In July we put two questions to 28 people.

The first asked whether they were aligned with where we're taking the company. The second asked whether to hold off on a set of measures that could have included voluntary resignations, furloughs, temporary pay reductions, or layoffs.

Back in the spring I wrote this:

You can't vote on everything. That would grind the company down fast. But if people only vote on harmless or symbolic matters, then you haven't really changed anything. You've just dressed up the same power structure in better language.

Our first two votes were a policy update and a public statement we were being asked to sign. Neither one was going to cost anybody their job. This was the first time we put something to a vote that could directly affect people's livelihoods.

Getting to the Vote

Before anyone voted, we walked the team through the actual financial position with the numbers on the screen.

For context, StatBid is debt free, and we have always run the company cautiously. We are also in a period of significant investment in R&D as AI changes our industry and the work our clients need from us. We weren't making this decision in the middle of an immediate financial crisis. The question was how much risk we were willing to carry while continuing to invest and when, if necessary, we should reduce expenses.

This built on something I described in an earlier post. Over the last year, we started sharing a common-size financial statement with payroll broken into production labor, back office labor, and R&D. We share it so people can better understand how we allocate capital across the business. That context helps them make more informed decisions when we ask them to vote and gives them better information to identify problems and improve how we operate.

Roy recorded a video explaining where we stood and what we were considering, and we sent it to the whole team. I then booked one-on-ones with all 28 people eligible to vote so I could hear their questions and concerns.

We're small enough that I can still sit down with everyone. That obviously will not work forever. At three times our current size, those conversations would probably need to happen through coaches or smaller groups. But I don't want to lose what we got from the conversations just because the format eventually has to change.

The one-on-ones produced more than 60 questions. I removed the duplicates, grouped the rest by theme, and published all of them on a Confluence page.

People wanted to know how our financial triggers worked and what days of cash on hand actually meant. They asked what the different measures could mean for their pay, their teams, and their severance. They asked whether we're trying to run three companies at once, whether our version of democracy has limits, and whether Roy and I ultimately keep control.

I answered every question in writing. It took a long time. To be honest, by the end I was ready to publish the page and stop debating the wording.

Roy then went through the page and added his own comments. He pushed back on some of my answers and expanded on others. More people followed with comments, questions, and reactions of their own.

Nobody had to rely on my summary of what everyone else was thinking. They could read the questions and the discussion for themselves. The page is still available to the team.

The ballot was probably the smallest part of the process. If we had simply emailed two questions that Roy and I had written and asked everyone to click yes or no, I don't think the vote would have meant very much. Most of the team's influence happened through the questions and the discussion leading up to it.

We also kept the two questions separate. Combining them would have forced people to weigh their belief in the company's direction against the potential effect on their own compensation in a single answer. The alignment question used a five-point scale. The question about holding off on the measures was a straight yes or no.

We Still Had to Chase Down Votes

This was our third vote, and for the third time we learned that giving people a vote does not mean they will automatically participate.

I already admitted this after our first vote in the spring. Then we did it again on the most consequential question we've put in front of the team.

Part of the problem was probably how we distributed it. The announcement went into a Slack channel and nowhere else. One team member was traveling when it was posted, saw it late, and told me afterward that they assumed a decision this important would come by email.

That was a reasonable assumption. We had never established which communication channels should be used for decisions at this level.

About half the team did not vote until someone followed up with them. We ultimately reached quorum, but it took individual reminders to get there.

We clearly need a better process for announcing votes, explaining what is expected, and following up. Putting something in Slack is not enough.

What We Decided

The team expressed strong alignment with the direction, and the second question came back close to unanimous: hold off on the measures and reevaluate in October.

I didn't want to over read that result. The level of agreement made me look more closely at how we got there.

For one thing, the strategy the team endorsed was still written largely by Roy and me. The team voted after we developed it rather than helping develop it. I think involving more people earlier in that process is probably the next step.

Still, the comments leading up to the vote were substantive. People were clear about how uncertain the market feels right now and about the personal risk of waiting. They were also clear that they believe in where we are taking the company.

We also tried to be explicit that waiting has a cost.

If we make reductions later rather than now, there is a risk that we will have less money available to help people through the transition. A vote to pause meant accepting that risk. We included it in the FAQ before the vote, along with several questions where the honest answer was that we didn't know yet.

There was also a defined limit to what the team was deciding. The measures we were voting to pause were tied to one cash threshold, and there is a second, lower one underneath it. If we fall below that floor we move forward with the austerity measures regardless of the vote. The team was not being asked to take responsibility for every financial allocation decision, and we tried to make those boundaries clear before anyone voted.

Is the Vote Real if Roy and I Can Veto It?

A variation of this question came up during the process. Is this actually democratic governance, or is it a symbolic gesture when everyone knows Roy and I can reject any result we don't like?

It's a fair question.

Roy and I each own 50 percent of StatBid. We intentionally structured the company that way instead of giving one of us 51 percent and the final say. In practice, we have always worked to build consensus between us, and neither of us has ever regretted that decision.

Under our current ownership structure, though, Roy and I still retain the ability to override a vote. That is an emergency power and we treat it like one. The bar is that the company itself is at risk, not that we disagree with where the team landed. It also takes both of us, so neither one of us can reach for it alone. If it ever became an escape hatch for a decision we didn't like, the process really would be symbolic.

This was not a situation where we believed a veto was necessary.

In the longer term, Roy and I are working toward a more distributed ownership model that better reflects the governance structure we're building. We aren't there yet, and the tension between democratic governance and concentrated ownership is not resolved.

What I Took Away From the Vote

The 28 one-on-ones were a lot to fit into a short period of time. I expected them to be useful. I did not expect how much energy I would get from them.

I started every meeting with the same question:

"You're going to be asked to vote on the direction of the business. What information do you need to feel confident casting that vote?"

The responses were awesome. People asked hard questions about our finances, our strategy, and how we were allocating capital. They also brought forward ideas for improving the business. They weren't just trying to understand how the decision might affect them. They were trying to understand the business well enough to help make the decision.

That was where I could really feel how democratic governance might work at StatBid.

After the vote, Roy and I both felt a real sense of joy about what had happened. Some of that was relief. This was the most consequential vote we've attempted, and we made it through. But most of it came from seeing how seriously people engaged when they were given the information and a meaningful role in the decision.

Our team alignment and engagement measures have also held through the quarter, though I can't claim the vote caused that. They were already strong and we don't have an easy way to isolate the effect of the process. What I can say is that they didn't decline during a period when people had good reason to be worried given the nature of the vote being put before them.

We still have a lot to learn. However, I came out of the process with more confidence that democratic governance can work at StatBid and a better understanding of what it takes to make it work.