Source & Binary has ceased operations

Sad news: Well, not news. Sad slightly dated information: Source & Binary has ceased operations. Substrate 2024.08 was the last official release of that software under Source & Binary’s stewardship. Travis and I have jobs. (More on these last two topics in future articles.)

Also: Not actually that sad. Being an entrepreneur is not central to my identity. The relief I feel not being on the hook for sales and marketing is wind beneath my wings.

I have a lot of deficiencies as an entrepreneur. I don’t really understand the difference between software and a product. I can’t tell a good go-to-market strategy from a bad one. I can’t even easily define what a go-to-market strategy is.

Nonetheless, and if for no one else but me, I think it’s important to account for Source & Binary’s successes and failures as best I can.

The business

The consulting practice that predated focusing on Substrate as a product was actually pretty successful. If I’d stayed on that course in 2022 I’d probably still be on that course now. But the idea of being a product company was too enticing to ignore so here we are.

At our peak, we had eight paying customers. On the one hand, that’s not very many and it wasn’t enough for Source & Binary to remain a going concern. On the other hand, I worked hard for those customers and am proud of what we did for them. We lost one customer when they were acquired and we couldn’t convert the parent company. We celebrated their “graduation” and tried to imagine how we could’ve expanded into the parent company. Otherwise, folks seemed happy.

Our operating expenses were near zero. We were two members of an LLC not taking salaries and paying for a modest selection of SaaS. Despite having hundreds of AWS accounts, we were mostly running near-free serverless control planes and using free AWS services like IAM and Organizations. Our COGS was effectively zero, too, since everything ran in our customers’ AWS accounts (where they’d want critical systems like Substrate to run).

We had no investors pushing us to swing for the fences, bet the farm, or pivot. (Perhaps we could’ve used some of that tough love.) But we also had no investors to repay, so the decision to cease operations was entirely our own.

What really drove the (lack of) health of our business, though, was this: We never discovered a repeatable sales and marketing strategy.

The Hail Mary

In the fall of 2023 we started to set deadlines for closing new customers. We ramped up our content marketing and cold outreach. We engaged outside help with messaging, pricing, and packaging. For all we did, we didn’t do anything useful because we closed zero new customers.

We also considered adding a second product, one that we would fund with venture capital, and had dozens of customer development meetings to test some ideas. Nothing convinced us we had a good enough idea to commit to working on it for ten years and being responsible to employees and investors.

On our way to getting jobs we tried to arrange an acquihire, as much for the experience as anything, since it seemed like such a longshot. No bites. Sometime in April or May we decided to throw in the towel.

The responsible dissolution

The day we decided to cease operations, we turned off recurring billing and figured out how long we’d already been paid to support our customers. Most folks paid quarterly and, through sheer luck, no one had recently re-upped an annual contract. We were obligated to operate through August.

Before the next Substrate release we removed the minimal, tasteful (defensible!) telemetry that Substrate collected. This was as much a matter of good Internet citizenship as anything. We didn’t want to be responsible for zombie Internet traffic if we could help it.

More substantially, we committed these final few releases would provide an optional off-ramp for Substrate users who want to switch to AWS IAM Identity Center. We knew that, despite the terms of the Substrate license being specifically designed to protect customers from having to start all over, some folks would want to migrate from no-longer-supported Substrate to very-aggressively-marketed AWS IAM Identity Center. (More on this in a future article.)

With customers sorted, we were left with the actually monumental (and ongoing!) task of dissolving a company. First item of business: Closing hundreds of AWS accounts. You should have lots of AWS accounts but you should not have this many AWS accounts in this many organizations. The month-scale rate limit on closing accounts is hateful.

All the rest of the SaaS was easy to stop paying for, though I am still kicking myself for not exporting our Slack history before turning it into a free team. I actually kept the Google Workspace because I’d been serving my personal mail out of my work mail for four years and found it easier to just keep going.

The last steps will be formally dissolving the LLC, distributing the remaining profits, closing the bank account, and filing our 2024 taxes.

The failures

Each of our customers got something slightly different out of Substrate. It was a crutch to some, a power tool to others, a cheat to still more. This could have been fallout from the consulting practice but, no, even customers who were never clients saw wildly different benefits to adopting Substrate.

There was one customer we considered to be the embodiment of our ideal customer profile. They were enthusiastic users, upgraded pretty often, and used it daily. Unfortunately, their use of Substrate turned out to be expedient and, though I don’t believe they’ve moved off Substrate, not at all sticky. Substrate was a nice-to-have.

“Doing AWS right” turned out to be poor motivation for everyone except the handful of us who’ve been burned by the one big AWS account before. “Better” wasn’t enough. Engineering time saved was a poor unit for measuring Substrate’s value.

I chose a too-small problem to solve.

The future

I have a long and storied history of mistaking the cool, interesting, and even useful thing I want to build for something marketable and sellable.

I may have learned my lesson this time. I’ll stick to jobs from now on.

This article is part of a series on Source & Binary, my company that operated from 2020 through 2024.