After 2 months of hard work, we were super happy to show the world Airbyte’s MVP on 09/24.
When we say MVP, we mean it. Airbyte is an open-source data integration platform that syncs data from applications, APIs & DBs to warehouses. So our value proposition is directly correlated to the number of connectors we have. On MVP launch date, we only had 3 source connectors and 3 destination ones. And potential contributors couldn’t build integrations with us yet.
We want to share with you what we learned during this very early launch, and, consequently, when sometimes you should go towards a soft one (vs. a hard one - see differences below).
What is a soft launch?
A soft launch is when you want to reach out to a finite number of people for your launch. It can be done via email outreach, social posts, or community groups.
A hard launch, in contrast, is about trying to get as much exposure as possible. So you want to post on ProductHunt, HackerNews, get some PR on TechCrunch, and reach out to influencers in advance so they can put you forward on the launch date.
Why would you ever opt for a soft launch?
We had a choice between releasing very early but without a big value prop, or releasing a bit later with more connectors and better value. We opted for the former with a soft launch: our own social networks and Bookface (YC’s internal forum). We didn’t post on HackerNews or ProductHunt, and didn’t try to get any PR, or any influencers’ support.
The main reason for that is we knew our number of connectors was very limited, and therefore our value proposition was, too. Plus, our goal would be to get as many new contributors to our open-source project as possible. But we didn’t have the framework for new contributions yet. You don’t have many shots at successful hard launches, so you’d better be prepared for that. Don’t do a hard launch unless you know your product resonates with your audience, which means a good activation and retention rate. And make sure your launch publication avenues are targeted to your audience.
So why even do a soft launch? We wanted to:
- know if our mission resonated with our audience
- know if people would be interested in contributing
- strengthen our knowledge about what companies use, and their issues with existing solutions
- get more companies in our lead pipeline, to better prioritize our list of connectors to support
A startup’s job is to learn as fast as possible, in order to get to product-market fit faster. The soft launch was a way for us to learn a great deal, so that we could adjust whatever needed to be changed to get to the PMF faster.
So what was the impact of our soft launch and what did we learn?
Community growth and activities for our 0.1.0 soft launch
Here are a few metrics:
- 42 new newsletter subscribers
- 120 new stars, 7 new forks and 8 new watchers on GitHub
- 38 new Slack users
- 200 new Twitter followers
This may seem small, but we got all this from just 2 LinkedIn posts from both founders, and one post on YC’s own forum.
We use the Orbit model to track the growth and engagement of our community. We’ll write more about our KPIs in a separate article. But in a few words, we segment users in 4 different orbits: observers (orbit 4), participants (orbit 3), contributors (orbit 2), advocates (orbit 1). The lower the orbit, the more engaged the user is. The orbit level is defined based on the actions performed by the user (subscribe to newsletter, follow on Twitter, fork on GitHub, open an issue, or a pull request on GitHub, etc.). Here is what we observed with the soft launch.
There was no orbit 1 yet for a simple reason: our MVP doesn’t let contributors contribute in terms of code (pull requests) yet. But we had quite a few people who wanted to contribute, which was a good sign for us! They just need to wait for the next version, which is coming in a few days.
Here is our activation funnel when our users deploy our app. Most of our users opt for anonymous metrics, which still enables us to improve the product without tracking them personally. The soft launch showed us that we need to address a serious drop when users need to specify the metadata for their sources. Our next version will offer a connector with pre-filled demo data so users can see that Airbyte works with little effort.
Without the soft launch, we wouldn’t have identified those issues and would have had our hard launch with both drops; that would have been a big missed opportunity.
Getting on the map for domain experts and active contributors
The soft launch exposed us to a lot of new teams we were not in contact with before. It really increased the top of our funnel, and above all, we were now a dot on the map. People started to mention us in Slack workspaces in conversations about open-source alternatives to Fivetran and StitchData, and to notify us. We had an open debate in the DBT Slack. We started to exist and to get more and more feedback, and new perspectives for us to strengthen the way we describe our project.
We now believe more strongly than ever in our mission and our technological choices.
One last thing the soft launch did was to increase our impetus: it moved us forward in a lot of discussions, from mere awareness of our product to consideration to contributions with some teams that hadn’t considered it before. This brings us to our next steps in the coming weeks.
Our next steps
1. Contributing and using their own connectors
Our first priority is to create the best developer experience (DX), so they can start contributing and using their own connectors directly through our UI or API. We will have a first iteration on that in mid-October.
2. Multiple destinations
Airbyte’s 0.1.0 (MVP) version only supports one destination. It would have taken us one more month to support multiple destinations from the get-go, so we chose to simplify the product as much as possible to get on the map as soon as possible with our soft launch.
However, this is something that we want to support before our hard launch, as a lot of teams we talk to use several destinations. They could have several Airbyte instances, but that wouldn’t be the most convenient. We want Airbyte to be perceived as the solution every team is dreaming of when they think about their data integration issues.
3. More and more pre-built connectors
We will focus on the most used connectors, and especially destination ones. Having very high quality connectors for all the data warehouses is of the utmost importance to our users. This is the same for all the main source connectors (Salesforce, Facebook, Google, Hubspot, Marketo, GitHub, etc.).
Some will be built by the community, some by us. In any case, we will be very active on the maintenance of all of them.
With the MVP, our goal was that it would only take you 2 minutes between deploying Airbyte and moving data. So we only addressed the case of a single node. But companies need us to support multi-node, and this is part of the next few months’ priorities, too.
Continuous iteration and improvement along the way
Of course, along with these milestones, we will keep iterating on the product to improve our activation rate (onboarding, typically) and other adoption metrics. Whether a product is open-source or not, it is still a product with a UX (or DX, in our case).
We hope you might consider more soft launches in your roadmap. They are great intermediary steps to give yourself higher chances to succeed on your hard launch.
Don’t hesitate to subscribe to our newsletter to learn about our progress, our community and our journey!