In just four years, Adam Pisoni helped grow Yammer from a team of five with an idea to a company of 500 worth a billion dollars. Today, he’s back to square one, working on a brand new company with just a handful of people — and one question dominates his mind: “What do I want to do differently this time?”

“Everyone sets out to build a great company where great employees all gets along, work hard and profit,” he says. “In reality, more commonly you’ll find yourself not trusting the people you work with and running around like madman trying to keep track of what everyone’s doing. You’ll end up slowing everyone down trying to make sure everyone is making the right decisions. You’ll feel like if you don’t have the pedal to the metal the whole time, the whole thing will collapse.”

At the heart of this problem is a widespread misconception of what it means to be a company leader. If you’re a founder, you’re probably laser-focused on building a great product. But earlier than you might imagine, you need to shift your thinking away from building a great product to building the company that builds the great product. It’s a subtle but powerful distinction. If you’re still required to answer every question, you’re never going to scale, Pisoni says.

He took the stage at First Round’s recent CEO Summit to talk about this disconnect in thinking that plagues too many founders and leaders, whether you’re at the head of a team or a company. He also shared the tactics that have worked for him to get out of the weeds and make the biggest impact possible with his time. These are lessons he plans to apply at his new company, Always Be Learning.

Conway’s Law and Finding Your Way Out of the Weeds

“All those times when I was making sure every little thing was working I was missing a whole universe of other things that I should have been thinking about. I didn’t even know they were there and I waited too long to find out,” Pisoni says. “Instead, I was running around trying to do too much myself when we already had dozens of employees. I was in most meetings. I was constantly being asked questions and providing solutions. But not in a good way.”

Here’s the real danger: As a leader, you’ll probably feel great about how busy you are. It feels like you’re adding so much value, when it’s just a sign that shit is rolling up hill to you too quickly.

He had no time to stop and think. But when one of his engineers brought up Conway’s Law, he froze and decided it was time for some introspection. Basically, Conway’s Law says this: Companies create products and services that are a reflection of themselves, the way they’re organized, communicate and work. Most often products and services get structured to mirror the way the companies producing them are structured.

At Yammer, Pisoni initially oversaw a small squad of engineers to build the original product. They all worked on the codebase, but as they hired more and more people that same original codebase became increasingly unmanageable and monolithic — just like the team. It was only when they started to break the engineering force into smaller teams that they were able to intelligently break up the codebase into smaller services. “The way in which we organized changed the shape of our technology,” he says.

When he was a web developer in the 90s, there was hardly any emphasis on process. Engineers would simply edit live files on web servers. There were no tests. No version control. “That would be absolute insanity today. Now we know that the only way to have a large group of engineers work together is to have a development methodology. But still too few companies lack an organizational methodology — a system or process, outside of engineering, that helps your company operate effectively without requiring you manually managing everything that happens.”

One area where Yammer got it right was continually iterating on its development methodology. The team slowed down as it grew, prompting them to question: “Why can a hack day team build something so much faster than we build on a day-to-day basis?” They realized they were working on too much at once and the communication about what they were building was slow and inefficient. People were stretched thin across multiple projects too much of the time.

“We ended up creating a rule where we wouldn’t even start a project until everyone who was needed could be dedicated,” says Pisoni. “Then we’d put them together in a small, autonomous team that didn’t require any permissions or dependencies to get things done. They’d work on a project for no more than 2-10 weeks, then that team would blow up and everyone would go work on other projects. That went a long way toward obsoleting the need for leadership to manage everything. For example, this system meant I didn’t have to dictate architecture or coding conventions because by having people rotating through the codebases so quickly, we ended up with emergent consistency.

As a leader, your goal should always be to build structures and processes that don’t depend on you and ideally don’t need you.

The Power of Cadences — and How To Bring Them To Your Company

So how do you build a system that allows big, important things to happen on their own without your managerial oversight? Pisoni’s number one piece of advice: Battle rhythms.

In the course of running your business, leaders have a bunch of work to do and decisions to make. But mostly, it happens randomly and in a reactionary way. There’s no rhyme or reason to when things get decided, when to shift strategy, when to reevaluate. You have no idea how much time and efficiency you could save by creating regular cadences around these actions. Cadences create alignment and transparency. It accelerates execution and — even though it sounds like additional process — actually ends up slashing the number of meetings and checkins needed. There are a few areas where the most momentum can be gained, and — incidentally — where the vast majority of startups run into trouble.

1. Roles & Responsibilities

“Most startups look at big companies and conclude that their rigid org charts and titles slows them down. And they’re right. That does slow them down,” says Pisoni. “The mistake comes where you think the best alternative is to have no structure, no titles, no clear accountabilities. This actually slows you down more! Without clearly defined accountabilities it’s easy for companies to become paralyzed by decision debt and arguments. I’ve never seen a company that scales to any size without realizing they need structure of some kind.”

You don’t need rigid titles and org charts to have clear accountability however. There’s actually something in between: Defining clear roles and responsibilities that aren’t tied to titles or the org chart. These roles should be temporary and flexible. As an exercise to come up with these roles, think through the decisions that have to get made now and in the future. Ask what are the questions that keep bubbling up — especially the questions that you and your leaders are always answering because it’s not clear who should decide? One of the best resources on this is the Citizen Code: Self Organizing Constitution. It contains a lot of good information, but the core ideas are these:

• Create a public document that anyone at the company can read which defines roles and responsibilities. For each role list: important metrics for that role, which person(s) hold it, how long the term for that role should be, and what areas of accountability that role owns.

• As a leader, commit to never undermining the people who are assigned accountabilities. You have to give them full control. Remember, if you make a mistake, it’s easy to change.

Once you’ve done this, make it clear that the roles you’ve just assigned aren’t permanent. It’s a role, not a title. It’s not an org position. “At Yammer, I was CTO. I had a VP of engineering and directors of engineering,” says Pisoni. “A question would come up such as, ‘Who owns onboarding of new engineers?’ Onboarding was an example of something that fell between the cracks of our org chart and titles. Whenever there was an issue that someone’s title didn’t directly correspond to, it would end up falling up to me. I’d feel great about being able to make quick decisions and move things along, but I didn’t realize how much those types of things slowed us down. If it had to go all the way up to me, it was because the lack of clear responsibility was creating problems.”

In retrospect, he wishes he would have defined roles tied to the major issues facing the business at the time and then committed to reassessing them every month.

Every month, Pisoni recommends having a Roles and Responsibilities check-in with the following format:

• People currently holding roles should give a quick retrospective on how they did in their role.

• Next, allow people to propose changes such as adding or removing roles which are no longer needed. Or suggesting new or modified accountabilities for existing roles. In some cases, the issue might not even exist anymore, in which case that role can be retired as the person moves on to new challenges.

• To decide which proposals to accept recommends a process called Integrative Decision Making. IDM allows a large group of people to deal with incremental changes and proposals without relying on rank to be the final judge. With IDM, anyone can make a proposal. Once a proposal is raised, there is a round of clarifying questions, a round of reactions, a chance for the proposer to make amendments, and finally a round of objections. The key is that the only valid objection is that the proposed change will cause material harm the business before it can be corrected. The idea of this process is to create a bias towards trying ideas wherever they come from. This works because even if it’s wrong, you can try something different in a month.

2. Monitoring Progress

“I absolutely hate status meetings. Worse than status meetings are status reports that roll up,” says Pisoni. “But then how do you keep everyone informed and aligned when things are constantly changing?”

Again, you don’t have to go to extremes as a leader. You don’t need tons of meetings and you shouldn’t get rid of all of them either. Pisoni saw how this might work firsthand while visiting General Stanley McChrystal, who formerly oversaw all American troops in Iraq and Afghanistan.

“He had this problem that is significantly worse than anything you’re dealing with — tens of thousands of people who needed to be precisely aligned in an environment that literally changed daily,” he says. “He ended up creating a process where every single day for 90 minutes, thousands of people from all over the world would dial in to a rapid-fire update meeting. Each group had 90 seconds to say what they did, what they’re doing next and what the risks or blockers are. Then there’d be a couple minutes for quick, clarifying questions.”

Pisoni recommends startup leaders do something very similar weekly. The goal of this meeting is not to solve problems or to figure out who should be doing what. It didn’t make any sense to try to solve problems with that many people in a limited amount of time. General McChrystal realized that all the real value would take place in the conversations during and after the daily call. The real challenge was giving everyone the information they needed to even have these conversations at a regular cadence. During the meeting, people can be chatting in a backchannel chat. After the meeting people will reach out to each other to confirm what they heard and ask further questions. He recommends the whole company be invited to listen in or participate, making it a truly cross-functional meeting.

McChrystal himself realized that his job in these meetings wasn’t to solve people’s problems. Rather, it was to model the behavior of finding the best person to solve any given issue. He never claimed to have all the answers or to be able to make every big decision. He would take his important conversations offline and encourage others to do the same.

It’s vital that everyone in your organization has an accurate and constant sense of the progress you’re making so that they know when they can and should jump in to help, or when their expertise might come into play. Putting a stake in the ground and saying, ‘This is where we’re at,’ lets many distributed pockets of authority work smart without relying too much on central authority figures.

3. Retrospectives

On a good day, a startup leader can only hold a few vital goals in their head. One of them should always be to stop repeating mistakes.

Unfortunately, the most reliable way to do this is to get involved in everything to make sure people stop making the same errors. This doesn’t scale. So, what can you do instead? Hold regular retrospectives. Instead of teams holding their own private retrospectives, he recommends having a company wide retrospective meeting where any teams that finished a project or managed a problem over the past month can present. If you do them frequently, they don’t have to be as detailed. Done well, each project can be limited to 5-10 minutes. It’s the practice of it that has the impact.

Another quality of successful retrospectives: Focusing them around hypotheses. Don’t just use the time to look back at what you did. Really get into the nitty gritty of whether it worked or not. And even if it did work, could it have been better? “Always distill what hypothesis you were trying to test. What did you learn from the success or failure of that hypothesis?” he says. “Beyond that, what worked about the process you used? What didn’t work?”

Using metrics during these sessions is important. If you hold retrospectives, you want to look at key metrics for the projects in question — and they should be outcome based, not output based. For example, a good metric might be customer retention. A bad metric would be number of powerpoint decks created. That’s not an outcome-based metric. Outcome-based metrics are the ones that matter to your customers.

Finally, consider what you would do differently next time. “The way I like to ask this question is, ‘What weren’t we doing that we should start doing next time?” Pisoni says. “What were we doing that we should stop doing? What should we continue doing?” Those are the few, salient ideas you want to capture from everything you do.

4. Planning and Prioritization

Many startup founders are so embroiled in the present moment that they don’t look ahead to shifting course or switching up priorities. But it’s inevitable. The worst thing you can do is lock yourself in a room with a few other leaders at your company and emerge with a whole new set of projects and priorities. People will freak out and feel left out. They won’t take on accountability with the same enthusiasm. And every time you do this, they’ll be slightly less bought into your vision and long-term success than they were before. Part of the reason the organizational methodology of battle rhythms works is that it gives you enough evenly-spaced check ins and forced communications that you won’t fall into this trap. But you still need to be mindful of it.

Pisoni recommends having an all day, or even multi-day planning and prioritization meeting at least once a quarter. This meeting is part uber-retrospective and part planning and prioritization. What did we learn last quarter? How should that influence our strategy moving forward? Then, global priorities should be set across the whole company. Everyone should be focused on the same high level goals. Maybe it’s growth this quarter or revenue. Finally, based on your new priorities, decide what you’re going to STOP doing.

It’s surprisingly difficult to tell people to stop doing something. Here they were just super invested in their work on a task, and now you’re telling them to desist. People don’t let go easily, especially when there’s insufficient explanation or gratitude. They want to make an impact. This is just one way that change will create tensions within your org.

5. Managing Tension

No matter how good you are and how tightly aligned your team is, there’s going to be tough, ongoing tension within your company, and you can’t ignore it.

Two weeks ago, Pisoni met with the founder of a small tech company that said, “You know, our sales department is frustrated because they want engineering to build features that help them close deals. But engineering are prioritizing growth features instead — so they’re just working on stuff for the top of the funnel like virality. This is a tremendous tension within the company.”

If he’s learned one thing from his experience at Yammer, it’s that you can’t talk people out of tension. Trying to do that is a huge mistake almost every time. You can’t go to sales and say, “No no no, it’s not what you think… product and engineering respects you, they just don’t have the time now, but it will be fine.” That’s only going to make things worse. So what should you do?

“There this process that I recommend, where instead of digging in and trying to resolve tensions as a leader, your only goal should be to make tensions public,” he says. “You’re just going to say, We’re going to have a regular meeting at a regular interval where we’ll outline what the big tensions are.’”

Maybe you decide to integrate this into one of your other standing meetings — like a status or planning meeting. It’s as simple as listing them out. You just want everyone to acknowledge that they are happening. No one should try to talk anyone out of any tension they express.

“The beauty of this is that you don’t have to fix all of the problems that come up at this meeting,” says Pisoni. “Now you have the permission to say, ‘Look, we have to focus. We can’t deal with all of this at once, so let’s pick three tensions we’re going to work on.’ Suddenly, you as the leader aren’t responsible for all the tensions all the time. You gave people the direction and momentum to deal with the top issues on their own, and you can say that the rest will be handled when they come up in future meetings. In most cases, tensions will resolve before you ever have to prioritize them.”

Implementing this advice isn’t going to be easy, but it is the fastest way to make big change at your company. It’s an immutable truth that humans are resistant to change. They don’t like trying things that are unfamiliar or uncomfortable. As a leader, you shouldn’t be handling all the problems at your company, but you do need to be able to weather this storm as you distribute decision-making and scale your capacity. That’s in fact most of the challenge to running a quickly growing company. You need to increase your capacity faster than you increase headcount with the lightest weight structure possible.

As a leader, you can’t make decisions as fast as you’re faced with them. You’ll just accrue decision debt and think it’s your job to get out of it by making decisions faster and faster. Instead, make it your job to build structure so decision debt doesn’t accrue in the first place.

“You’re ahead of the game if you recognize that solving the problems you have today are going to create new problems tomorrow. You’ll find solutions to those problems and the cycle will go on and on ad nauseum until you die,” says Pisoni. “That might sound depressing, but once you get used to it, it actually takes tremendous weight off your shoulder. You realize it’s fine. That’s how life is. It’s not that you’re constantly failing. You’re constantly reframing and moving forward.”