Hi all,
We’ve decided to start doing a monthly update for the Bubble community. The goal is to give the community some more insight into what’s going on at Bubble: what we’ve been working on and what’s on our minds.
Emmanuel and I haven’t been as present on the forum as we were in the earlier days; as we’ve grown the team and community (we just crossed the 400,000 signups mark!), our time has gotten spread pretty thin. It’s been great to see the rest of the Bubble team pick up the slack, but I think it’s important for us to periodically touch base with you all. Hopefully this post gives some insight into what’s going on with the company, and I plan to periodically check the replies for the next few days and try to answer questions where I can.
Changes we made this month
-
We launched a series of tutorials on how to build popular websites using Bubble
-
We launched a podcast where we interview Bubble users and other no-code innovators
-
We launched a series of bootcamps to help people ramping up on Bubble accelerate their learning
-
We added the ability to upload large files, which involved overhauling our file upload infrastructure
-
We now support A records for customers using CloudFlare. These are anycast records, which means that users can now host or redirect from the root (i.e., with the “www.”) of their domain in robust way, and is an important step towards making our overall hosting infrastructure global and world-class.
-
We releases a number of follow-up improvements to our option sets feature
-
We added a page for viewing the progress of bugs, and changed our team’s internal processes to keep it up to date.
We also made a some important behind-the-scenes changes:
-
We partitioned the database tables we use to store the change history for application data; this allows us to grow them and clean up old data in a scalable way. We also improved our monitoring capabilities for storage usage in our main cluster. This is important to keep our main user databases healthy as we get more and more customers.
-
We recently changed the way we handle incoming support requests. We now have the success team take turns playing separate roles: some people are quickly triaging requests and replying to the ones that can be answered in a minute or two, while others are doing deep dives on the ones that require significant investigation or debugging. We think this will help us respond faster to the easy requests, and help the people working on difficult bugs work more efficiently and get to the bottom of them quicker.
-
We switched the provider we use to monitor our infrastructure. We collect thousands of metrics streams on all aspects of the health of our servers and databases, which we use to keep our infrastructure healthy and respond quickly to problems. We switched providers because of features that allow us to better organize our metrics as we add more systems to keep track of, and to monitor more things in cost effective manner so that we can maintain high visibility.
This month in numbers
-
Total customers who reached out to us through bug reports or support@bubble.io: 1,408 (up 52% from last 30 days)
-
Total received messages: 4,744 (up 64% from last month)
-
Average response time to messages (4h 30m counting only our work hours; 12h 27m in absolute terms)
-
Total bug reports: 508 (up 116% from last month)
-
Bug fixes that we deployed: 35
Things on our minds
-
We’ve received a number of reports from users that they’ve had elements or workflows disappear while using the editor. We always get some number of these reports, and many of them are due to people accidentally deleting something and not noticing til later. But in the last couple weeks the volume has been higher than usual. These issues tend to be hard to investigate, because we have to look at our logs and guess what might have happened, so we don’t have definitive answers yet. Our best guess is that our recent changes to our version control and deployment feature might have led to the increase in these reports. We’re actively working on a project to make our version merging more reliable, so hope that will solve the problem, but we’re also investigating several of the most recent bug reports to make sure there isn’t something else going on.
-
This month was a fairly bad month in general for production breakages and downtime. Several of these were self-inflicted: we pushed bad code, which either caused an outage or broke an important feature on user apps, and had to roll back. Others were caused by scaling pains as we change our infrastructure to handle the growing volume of usage. And we also had some caused by issues with upstream infrastructure providers. We’re worried about the negative trend, and thinking about ways of mitigating it. One thing we’re doing is investing more time in building automated tests: that’s an important part of making sure we don’t make the same mistakes twice, although it’s not a cure-all because it’s easy to test for things that you anticipate but hard to test for issues you weren’t thinking about. Another idea we’re discussing internally is rolling out code to users on our free plan prior to rolling it out to customers on our paid plan, so that people with apps that are in production are less likely to be impacted by bugs. We don’t currently have the ability to do this, but an infrastructure project we’re working on that we hope to complete in the next couple of weeks would open up that capability. Finally, we’ve noticed that a lot of bugs keep coming back to several parts of our code that aren’t very high quality. We keep on doing small patches and fixes to them in response to user reports, but often those patches actually cause new bugs. So we’re planning on prioritizing spending the time to just rebuild those parts of the code in a higher-quality way so that we fix the problem for the long-haul.
-
Relatedly, we’ve been thinking about ways to build trust in our bug reporting system. We’ve been doing some reflection on this and realized we haven’t been doing a good job as a company of telling users who report bugs to us through [https://bubble.io/bug_report] what’s going on with their report; we’ve heard a bunch of users tell us that they feel like the bug report form is a blackhole and things disappear into it. It’s important the community trusts that medium of communication, because the quickest way of us getting to the bottom of a situation is if we get a detailed, clear report, which often involves confidential details about a user’s app that can only be shared through a private communication channel. So it helps us respond more quickly if people’s first reaction when they see a problem with Bubble is to send a report in. But that will only become a habit for the community if it gets rewarded with a fast response and clear communication. The bug progress tracking page I mentioned above is our first step in doing this; we’re also making changes to the way we triage and handle bugs to try to up our game here.
-
Another big thing we’ve been dealing with is the impact of COVID on our company and team. We’ve been relatively lucky in the grand scheme of things: all our work can be done remotely, so far no one on the team has had serious health issues, and while we’ve lost some of our customers because COVID forced them to cut costs or shut down their business, it’s been a small percentage of our total users, and we’ve gained more customers than we’ve lost over the past month. That said, transitioning from an all-in-person company to an all-remote company has been a big change, and we’re still getting used to communicating with each other entirely over chat and video. Several of us have been stuck in small NYC apartments by themselves for over a month now, so we’ve been doing various things to socialize remotely as a team to try to keep everyone from getting too isolated. We’re also monitoring the larger economic situation, and trying to trim back on unnecessary costs and improve our internal financial reporting, so that if we start losing even more of our customers we’ll be able to keep the business healthy.
What we’re currently working on
In addition to the initiatives mentioned above, we’re actively working on the following:
-
We’re alpha-testing a feature to allow importing much larger CSVs than our current CSV import feature can handle.
-
We’re alpha-testing a videochat plugin to make it easier to build apps that have video chat as a big component.
-
We’re alpha-testing an integration with Google Optimize to make it easy for users to do A/B tests in Bubble.
-
A complete rebuild of our system for doing calculations on top of application data. The current system can sometimes result in the calculations getting out of sync with the app, which can cause things like elements disappearing or pages not rendering correctly. Some of the recent bugs related to our version control feature have been from interactions between that feature and this system. This is a big project, but unfortunately the only way to completely eliminate this category of bugs is to replace the system with one that works using a different overall approach.
-
We’re in the process of rolling out a connection pooling layer in front of our main cluster databases. This is big infrastructure change (and was the source of a couple of our more recent outages), but it’s critical to let us keep scaling up our systems. Right now, one of our big performance bottlenecks is that we’re unable to add as many servers to the main cluster as we’d like because of pooling limits, which leads to slower response times throughout the system. This infrastructure change, once complete, will unlock that, which is a major victory in making Bubble faster and more scalable.
-
We’re also planning out a project to be able to seamlessly transfer apps between databases in our main cluster without causing disruption to the app. This would allow us to rebalance databases when one database has too much activity on it, which would enable us to have more even, predictable performance.
-
We’re doing an overhaul of the code that powers the Input element; this is one of the areas I mentioned above where we have a constant stream of bugs, so we’re investing time to improve the overall reliability.
-
We’re about to start working on an upgrade to our SSL LetsEncrypt integration. This won’t add any direct user value, but if we don’t do it this month, the current system will stop working.
-
We’re working on a complete redesign of our editor. This involves changing the visual appearance, moving components around, and smoothing out some of the jankier interactions. This is a huge project that’s been going on since last year and that we don’t expect to finish before the summer (we currently have two and a quarter engineers working on it almost full-time).
-
We’re doing a ground up overhaul of all our educational and reference content, including the reference, manual, and video tutorials. This is a big, multi-month project to get our learning resources to the next level.
-
We’re continuing to scale up our customer success team and professionalize its operations
-
We’re hiring for our growth team to help us connect with the entrepreneurship community and support our customers
-
We’re sketching out changes to the design of our homepage to make our branding more modern and appealing
As always, thanks for all the contributions you’ve given to the Bubble community – we appreciate all the community does to help each other out and to support Bubble. We look forward to seeing what you build in May.
– Josh and Emmanuel
.
Congrats on the 400,000+ sign ups! 
thank you!