The best community metrics!
What to measure and how to measure it to keep people in the center, create positive impact and grow financial outcomes
A couple of weeks ago I published an essay focused on 6 key elements for building and growing healthy communities.
One feedback I received many times was that I should dive deeper: it’s not only about the elements, but how you measure them.
Based on that, I started to work on this essay… But while doing it I encounter two challenges.
First: community-based companies and all the web3 paradigm are at the very dawn of a new model for businesses and economics.
In the words of Morgan Beller:
For most web 2.0 creators and content, it’s about how many eyeballs can you get? How many followers do you have? How many views do you have?
[H]ere in web 3.0, the real metrics are commerce. How many people purchase an NFT? How many people purchase a social token that gives you membership into a community? And then how many of them participate in that community and take some action in that community?
This paradigm shift is not a superfluous nor isolated one. This will change the face of businesses and it is happening amidst the second Capitalism vs Socialism crisis.
We have learnt the hard way shareholders are not the only nor the most important stakeholder for a company, and that profit is a good metric but by all means, is not enough when evaluating the whole.
Those mistakes can be reborn in this new era. And we must avoid them if we want to create a healthy and positive new chapter for the internet, capitalism and the world.
Second: there’s a lot written about metrics and mental models to evaluate companies and community-based projects. What could I contribute to this?
To the first challenge, my answer was that if I were to tackle the metrics for a successful community-driven company I needed to be aware of this and reflect on its implications and opportunities.
To the second one, my answer was to try to organize the metrics in a way that make sense for community-based projects. In the words of Alex Angel, Comsonr’s Chief Community Officer: “It may not be intuitive how everything ties together”1
Understanding the principles we’ve been discussing in past essays and proposing a taxonomy may be a good way to add value in this arena.
Let’s jump right into it, then:
I. Discovery
TAM: although many communities are probably niche by definition, I believe community-based projects need to have a good awareness of their projected potential universe. I would suggest following a bottom-up approach. This essay has a great example of how to do it: https://thepathforward.io/how-to-estimate-your-potential-audience/
Conversion Rate: it’s the result of the number of new members over the total interactions. For example, if your community won 50 new members in a week where you connected with 1000 people across diverse media, you have a conversion rate of 5%. — Note: since communities are nurtured by content, I assume interactions/connections should be counted on all your channels together.
II. Acquisition
CAC or Members Acquisition Cost (MAC): is the full cost of acquiring users, stated on a per-user basis. You should track separately organic acquisition and paid acquisition. And the standard formula is: [total acquisition costs] / [number of new members acquired through paid efforts]
Time to convert: is the total time it takes for a follower to become a paying member. I’d analyze this on a cohort base or on a segment base (imagine your community-based project have 3 different subsets of users, the turnover time may be different for each one). Heads up with two critical considerations:
Not everyone will or should convert into paying members! Each segment of your users fills a critical role in the full fabric of your community. Even the lurkers? Yes! As Rosy Sherry would’ve said it: they are not lurkers, they are consumers, shy, not able to engage…2
Conversion is not a definitive state. You may see people becoming members and then going back to not paying… those changes are natural in any community and I will address them in the next section.
III. Activation
Stage distribution: take all the people in your community and identify their current stage, for example, consumers, active users and super users. You can have the stages that better suit your community, but I’d recommend being clear at least in three levels: passive, active and super active members. Then map out the distribution: Passive = X%, Active = Y%, Super active=Z%
Stage change over time: As I said in the previous section, conversion is fluid. All community-based projects should approach conversion as a stage change and monitor how and how often people move from one stage to another. For example, your community may start with these stats for a given week:
Community total universe: 11,550 ppl
New followers: 1000
Passive followers: 10,000
Active users: 1000
Super Active users: 500
Paying members: 50
Then, the next week the stats are:
Community total universe: 12,500 ppl
New followers: 1000
Passive followers: 10,100
Active users: 1000
Super Active users: 400
Paying members: 75
For some reason, your super active users decreased while your paying members grew. Also, you see an increase of passive followers of the same number as the decrease of super active users… Beware: you shall not assume they are the same persons, these metrics show correlation, it’s your job to work out the causation.
Monthly active users or MAU: as a zoom-in into the previous metric, you want to know what percentage of your members have taken any action within the last 30 days. Based on the stats given above, your MAU is 11.2% or [active users + super active users] / [Community total universe] * 100.
That’s not a very good MAU! If we believe Commsor we should seek for a range between 30% and 70%.
Also, seek stability in this metric. Peaks are sexy and lows are frightening, but if you are focused on keeping your healthy MAU stable, you are up to a win.
MRR or Monthly recurring relationships: This metric I stole from Rosy Sherry. And I think it makes all the sense. If the basic element of any community are the relationships, we should be measuring them. To design a useful metric for this I defined relationship as any interaction between two members, excluding the interactions with the “community account”. This exclusion is an artificial restriction I introduced to force a focus on the connection among the members assuming there is a value-driven connection to the community account. If we don’t exclude this, we may be building a media project, not a community! Being said that, the basic3 MRR formula is: [total interactions among all members] / 30
Average Content Per User or ACPU: is the total content created by the members divided by the number of all the people in your community, for a specific time.
Let’s say that from the 12,575 people in your community you have 10,000 content instances created in a month. Following the formula, your community has an ACPU of 0.79 in that month. Much wow, right? Almost every person creates one piece of content…
But if we know that from that 12,575 only 1,400 members are active, we can re-calculate our ACPU to 7.14. That’s a much better result!
Number of reactions per content: Reactions are the medium through which we express some relatedness to a given content. I think this is a good proxy for the quality of the content, and like such, I would suggest measuring it in two levels:
Number of reactions per content: it’s the total reactions per each content in your community. If we had 10,000 content instances, your table should show those 10,000 data points with their respective number of reactions.
Monthly Average Reactions per Content or MARC: is the [total of reactions] / [the total content instances] in a month.
Effective Promotion: one last metric for activation is how much our users are promoting the community. There are many metrics for this and they depend on the promotion model our community has.
NPS: is a good proxy for the willingness to promote a community, but it does not give us a real number. I’d suggest using this metric in the first months of the community and moving forward to more specific ones later.
Monthly Average Effective Promotion or MAEP: Measuring how many members promote joining the community and how often they do it gives us a clear understanding of the promotion dynamics our community is creating. I’ve used direct surveys and promotion-referral links tracking to understand the effective promotion of my communities. The formula is [total of effective promotions] / [total of effective promoters] in a Month.
IV. Togetherness (‘Retention’ in traditional terms)
Critical Behaviors: when we want to measure values alignment we should focus on the behaviours our members exhibit. Values only matter when people act based on them. To measure this, you should identify around 5 critical behaviours that your member should exhibit in a clear fashion. Once you have the list, measure the percentage of active users doing those behaviours in a given time period.
I would suggest not to measure economic behaviours here. We want to measure togetherness, not financial viability. The behaviour of subscribing or giving a tip will be reviewed in the next section.
The behaviours should be repeatable (commenting, submitting something, etc.)
Rate of Returning Members or RRM: is the percentage of your active and super active members that engaged in your community more than once in a time period. The formula is [total number of returning members] / [total number of active members] * 100.
Let’s say that from our 1,400 active users, 850 returned to the community more than once in a time period. Our RRM is: (850/1400)*100 or 60.7% Not too shabby!
Rate of Community Generated Content or RCGC: if we measure ACPU as an activation metric, the rate of ACPU versus the “community account” generated content is a good indicator of the ownership the members feel regarding the community. The formula is [total of member-generated content] / [“community account” generated content] and if the result is bigger than 1 your RCGC is healthy. Following with our example:
10,000 member-generated content
60 “official” content (2 per day in a month)
RCGC = 166.6 member-per-official rate 🔥
Or this other scenario:
30 member-generated content (one per month)
60 “official” content (2 per day in a month)
RCGC = 0.5 member-per-official rate 💔
Happiness GMV: This metric was coined by Sarah Tavel. I think this is a great way not just to measure retention, but to understand it as a togetherness factor as well. The way to measure it, following Sarah’s approach is:
Based on first-principles analysis, define what is the members’ experience that will lead them to stay and return to your community (if you are starting, make your best guesstimation)
Measure the percentage of your (potential) members that get that experience. The higher your %, the better.
Track this over time and optimize your members experience to reduce the imbalance between your members’ expectation and their experience in your community.
Social shares per content: is the sum of all the times a given content piece is shared on other channels-communities-media.
Pending: driving meaning is one key element of a healthy community, and I believe clusters and topics distribution may provide a good insight into this element, but I have not been able to find or craft a relevant metric for it. Any suggestions?
V. Growth & Revenue
Compounded Monthly Growth Rate: it’s the rate at which your membership is growing periodically4. The formula is: [Latest Month Number of Members / First Month Number of Members]^(1/Number of Months) -1
Churn: remember we said that conversion is fluid? Churn is the effect of this fluidity in terms of growth. For community-driven projects churn can be unfolded in two different kinds of churn:
Members Churn: Focus on the growth-driven engagement of your members, not their financial outcome. The formula is: lost members/prior month total of members. In our example:
We started with 11,550 people (Month 1)
Next month we reached 12,500 people (Month 2)
But we saw an increase of 1,000 new members (Month 2)
We lost 50 people
50 / 11,550 = 0.004 Churn rate
Subscription Churn: This is very similar to the Members Churn, but here we want to focus on the financial side of it. The formula is: ([Montly Recurring Revenue (MRR$) lost - MRR$ retained) in a given month] / MRR$ of prior Month)*-1. You want negative churn. In our example:
We started with 75 paid members (Month 1)
Each member pays $5 / month
MRR$ = $375
Next month we have 70 paid members (Month 2)
(($25 - $350) / $375)*-1 = 0.86 Subscription Churn Rate 😕
Or this other scenario:
We started with 75 paid members (Month 1)
Each member pays $5 / month
MRR$ = $375
Next month we have 80 paid members (Month 2)
(($0 - $400) / $375)*-1 = -1.06 Subscription Churn Rate 🤑
Monthly Internal Commerce or MIC: these are the sum of the total transactions between your members, both in number and in financial (MIC$) terms in a Month. These transactions depend on the nature of your community, must be neatly related to the value of the relationships happening there and can happen in very different ways:
NFTs commerce
Merch commerce
Subscription to a community member in your community
Tips to community members
Etc.
Average Community Network Value or Avg CNV: is the value of your community in terms of the number of members and the density of the relationships between them. As communities by definition are affected by NƒX knowing this metric casts light on how valuable your community is not for your members but to other stakeholders and the market. I propose Avg CNV is the average of the growth law5 of members [n * log(n), where n=number of members], the MRR and the MIC. The formula would be: [n*log(n) + MRR + MIC] / 3. In our example:
Total number of members: 12,500
Total interactions in the Month: 25,000
MRR: 833.33
MIC$: $415
Avg CNV = [12,500 * log(12,500)] + 833.33 + 625] / 3 = 10,906.70
Avg CNV$= [12,500 * log(12,500)] + 833.33 + $415] / 3 = $10,836.70
What do you think of this approach? Let me know and let’s build together a better metrics stack.
https://www.commsor.com/post/community-metrics
https://rosie.land/posts/lurkers-are-people-too
I think there should be a complete MRR formula that reflects the density of the recurring relationships… but I haven’t been able yet to come up with a clear formula. If you have any ideas, please let me know!
https://a16z.com/2015/08/21/16-metrics/
https://a16z.com/2016/03/07/network-effects_critical-mass/