to be a bit leakier in the funnel than 'organic' sign-ups, but still the bigger fish to fry... Instead of chasing 100's of partially committed users, we could instead consider how to attract a small slice of the 7 billion non-users.

We're a weird place. The site is deeeeeeply anachronistic. But we just need to find ways to connect to the people who area good match for us... It's probably worth imagining how many people in the world might be good matches for metafilter? Probably a million or more, I would guess - it's a matter of learning to find and connect with them.
posted by kaibutsu at 5:37 PM on August 22, 2022 [4 favorites]


Thanks for the reassurance jessamyn. I'm sorry for chiming in with that, but whenever I hear about something I like deciding to go subscription a twinge of fear goes through me. I expected to be financially solvent years and years ago, but it never happened.
posted by JHarris at 12:51 AM on August 23, 2022 [6 favorites]


You have nothing to apologize for. This place is a home for so many people and I know the idea of change, even if it's potentially positive change, can be nervewracking if you don't know what might be coming.
posted by jessamyn (staff) at 5:04 PM on August 23, 2022 [7 favorites]


One thing occurs to me as a possibility regarding the 2021 unused accounts: interactive online leisure activity across the board took a hit as people went “back to work” (in quotes because, well, everything pandemic goes a bit differently than described). People may have signed up envisioning endless days of scrolling, only to discover that those days had abruptly ended. In my own non-profit work, we’re looking sceptically at all 2021 data when it comes to forecasting.
posted by obliquicity at 10:59 AM on August 24, 2022


« Older Do we know the demographics of metafilter users?   |   On Cooling Off MetaTalk Threads Newer »

You are not logged in, either login or create an account to post comments