A playbook for a successful developer community operations team


Alfredo Morresi

Job title

Developer Community Operations Lead




DevRelCon London 2023

In this talk from DevRelCon London 2023, Alfredo Morresi, Developer Community Operations Lead at Google, discusses the importance of data-inspired insights in community operations. He shares lessons learned from his experience in managing community programs and highlights the need for a solid data culture, technical skills, and thoughtful automation. Alfredo also addresses challenges in obtaining data from different teams and navigating data privacy regulations like GDPR.

Watch the video

Watch the video on YouTube >

Key takeaways

  • Data is essential for generating insights: Raw data is the starting point, but it needs to be organized, cleaned, and categorized to become usable. Visualizing the data through dashboards helps stakeholders understand what is happening.
  • Data should be explained with a story: Simply presenting data visually is not enough. It is important to build a narrative around the data to derive insights. This requires going beyond just presenting the data and explaining its meaning and implications.
  • Community data is difficult to work with: Community data is often messy, inconsistent, and influenced by external factors. It is not the same as big data, and the approach to analyzing it should be different. Triple-checking the data and adding context are important steps in dealing with the challenges of community data.
  • Building trust and collaboration is crucial: In order to successfully implement a data-inspired insights culture, trust needs to be built with stakeholders. Starting with small wins and gradually providing more complex insights helps build trust and positions the data team as collaborators.
  • Planning for growth and resources: As the data infrastructure grows, it is important to plan for the increasing complexity and resource requirements. However, it can be challenging to secure additional resources, so it is necessary to advocate for the necessary tools and infrastructure.
  • Consider legal and privacy regulations: When dealing with community data, it is important to consider legal and privacy regulations, such as GDPR. Working closely with legal teams and setting realistic expectations with stakeholders is crucial in navigating these regulations.
  • Start small and iterate: It is better to start with small wins and iterate over time rather than waiting for a big breakthrough. This allows for the gradual development of knowledge and trust, and also ensures that insights are delivered in a timely manner.
  • Overall, building a data-inspired insights culture requires a combination of technical expertise, domain knowledge, collaboration, and a strategic approach to data analysis. It is a continuous process that requires adapting to the challenges of community data and delivering insights that have a positive impact on the community program.


Lucy Jones:

Our next talk, a Playbook for community operations is a wonderful example of how your profession has evolved. When I first came on the scene, people were talking about how roles in DevRel would become more diversified, more specialized, but they weren’t talking about community ops. It wasn’t a thing back then. Yet here we are about to understand everything that community operations entails its place in the world and hear the key learnings from someone who knows a thing or two about a scaling developer, community programs, and B, the engine needed to fuel those programs. Alfredo Morresi was the guy who first stood up developer relations efforts in Italy, Google. He then was in charge of all developer community programs across Europe. And three years ago Alfredo was handpicked to figure out, set up and lead community operations for Google globally across something like 3000 developer community chapters we’re in safe hands. Please be upstanding for Alfredo Morresi.

Alfredo Morresi:

Yes, this is me coming from Italy, straight from Italy and in a kind of nutshell, I define myself a servant leader but also a community scientist because I love to explore the community space and also a developer because before joining Google to work with communities, I was a developer and still consider myself a developer. So I did the community developer, community operations team in Google and I’ve been dealing with communities for, I dunno, more or less 20 years when I first joined or started my Linux user group in the city where I was living. So back in time when Linux was cool stuff in addition to community and my job, I think I also have passions like anyone here and my passions are for sure being a dead. I love everything regarding extended and reality. And I also have a passion for running on the beach, snowboarding, preparing and eating tiramisu like if you want to make me very happy, let’s talk about tiramisu.

And if I have to pick a color to describe myself, I’m definitely a yellow person. Even if today my outfit doesn’t perfectly reflect this tendency to yellow. So Lucy mentioned developer communities in Google. Just to give you a little bit of overview, there are many community programs in Google as you can expect I support or my team supports three community programs. First one is Google Developer Group and those are like community of developers that are sharing the same technological interests with other people to learn together. Then there are Google developer, student student clubs. Are these community within universities, generally tech universities that students participate too in order to learn how to become a developer and build stuff with other developers within the university. And finally there is the Women technical program is a community program aiming to offer women and unrepresented gender a possibility to thrive in the technical world more or less.

There are 3000 chapters all over the world and today I’m here to sharing what I’ve learned in working for community operations, dev community operation for these three community programs. So before I start, just a disclaimer, community community management is not a perfect science. Someone defines it an art. I mostly agree even if we have good elements already and also community operation is not written in stone. So this is my experience. If you have different experience, it’s okay. And by the way, really I’m really curious to know other different experiences. So after the talk, reach me if you want to discuss about, Hey Alfredo, you told this, but I think or we are doing that, I’m perfectly fine with this. So let’s start with a quick audience check. How many of you have heard the term community operations? Raise your hand. Okay, how many of you are working or have worked in a community operations team?

Okay. Okay, good. So the second question, I think we can skip it because it’s for an audience that it’s already a little bit into developer operations, but let’s say the question stays and I will try to reply to it. What are the core responsibilities of a community operations team? Well, luckily we don’t need to start from scratch and there are different people that have built a little bit of culture on community operations. And here for example, it’s Kate defines it as the layer in between the vision of the community program and the execution of it. So community operations stay in between Tiffany and other person, very active in the community operation space has detailed the many raw responsibilities of a communications team. As you can see, there are many from process reporting, automation, managing of the technological stack. And again, it’s okay if you don’t do all of them, just some of them or if you do something different than what’s what is written here.

But both of these talks and converge to one core responsibility of a community operations team. And this core responsibility is manage what happens behind the scene to make the community program run smoothly, manage what happens behind the scene to make the community program run smoothly. Okay, this is what should be the core responsibility of our community operations team. I absolutely agree with this definition. I think it’s a great one. Gives a clear sense of purpose to the entire team and at the same time it’s flexible enough to do not be focused on just the one single kind of community program. Because as you may know, there are many different kinds of community programs, but I think there is also something we can add to give to this definition even more value. And in order to do these, let’s explore together what is important to make an indispensable community program.

To be honest, this is not something I define someone here in this audience have defined it, but there are two core elements that we need to account to create an indispensable community program, the business value and the community needs. What’s the business value for the company running the community program and what are the needs the members of the community think the community believe the community can solve for them? When we work at the intersection of the business value and the community needs, we have an indispensable community program. We have a community program that works. We have a community program that create value for the company and solve the needs of the community member at the same time. So keeping this sweet spot in mind, I think one of the core responsibilities of a community operations team is not only to do everything behind the scene, to have a smooth community program, but also fill the community team with data inspired insights to work on that sweet spot even more, even better, and to be even more precise, fool the community team with useful, timely and reliable insights, data inspired insights to work on that sweet spot even more, even better.

This is a big shift or at least it was for myself and my team in the way we consider ourself and in the way also the stakeholders we are working with consider us because at the beginning we were the data nerd doing dashboard spreadsheets and automations. Now we are the partners that are helping our stakeholders to take good decisions on the strategy of the community program and also of course to check on the execution of it. Another way to say this, it’s like using time like line dashboards and are in the past they describe something that has happened. Process automation is the present. It’s like dealing with what is happening. Insights is the future. It’s understanding what would happen if, what are the options I have and I can choose from in order to design a better community strategy. This is crucial, important and we will see how much this also will reflect your day-to-day operations, keeping these data inspired insights culture into account.

So how to generate these insights. Let’s explore a little bit some of the lessons and consideration of course shouldn’t be a surprise that in order to generate these data inspired insights, data is your best friend. Of course, I don’t want to do a to on community data. This is not the topic of this talk, but there is one single lesson I want to share with data. This is raw data, okay? This data, it’s in its primordial form. It’s like generated is what you get from your community platforms, from your processes, from what people are producing. It’s intimidating. It’s only good to be stored somewhere, that’s all. But once you store it, you have a first pass to the data and now data is sorted. So it’s splitted in different categories, it’s cleaned up, it’s still non really usable, but at least it is less intimidating.

Another pass and your data is arranged, it’s connected. And this is the stage where the techie people in the team can look to, okay, tell me more about this community, what information I have about this specific entity. Or okay, this is the impact of this initiative we’re running. So it’s okay, but still something not every people, not all the people can access to another step. And data is presented visually. This is where you have your wonderful dashboards, visualization, the stakeholders can understand what’s happening and they are happy. You are happy and your job is done. Not really because many operations team, community operations team stop when the data is presented visually they say, okay, my job is done now rest of the world, look at the data, get the information you want to get from it. I think in order to fulfill these data inspired inside culture, we need to move one step farther and we need to explain the data with a story because from the story we build a narrative and from the narrative we get insights.

So my suggestion is as a team, do not stop here, go there. As you can see it’s quite easy. This is where the complex stuff happens happens, but this is where you get most of the value in running your team. And this is where basically after a while people start consulting it in understanding, tell me more because you know what? You have this specific domain knowledge of data that allow you to provide me the insights and need to run my community program. So what are some of the lessons I’ve learned while building these data? Insights data, data inspired insights, culture first community data is difficult. It’s dirty. I mean it’s not perfect. It has several gaps, could not be consistent. So it’s not big data. Don’t imagine you can do what you do with big data with community culture because generally this is not the case.

You can get very good insights, but the approach should be different. This is also what we have, so we have to accept the fact and deal with it. So again, triple check the data you have always add the context to the information or dealing with because the standard is community data is a difficult one to deal with. Second, there are a lot of external influencing factors for this data. I give you a very simple example. Let’s take an in-person community program and you want to track something about it and you may track the number of attendees to the events organized by this committee. And also satisfaction score. This is quite standard. Have you ever measured the weather condition the moment event happened? And I tell you why because when it trains, very few people go to the event. So the attendee number will be lower than usual even for a very successful community.

And at the same time the satisfaction score rises because there are very committed people, very motivated. So they go to the event, they say, yes, it was great simply because I had to walk through the rain. So I cannot say it was a very bad event, but in any case, I’m not suggesting to correlate the success of your community program to the weather condition. But it’s just an example to tell you how much data community that you have can be influenced by factors you will never consider a never factor in your analysis. This is important. And third, perform a large scale collection of data is not always easy or not even feasible in some cases.

At the same time, don’t be okay with all the data you have. Whatever is the simple size. And again, I tell you why. If you’re measuring the satisfaction to an event, one single on a scale from one to five, one single on happy person can totally shift the perception if you do the average of other four super happy people attending the event. So if you simply do an average, you will see a number that doesn’t reflect the reality of one single on unhappy person and four very happy people. So my suggestion is knowing you cannot collect large amount of information, always set your bar for what is the reliable data I want to have in order to perform further analysis. And if you don’t meet this bar, go on and find another kind of analysis you can perform. Okay then in this data inspired culture, small wins are important because there is an important factor in building this culture.

Not the data but the people you’re working with. You have to build trust. It’s a change management process. They have to accept a partner telling them what they could do or what they could do better or even point the finger to, okay, this didn’t go very well. This is hard because we are humans. And so it’s important. Instead of performing a super duper complex analysis where you put everything together and you spend month in working with your data, no one knows what you’re doing. It’s better that you start releasing small wins. You build the trust, you start positioning yourself as a collaborator. You don’t wait for the big wow moment where you discover the truth about your community program, but you move one step at a time. And by the way, I keep saying data inspired insights because data is difficult and so you can use the data, but you have to add a lot of knowledge of the community program and over time also tend to these small wins.

You build this knowledge, you still have an impact or a positive impact on the community because not all the trivial insights are non useful. Some of the tri insights to get the easy ones cannot have a very positive impact on the community program. So do not discard the simply things that are simple to get. Think about how much impactful this insight can be in the development of the community strategy. And finally, these small wins, as I told you, allow you to plan for more complex insights to provide over time. Again, don’t leave your people waiting ages for you to provide something to them. Give them something in a very ative way. And so build trust over time and finally come with this. Okay, now there is time for complex insights I can provide you to say it’s a marathon, not a spring, not a sprint.

You need to build the knowledge of the domain in order to give context to this dirty data, to this difficult data. You need to build the trust of your stakeholders and also you need to be creative in getting new data. I think sometimes we have to hack a platform to get the data we need because otherwise there is no other way. Sometimes we have to literally pray other teams to provide us the data they have because we think there is something cool in this data we can use, but we are not still sure. So it’s something you have to do and remember, you have to follow the data because the data won’t follow you. So it’s our responsibility to follow the data because otherwise we have very few things to work on and of course like I said, told you correct insights are complex, complex to obtain, so you have to work on them. Often you have to perform analysis to discover you are going nowhere and then you come back and then you try another approach and then you add another piece of data and then you discover it’s a correlation. But then no, it’s not a correlation because it’s analyzed the data deeper. It was just a coincidence. It’s a marathon, but it’s totally worthwhile. One.

Another element to consider in building your communications team is how to grow over time. This is very true for teams where there are different people, but this is also kind of true for team where there is one single person doing all the community operations simply because if you know what is waiting for you in the future, you can better plan the present. And you don’t need to take rush decision just to, oh wow, this was totally unexpected, I don’t know what to do. Okay, we’ll take this compromise because otherwise I will die. Okay? It’s something that you can plan in advance so it’s better to know what’s going on. The first element is to consider that your data will grow relentlessly. The more you have these data inspired insights culture, the more you collect data, the more the sides of your data infrastructure will grow.

So something that today you manage, you store in a spreadsheet tomorrow you’ll need a database, something you visualize today with a dashboard for all the stakeholders you have tomorrow, you will probably be a business int intelligence software to cut the data in the way people want to see it. So it’s important to account for this growth. And the bad news is no one will Thank you for keeping the data infrastructure in order. While this data is growing, no one will give you free budget. Oh yes, please upgrade the database engine. Oh Grace, please buy a new tool to do data analysis. You have to fight for these resources. If you are lucky, they will see the results without even knowing what you have done in order to obtain these super complex results. But this is the reality, so we have to deal with it and embrace what we have.

In order to get more insightful insights, you have to collect more complex data. You start from something, you build your data culture and then you have to look around to find new data to get more complex insights. And also my suggestion to manage the community operations team is to consider it sort of technical team and consider basically the infrastructure. You have a software product, okay? This may not be new to this audience. Developer relations, of course we know how much software is important, but consider for some community programs that are not technical at all, can be oh wow is a software, oh, it’s scary. So it’s something that some other nerds should manage. No, my suggestion is to go for this approach and also leverage what we already have in the market in term of best practices in software development. There is so much territory around don’t go again, top-notch, start small, but start toward that direction.

Everything will be easier. For example, in my team, we approach the work we do using iterations of two weeks. We have a backlog of requests with bugs and features. We build roadmaps on this backlog. And what I’m trying to do now is to consolidate the role of a data product owner. So because data is a product, and so you need a person growing this product toward the needs, other stakeholders, other users, in this case, our internal company users have for this product, instead of saying, I take requests from everyone and then I will do everything for everyone.

Important to hire for the right role. Again, not rocket science. Everyone knows first and foremost, the team has to be built on the people you have. So the right people will build the right team, focus on it, and one single technical person can really make a difference in the way your data infrastructure is managed on the other side, because you need to have the community context to deal with this data. You remember, it’s important to have a role with both souls, community souls, understanding what you’re doing with the community and technical soul. And to be honest, it’s very, very difficult to find this kind of person on the market. So generally, or you start with a technical person with some sort of attitude towards community management and teach your community program on board this person your community program or you take a community person with some sort of passion from writing code and try to move this person more toward the software development world.

In any case, it’s something generally you have to do because it’s very, very difficult to find these expertise outside. Final point processes. We spoke about automation. What are the lessons I learned about automate yesterday in a talk it was written rootless automation, automate everything, yes. And let’s see this end. What will be to put simply pick up the automations or the processes to automate that are also able to generate additional data you can use for your analysis. So be picky in what you automate. And I will also detail the reasons for being picky. An example, you want to automate how people are asking to build to create a new community in your community program, a new chapter of your community program. And you are asked to automate the process. So intake of the request, analysis of the request. So there is an hour flow and also the communication you were approved and these are the next steps.

Thanks. But for some reasons we decide to not move forward with your request. It’s easy to automate and once you have automated the process, you may think your job is done, but what if you start getting the data from these out processes that you have automated in order to generate new insights. For example, you correlate where these requests are coming from with where your communities are. Are you having an organic growth only where the communities are already present or you receive requests also from areas where you aren’t present, your communities aren’t present, are you rejecting more requests than usual in particular areas if yes may be a different in culture of this particular area with the cultural frame you’re using to evaluate this request. Those are important insights and thanks to the process that you have automated, you can get them something that at the beginning you didn’t have another important part.

Yes, instead of ruthless automation, thoughtful automation also because every time you automate a process, it’s like spending some budget, some complexity, budget for two reasons. First because you make the debt infrastructure bigger and so you have to maintain a bigger surface if you want to change something. One approach is to, okay, one situation where I have to update only three processes, so I change something in the data, I go to these three processes and I’ve done, if you have 100 automated processes, every time you make a change in the data infrastructure, and as I told you, you have to make change because the data complexity will grow. You have to maintain 100 automated processes and it’s time consuming, it’s a results. It’s kind of a budget consuming stuff. And also because once you automate something, people expect it works forever. And so if you ask them to go back, because at the beginning we had an idea, but then we tried and the idea wasn’t great, we have to go back.

So be thoughtful in what you automate and measure, measure, measure. I mean what if to automate something, I take two weeks time of a person versus the same task that takes one hour. It’s very boring, but takes one hour every week. You know what? For one year you can still run it manually because I could spend the same amount of resources to automate something that has a bigger impact or may save more people in aggregate for the people of the team working with it. And finally keep things. It’s okay to keep something manual for one single reason, what I call the important nuance. The fact that when you talk with people, those are communities. When you talk with people, you can discover something very interesting. Your perfectly automated process won’t tell you. So to wrap up, the goal, at least from my point of view of a community operations team, is to fuel the community team with useful, timely, and reliable data inspired insights to work on the sweet spot of the indispensable committee, even more, even better, thanks to a solid data culture, the ability to deal with a growing complexity of the data and also technical skills.

Also thanks to a thoughtful approach to everything regarding data and automations. And also thanks to you for having listened to me.

Lucy Jones:

Thank you, Alfredo, thank you so much. Do we have questions for Alfredo? Lots to take in one right there.

Audience member:

Thanks Alfredo. Fantastic talk. I have a couple of questions. So I mean very often in large companies I think different parts of the processes are, or where we could source data are owned by different teams who may not have ever alignment or whatever it is. That’s one thing I think is if you could provide some guidance around that and how to get that and also to do with say GDPR or other kind of PI type concerns. Do you find that maybe in Google even limits the kind of questions you could ask? Maybe very interested in asking.

Alfredo Morresi:

This is generally where I say you should follow the data because data doesn’t follow you. Okay, so you need to chase the other team and ask data for them maybe telling them why you need this data. Sometimes it really depends on the people. There are people that are okay in sharing data. You have, there are people that are very conservative, oh, why are you asking me this data? What do you want to do with it? So it really depends on the situation you have in front of you. Generally the small wins can also be used to buy the other teams. And so they say, you know what? I want to calculate this and this and this, and it’s true, I’m doing it for myself. But what if there is also a small advantage, a small win for you, it’s free. Unless of course you have to work to give me this data, but if the data is there, you basically your effort to extract data. For me it’s close to zero. Why say not to this opportunity I’m offering you. So it’s a little bit political, but it’s also a little bit working in a big company. No difference for GDPR.

It’s not only GDPR because GDPR in Europe. But when you deal with a global community program, trust me, there are many different. First you need to be good allies, lawyers inside of the company or externally telling you what is important to do and how you should also process, manage process and store this data. So don’t think you are able to understand the whole regulatory framework because it’s a matter for lawyer. So find good allies that are telling you what to do. And second, set expectations. If your stakeholders want this wonderful analysis on every single person part of your community program for the previous three years, tell them it’s not possible because we act in a kind of regulated user data privacy framework. So you may have aggregated information, so offer them other opportunities. Don’t say simply no, but tell them they set their own expectations in the right way because you know what is possible. They don’t know what is possible.

Lucy Jones:

Some good advice there. Any other questions for Alfredo? No, Alfredo, I’ve got a question for you. So you touched on it with thoughtful automations and it’s something that Matt and Rhonda were talking about yesterday. If looking back when you first set this up at Google, what were the kind of real game changers for you? What processes that really landed? Well,

Alfredo Morresi:

I mean one of the first big small wins was to be able to track the success of some of the activities we were doing because previously the people were able to do it looking in different places, putting together and sometimes doing differently based on who was doing it. And so it was a little bit of a non misalignment internally. When we showed, okay, we can look at the results through these dashboard, people say, oh wow, really? Yes. And by the way, the data was there, we simply rearranged it. Everything was there, the data was already ordered sorted. We simply build the dashboard, the visual part. They were so happy. They said, okay, now we want to run this dashboard for all the programs and this is where you say, wait, wait, wait. Yes and let’s see what are the most important programs. Let’s see what you can still collect. Because building this dashboard requires time, it requires the sources. I cannot build that dashboard out of nothing unless you give me 100 people or whatever. So it’s like always a trade off you have to take, but probably showing people what is possible. Okay. Was for me the most important initial win.

Lucy Jones:

Sounds like magic. Any other questions? Quick one or I think we are done. Can you please give Alfredo a big welcome? Thank you.


Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.