It is a good practice to sometimes stop and turn back. Sometimes you need to see what you left behind, what successes made you go that far, and what things you missed and never had time to enjoy or explore. The more often you do it, the better your decisions may be in the future. It is essential to do it at least once a year in open source.
This article will take you on a journey to the past to see what happened in 2022 at AsyncAPI Initiative.
If you are interested in previous project updates check other
Project Status posts.
At the end of 2022, we had 3114 members in the AsyncAPI workspace. But what does that really mean? Don't get me wrong; I like when the numbers are high, especially if I see them in my bank account.
The thing is that some accounts might belong to people that joined three years ago but don't even remember that they had an account in the AsyncAPI workspace. It would be good to check how this number relates to people's activity.
Another important item to understand is how come we have 3114
Below you can see how many members have joined AsyncAPI Slack in past years. The numbers are pretty neat! People join AsyncAPI Slack because they know:
- someone will answer their call for help
- it is a welcoming place for first-time contributors
- we participate in many mentorship programs
- this is the best place to get closer together with the community
My prediction is that it will be growing and growing cause existing members will tell others what place it is and why it is worth joining.
Weekly active members
Below I'm showing the average and median activity throughout the years. I think the median is more accurate as it eliminates super-high weeks (like typical conference-related noise) and weeks when we have lots of holidays simultaneously, like Christmas break.
Last year was the first time the median value was much lower than the average. It is higher than the previous year, but not something we got used to so far. We noticed a significant increase in Slack members, but I think we can say that this number doesn't correlate well with median activity.
We should look into the data more deeply and understand what happened between April 23 and June 21 when numbers got much higher than usual. For example, what the heck happened on April 29 when we had 583 active members? Was AsyncAPI Mentorship the reason?
Anyway, I definitely predict that data for 2023 will show an increase in activity as we plan to do a series of meetups around the world that will bring more and more people into Slack.
Below you can see all-time data that show the number of messages posted on each channel in Slack.
General channel is the oldest, and yeah, it is generally for general stuff... what a joker. Anyway, the most important is the discussion size in the
Tooling channel. Specifications are boring, and people are interested more in how it works. On the other hand, tooling is most attractive to tech folks; coding and talking about it is more fun.
Remember that AsyncAPI Initiative cares about both the specification and the tooling ecosystem.
New members' activity
As I wrote in the intro, each community must be welcoming to new people and give them space inside so they feel comfortable interacting with others and feeling safe.
Below you can see the top three members that joined the community in 2022 and were most active among other new-joiners. Their activity was clearly reflected in their contributions. In the end, all the people you can see below are now members of the AsyncAPI Technical Steering Committee.
Social media is not a silver bullet. I think social media should only be a news interface with the community, not a marketing tool.
Below data will show you that these channels should not be treated as promotion channels that convert to new users. You will see that LinkedIn and Twitter brought together 2,082 users to the AsyncAPI website. Now, compare it to this single Mozilla documentation article that referred to our website 5,542 users.
Social media is suitable for communication with the community. Promotion should target people in a location where they look for help. In other words, mentioning AsyncAPI on a WebSocket-related website performs much better than hundreds of social media posts. So we should also try to make sure AsyncAPI is mentioned in other locations. Why not mention AsyncAPI in official docs for Kafka, MQTT, and other protocols?
I'm so damn tired of LinkedIn. Seriously, a platform that drains so much data from others and earns money from job marketing cannot afford to give users access to historical data and see more than one-year-old stats. Like seriously...
We had 2,139 followers at the end of 2022. Every year's data since account creation looks like below.
We are not active on LinkedIn. We only republish whatever we publish on Twitter. No engagement is created. We treat it as a news feed, although LinkedIn != Twitter.
Numbers went down a lot compared to last year, but when you look two years back, numbers are very similar. Why? In 2021, we spent $1,200 on LinkedIn marketing; thus, the number spike for one year.
My statement doesn't change. LinkedIn is an important presence, and even without dedicated moderation, simple reposting brings people to the project. We should continue our work there.
Forget about the number of followers. We have over 3.5k followers, over 500 more than last year. For me, followers number are as reliable as GitHub stars on a repository. Not much...
Let's see how Twitter's presence converts to visits to the AsyncAPI website.
My take on it. This is acceptable. I do not think we have to do more there. My observation is that tweets that perform there are not technical tweets that could bring more people to the project, but "thank you" tweets and "appreciation" tweets. And do not get me wrong, we should still create such tweets, no question about it. It proves it is a social media, not a "technical geeks knowledge exchange center." I do not think any additional investment in the platform makes sense.
We use Google Analytics on AsyncAPI website to understand the most valuable content for you and learn how the interest is growing. The data we get doesn't 100% reflect reality because AsyncAPI is for technical people, and many technical people use plugins in browsers to block Google Analytics or browsers that block it natively.
We also have numbers that confirm that. I checked the last seven days of data from Google Analytics and compared it with the raw traffic data we get from Netlify servers. The dates I checked were February 13-20:
- Google Analytics noted over 24k page views.
- Netlify noted over 26k page views.
The "quality" of the user visiting the website is much better. In 2022, we had no paid marketing nor any viral blog posts, and finally, data is more reliable and natural.
There is a much larger increase in the amount of Pageviews and Sessions but no significant increase in User visits. My theory: In 2022, we didn't have a single super viral blog post like in previous years that would bring a few thousand people in just a few days (like September 9, 2021 - over 5k readers), people that are interested in an article, and not AsyncAPI in general.
I think we can say that it is the first time we have had a regular increase in users and that this increase is valuable as these users generate a lot of sessions and navigate through the website a lot. In other words, the bigger difference between User visits and Pageviews, the better.
It would make sense to retake the above table and check it against Google Analytics, but not base it on the total number of visits and sessions, but the median for each year. In theory, this way, we would remove the traffic that is hard to predict from the equation, as it is not up to us to decide if something will be viral.
The more docs we get, the more traffic increase we will see
One of the top goals for the AsyncAPI website should be to increase the content and benefit from the SEO. This happened in 2022, thanks to participation in Google Season of Docs. We had six dedicated docs contributors. We gained new
Generator documentation in the website.
In the case of Generator documentation, Florence and Pratik refactored all the documentation from the Generator repository and made sure it was published in an automated way on the website. It was published on the website on the 20th of December, almost the end of the year during Christmas break, a dead season for the website. Yet the new content managed to generate an additional 1,053 Pageviews to the website!
Concepts were published to the website on different dates, some in September and some in early December. This new content generated more than 17,646 Pageviews in over three months.
As you can see, more technical content, which was not even released at the beginning of the year, converted into many additional pageviews. In 2023, this content will perform even better. I know it for sure, as only in the case of Generator docs, in January 2023, Google Analytics noted 4,668 views! Well, you must admit that 48k Pageviews for 2023 would be huge!
Google Search Console
In 2022 we didn't perform any specific modifications on the website to improve SEO. So the only explanation for the increase in impressions and clicks is the increased volume of documentation I wrote about in the previous section. More useful content, better ranking in search.
Like on other channels, we notice growing traffic in our GitHub organization.
In the last three years, we increased from 301 contributors in Q1 2020 to 1,520 in Q4 2022.
From 301 to 1,520 contributors for things like commits, pull requests authoring and review, or issues participation. In just three years!!!
This huge difference also has a negative aspect. More contributions mean more maintenance work.
Negative aspect of growth
Unfortunately, the number of maintainers does not grow as much as the number of contributors. This means that the difference between the number of submitted and resolved issues will get larger and larger.
Three years ago, we dreamed about having thousands of contributions. We forgot to balance that number with the proper number of maintainers. Now we need to react and make sure we also get a large number of active maintainers.
Our most essential packages are the project that provides AsyncAPI JSON Schema files, the Generator, the React component and the Parser. Below you can see how downloads of these packages increased.
You do not believe in this data? Lemme repeat what I wrote last year:
The increase in numbers is insanely huge. People automatically say that it is because of build systems, local development, or that some single corporation started using it at scale. In other words, the increase could result from just one project that uses these tools as a dependency.
Let me challenge you:
- Corporations always use mirrored versions in npm. When you work on a project in a corporate network, you will always hit the internal proxy. Not included in downloads.
- Most build systems cache packages downloaded are ready to improve performance. Not included in downloads.
- When you work with npm locally, you also have a local cache that is used if you want to install something that you already installed. Not included in downloads.
An interesting observation is that in 2021 AsyncAPI Parser was the only user of the AsyncAPI spec JSON schemas project (
@asyncapi/specs), as the number of downloads was almost the same. In 2022, the AsyncAPI spec JSON schemas project was downloaded five million times, three times more than the parser. This means many more tools are dependent on the project.
In 2021 we could see that the AsyncAPI Parser is not only a Generator dependency; today we see the AsyncAPI spec JSON schemas project is not only a Parser dependency. This means that the increase in traffic in different channels and the growing user base also reflect the number of tools built on top of AsyncAPI tools. One of the big users of
@asyncapi/specs is now Spectral. Also, GitHub tells us that
@asyncapi/specs is used in 1,329 public repositories
The AsyncAPI YouTube channel had a massive increase in traffic.
YouTube is just another channel where we can see the audience is growing. The conference and other regular live streams bring a lot of watchers. In 2023, numbers might drop because we do not want to do the online conference, but yeah, let's see.
We grow. Numbers just confirm our observation.
- If you plan to invest in AsyncAPI Initiative, don't worry. It is still an on-top project.
- It was the right call if you already invested in AsyncAPI Initiative. The community is growing.
We do not see any negative trends indicating that the interest in the project got stale and that the community is not growing anymore.
I also recommend below conference recording to learn more about AsyncAPI Community.
Photo by Hannah Busing on Unsplash