Cloudflare Careers Day: Radar: A 360º Collaboration Story

Welcome to our next session in this Cloudflare careers day for Europe. My name is Marc.

I'm Product Director here at Cloudflare in Lisbon and with me today is Celso who is Director of Engineering and Ricardo who's our Lead Product Designer in Lisbon as well.

What we'd like to talk to you about this afternoon is how we build Cloudflare Radar which is one of the products we're building here in Lisbon and it's one of the products that actually just started a bit more than a year ago.

When I joined Cloudflare and JTC, John Crane coming our CDO came to me and told me yeah we have all this data and we love to share it with the public.

We love to build something where we share knowledge with the public and there's a ton of ideas within Cloudflare.

Just talk to some people and figure out what data we're having, how we could share it, how we can make the Internet better with that and then I went on my way.

It took a while. I talked to a lot of people and I got a lot of insight about what we can and what we should do and then I went looking okay well we actually need to build a product and I realized okay I can't.

I'm a Product Manager so I'm really good at talking but I'm really bad in building actual software so I needed help in building software.

So while looking for engineers, my lucky day came when Celso started and started building up the software engineering team and he then joined up and we started building Cloudflare Radar.

I think what we realized during the first phase or what I realized is a lot of data is great but it's super complex to find out okay how can we handle this and how can we actually find the right data and how can we still follow up on our promises that we keep everything private and privacy focused and then we got our great engineering team together and we built prototypes.

I think we tried out a lot. We built a ton of prototypes and then we had the first version we could use and then the Cloudflare hammer hit us when we were told well there's a Cloudflare birthday week in October and this is one of the projects we want to show there.

So we like had two months to build a product and I'm going to hand over to Celso because he's going to tell you how we made it happen in just two months from more or less nothing to building a product that is now used by thousands of people.

Sure thank you Mark. So hello everyone I'm an engineer director for Cloudflare in Lisbon.

I joined the company in late August last year right in the middle of the pandemic.

So fun fact I've never been to the office until now and that's right the minute I joined Cloudflare Radar was headed for launch on birthday week and it was scary for me because suddenly I just joined the big company still trying to learn the ways of Cloudflare how things work and I had a deadline in front of me to make sure that we were able to launch Radar on birthday week but at the same time it was also proof that Cloudflare is unique when it comes to culture and making sure that the teams work together and help each other to make things happen and that was totally the case with Cloudflare Radar.

We didn't have an engineering team a dedicated engineering team but we managed to build a group from people from the Intel team a couple of very experienced engineers I'm talking about Nick and Sri who are you know founders of Cloudflare Radar and they totally joined the boat and they were able to bring us the engineering muscle that we were needing to launch Radar.

We did a lot of prototyping, we did daily stand-ups with everyone on the team including the design team in London and we made it.

We actually launched Radar on birthday week and I'm very proud of the work we've done and of all the people that helped making that happen.

I think that is a living proof of how Cloudflare can actually make things happen in such short time if everyone works together.

Maybe Selsa you just mentioned and I mentioned it as well that we built a lot of prototypes but we also had a tight deadline and usually you say you can't experiment when you have a tight deadline you just need to build without being able to iterate.

How did we do it and how did you and the engineers make sure that we still are able to iterate and make the deadline?

What was the secret sauce of making that happen?

I guess there isn't a list of things you need to do to make that happen but I mean when you manage a team in my case it's an engineering team but I guess this applies to any team.

It's up to you to decide what's the best methodology for that specific team and that specific project.

How you're transparent in terms of what's the roadmap, what's the goals, what's the deadlines you have in front of you and making sure that you have enough resources and space to be able to experiment and to be able to try things that otherwise if you don't do that it will be very difficult to accomplish because the roadmap always takes all of your resources and I think we've been able to do that.

I think Cloudflare Raider is the kind of project that really allows you to do a lot of experimentation because it's database.

We got a few tools we use for instance a lot of the things you see on our public pages we do prototype all of that internally using tools like Jupyter notebooks, our data scientists are always doing studies, always experimenting with different formulas, different visualizations, sharing that with the product team.

Ricardo also helps with user interface ideas and then when we're finished with all of that experimentation we take decisions and we go for the public versions of that process.

So yeah, I think we've been able to do a good mix of things that need to be launched, making sure that our deadlines are in place but also finding the space to experiment and having the right culture and the right tools to do that internally.

Thank you. I still remember when we did the final designs for Raider we iterated like sometimes deep in the night my time because one of our designers is based in San Francisco and then she worked during her day and then we reviewed it in the morning.

I think there's good and bad with like different time zones but that definitely helped that gave like eight hours of work without my constant questions and then we launched an MVP and I just a few months after we launched the MVP Ricardo joined and found us still in our super experimental MVP we need to get it out mode and finally came to our rescue to have a bit more of design strategy.

Ricardo, how would you describe what you found when you joined and how did we transform it into a way of working by now that is maybe a bit more mature?

Thank you so much Marc.

Hello everyone, I'm Ricardo. I work with the Pixels and Words in the beautiful city of Lisbon for a living and I'm the father of two loving siblings whose affection derived most of my crazy ideas and exactly those crazy ideas were actually the thing that I felt the most was needed when I joined.

So I joined not even four months ago and what I found it was a team that had a lot of ambition and that there was a lot of dynamics for collaboration and there was a lot of room for improvement and a lot of room to establish good design practice and good design methodology.

So I found the team that was really motivated and most of all it was open to experiment new design approaches and new ways of working and new ways of making things happen.

So that was my initial entry on Radar. What are like some methodologies that you introduced where you think okay that really helped in terms of the collaboration but also in terms of making the product better for our customers?

For sure. So of course we now are in front of a reality that is different for all of us and working remotely and collaborating remotely it's the first challenge that we have but it's also an opportunity of course and the challenge was to drive some collective ownership of the design initiatives that we want to ship and deliver.

So it was the establishment of a really tight relation with engineering.

So I believe not only design but engineering as well should have a collective ownership over the entire development lifecycle.

So not only was the creation of design jams when we jump everyone on board for instance I got in meetings where we had an agenda to discuss some UI detail or some feature that you wanted to ship but actually the first 20 minutes of it was actually the engineers that were on the call discussing with each other to gather some understanding.

So was creating like design acting like a facilitator and then negotiating between all parties to get that understanding and to be able to drive some other initiatives like a synchronous collaboration.

So we all have our time to work our focus and most of the initiatives were meant to be executed and meant to be accomplished where we draw a framework and we have a way to achieve our goals and as synchronously we try to and we actually made it to collaborate and to get some insights and to get more alignment on what we really wanted to ship.

Of course I had the luck to immediately emphasize with João.

João Tiago is a really good front -end engineer and it's actually was it was actually a surprise because the team already had the dynamics the team already had the motivation and the ambition regarding design and how can design help figure out what we what should be the best approach to tackle a problem.

It was really it was a everyone was very welcome and very open to experiment and try new methods of doing design and actually as I used to say design is not a department design is a verb and actually the Radar team all jumped together and designed and that's what I love like it's not the department everyone on the team participated and likes to design as well so I think it was great that we got everyone together.

So what was kind of the reaction from the engineering team when we started like having those like joint design workshops and when we started to use a bit more methodologies and a bit more of a structured process in terms of getting everyone together and working on the design of Radar?

The reaction was great and that applies to Ricardo and more recently João Boto was doing also product management.

I mean if you leave engineers alone we can be creative but also not extremely organized and you know having help from design in terms of making sure that we think our interfaces and our screens at specification time and that there's a rationale behind all of those decisions and the same with product you know having proper specifications a good strategy on how why we're doing things and what's the best approach as to launching features I think that that helps engineering quite a bit and the team totally understands that and was looking for it since the beginning.

So the reaction was amazing and Ricardo has been participating in most of our daily stand-ups.

I know that Ricardo is also involved in other projects but we feel like a big family we feel like a team.

I think it's really working out. How would you say is kind of the what is the strength of Cloudflare when it comes to like also collaboration outside of the team?

I know Ricardo you've been also like using best practices that we got from Radar and then work with other designers to implement it in other teams and the other way around.

So maybe you can tell us a bit about like how the overall alignment and knowledge sharing with the design within the design team works.

Well sure of course.

I would say mainly Radar is a really good opportunity of course Cloudflare is full of good opportunities like we always are told to do so like find a problem and solve it.

In terms of aligning our product design methods and practice we at Radar have an awesome opportunity because mostly regarding data visualizations we have the opportunity really to mature how do we interact and what's the message and what's the meaning that we want to extract from a data visualization and it's kind of an experimentation laboratory where we have the chance to try really different things and this of course it's exported to the core dash of Cloudflare.

For instance we did the research on the colors applied to data visualizations and that really worked out because we were able to actually export that work on the categorical colors for data visualizations on the dash and that's really important because we not can not only share the knowledge but share the expertise and the domain expertise that we have on Radar and for instance there are there were other methods that were established but actually they were born on Radar and they are actually being exported to other practice mainly regarding for instance research where we engage with some of our user groups and we had the results from the interviews and then we had like a framework to analyze that research and to ask together some insights so we focus on product themes and we are were actually able to drill down some insights that can help us make more reasoned decisions and make us not follow that much assumptions but be more precise and be more to as also was mentioned to have a rational to do something and that's something that we as well engage with the product design team and share as a framework to get some insight that was really welcome and recently we we drove like I said an asynchronous collaboration and workshop and that was really well received by by the product design team because it was an environment like we all expect is very dynamic it's very people are very motivated people are very ambitious but together like a shared vision and together like to together the diverse thinking that we all have because we have different profiles we had to conduct a very inclusive exercise and not only the the data visualization scholar but as well the research and this recent initiative that we had it was actually a case study to to expand to product design team that will be then applied to other teams and to solve other products so we have in sort in such short period we have already a really good case on on making things happen with a strong rational.

I'm just interrupting the flow for a question out of the audience which is for Celso.

So the question is radar has such a big potential for further features and functions are there on open positions directly for the radar team?

Yeah first of all we all do agree that there's big potential on radar and we have a very ambitious roadmap until for the rest of the year as you know Mark and yes we'll be opening at least four new engineering positions for radar until the end of the year.

First ones will come out beginning of Q3 so hopefully in early July and then we'll spread the other positions until the end of the year but yes the team will keep growing.

We're about to hire an engineering manager for the team now. By the way the team is currently six engineers and we work closely with other teams at Cloudflare that help us with data or APIs but the team will keep growing absolutely.

Cool. Maybe coming back to the design and the development process. I think one thing I'd love to talk about is how we actually decide what we're going to do because I think radar is one of many Cloudflare products where it's great for us to get feedback from customers.

I think one of the best options to get feedback is just to listen what customers write.

I think radar is such a great public product where you just have to look on Twitter and get a lot of responses and a lot of comments on what we are just launching.

I think another thing for us as product managers is to look on Hacker News.

Every time we post a new feature or a bigger release there are some cool comments on there and that really help us to gather insights about how our customers feel.

But that was not enough so we did some more customer research as well and maybe Ricardo you can walk us a bit through how we as a radar team but also how we as Cloudflare make sure we're actually listening to our users and to our customers and how we can derive a great product from our customers feedback as well.

Of course, so we have our goals and our objectives and our ambitions for radar.

But of course, as I encountered and I fostered that idea, first of all we have to have the notion of human-centered design.

That's the first thing to accomplish and in terms of we have a product and we have a lot of ideas but we have to make sure that those ideas really meet users' expectations.

So for instance, we have a user group that is the industry analyst, we have a journalist as well as Internet couriers but those user groups have different mental models.

So they have different ways to perceive information and they have different ways of navigating on a web page for instance and this kind of different mental model and user representation is Basilar.

So we have for sure, we all know for sure, great products that have great user experience but didn't quite meet the business goals.

And as the opposite, we have a lot of product that goes to business goals and we have the features that we want to ship and believe that they are going to be and met user expectations but they don't.

So actually, the effort since the beginning I joined was to involve research, to really know who are our core users and having that in mind and having that characterization of the users, it acts like a decision helper.

So when we are wondering if we should move in this direction or this direction, when we have really user representation and we know who are we talking to and moreover and more important, we really know what people want to see, what attracts them, what will make them come back to the website.

So it's actually a mantra that we all follow to be human-centered design and this for sure started as an initiative but user research is a strong factor for achieving our goals in this product.

Very cool. Celso, Cloudflare is a company that builds a lot of products from engineers for engineers, which makes like product management but also engineering an interesting job.

How is the engineering team involved into bringing in their own ideas and making sure that as a core user as well as the people building radar, they are involved in all the processes?

Sure, there's a lot of things happening behind the scenes that are not visible on the public radar pages.

I think our main goal is to make sure that we're faster and better as we evolve and we've been doing a lot of tooling and automation behind the scenes.

So there's a lot of engineering work that doesn't necessarily introduces into visible features and that includes things related with data exploration.

We do have a lot of data at Cloudflare. I mean, I think the latest figures point to about 70% of the Internet traffic comes through Cloudflare and we do have a lot of data we can use to provide insights on radar and we're still learning and we're still experimenting on how we can use that data to provide value in Cloudflare radar.

So yeah, I mean, we do a lot of prototyping. We tend to try and automate as much as possible.

We do a lot of iteration just like what happens with product where you launch things, then you improve things and you keep improving and you keep iterating.

That also happens behind the scenes with technology and engineering.

That's a way of living, I guess. Yeah, that is it. Cool. And I think when you look back at the journey of radar, which is a product and reflecting on it, that's not even out for a year.

Like we just launched in birthday week, which was beginning of October.

So we are out there only for a few months and the idea is not even a year old and from the first discovery and the first brainstorming.

Now we talked nearly half an hour about radar and how we're working together and it sounds like we've been working together for years and built a product for a long, long time.

But I think it just shows a lot of the Cloudflare way of building products that we're not talking about years that we've been working together, but it's not even a year.

Like Celso, you joined in August and Ricardo just beginning of this year.

So even though I'm only a bit more than a year in Cloudflare, I still feel like the grandfather of this product.

And overall, this is kind of the speed that we want to work in and that's also the speed that we want to continue.

And I think overall, this is something where we can build unique teams, but it only works if everyone really pulls together.

And I think what we've learned over the last half hour, that radar really is one of those examples where we pull every resource we have in, we ask everyone to bring in their ideas, to voice their opinion.

And even though that sometimes leads into some discussions, but usually they have a very fruitful ending instead of just leading nowhere.

And if everyone is free to voice their opinion and free to voice their concerns, we're really, really building a great product.

Celso and Ricardo, thank you both so much for joining me today and talking about your experience in radar.

And yeah, we're up for the next iterations and for the next releases of radar.

Thank you so much. It was a pleasure.