Your business can only grow as fast as its narrowest point. Here’s how I find that point in my companies, fix it fast, and build a team that does it without me.
A couple of months ago, one of my companies hit a wall.
Revenue wasn’t crashing. It was slowly sliding, which is worse, because nobody panics and everybody stays busy. We ran the process below, and we just finished a record month.
I learned this years earlier as a software developer. I could write all the code I wanted, but if nobody bought it, the business had no money. More code felt safe because it was what I knew how to do. Most teams do the same thing: more of what they’re good at, whether or not it’s what’s holding the business back.
Your business is a pipe
Ideas become products, products get marketed, people buy, the team delivers. Now imagine one spot in the middle where the pipe gets pinched. Your team pours water in on the left, working incredibly hard, and only a trickle comes out the right. The whole business runs at the speed of that narrow point.
So when an owner tells me they need more marketing, more sales, more automation, some AI, my first question is: where’s the pinch? If delivery can’t onboard new clients, faster marketing just piles more people up in front of the same wall.
How to spot it: walk down the process and look for where work is piling up. The step right after the pile is usually your constraint. It could be:
- You, approving every invoice.
- Someone on your team waiting on someone else before they can move.
- Decisions and approvals that live in somebody’s head instead of a process.
- Customers who only want to deal with you, because the sales script never told them they’d be working with your team.
That last one is a little bait and switch, and it shows up more than you’d think.
There’s only one problem in your business that needs solving right now. Find it, solve it, repeat. Solving the right thing in the wrong order is still a loss. You don’t polish the car while the engine is on fire.
Check the numbers before you fix anything
A buddy of mine came to me a few years ago and said, “I need to redesign my whole program.”
“Why?”
“We lost some customers.”
“How many last month?”
“Three.”
“How many do you have total?”
“Three hundred.”
That’s 1% monthly churn. You can run an incredible business at 2 or 3%. So I asked him how many people he had on his marketing team.
“Honestly, Dan? Nobody.”
He didn’t have a retention problem. He had a marketing problem, and he was about to spend three to six months rebuilding a program that worked, which could have cost him far more customers than the three he lost.
Every bottleneck is a hidden KPI. Measure the right things and the business tells you what to fix. Measure nothing and you’re guessing. This is the theory of constraints, written for manufacturing. I picked it up from my dad, who worked at Siemens, where they brought in Six Sigma and lean manufacturing experts. It was shop-floor language in our house.
1. Map the steps. A process that isn’t defined is a process you can’t fix. Take a sales workflow: leads, qualify, book a call, show up, close. Have someone on your team draw it on a whiteboard with each step as its own box.
2. Ask the magic question. This is my favorite question in business, because it gets to the problem fast:
If I 10x the volume going into this workflow, what breaks?
Push ten times the leads through that sales process in your head and the weak spot jumps out. In the example I’m thinking of, show rate and close rate were fine for the industry. The break was booking: plenty of qualified people, too few getting on a call.
3. Put a number on every step. Without a spreadsheet showing each step and the conversion between them, you can’t know where it’s broken. I walked into one of my offices recently and counted six dashboards in one room. You don’t need the TVs. You do need the spreadsheet, and AI can help you build it. To know what “normal” looks like, ask your AI for the industry benchmark at each step, and make sure you measure it the same way the industry does.
Pro tip: I have an AI agent in GrokBot called Business Constraint. Every Sunday and Thursday it analyzes the data, the processes, and even the meeting recordings from every company I own, picks the number one constraint in each, posts it in Slack, and tags the people involved to reply if they agree. When they don’t, we talk. You can build the same thing in Claude or ChatGPT, but only once the workflow is mapped and the data is flowing.
READY TO FIND YOUR BOTTLENECK?
The free Scale Your Business Workbook walks you through the whole theory of constraints process, and how I run it in every department of my companies.
Get the free workbookSurge on it
Once you know the bottleneck, attack it as fast as humanly possible. The first thing in the way is usually an assumption.
I was at a conference recently and heard Steven Bartlett tell a story about an animator on his team who’d been editing for five years. Someone finally asked him, “What would it take for you to double your output every week?” Everyone braced for some big, elaborate process change.
He said, “I’d just need a new laptop.”
Nobody had ever asked, and he’d assumed nobody would buy him one. A $2,000 or $3,000 fix that doubled his output had been sitting there for years. Challenge every assumption around the bottleneck. The fix is often easier than you think.
Last year my venture studio, where we launch a new AI company every month, hit a talent bottleneck. So we surged on it. There were more than 2,000 people in my Instagram DMs who’d said they wanted to work with me, and nobody had looked. I gave the whole team an incentive to dig through their email and chats, find the top candidates, and move them through the pipeline like sales leads. We paid a $5,000 referral bonus, even when the person came from my own audience. I needed the talent.
Within six weeks, that bottleneck was gone. One bottleneck, one focus, fixed fast. That’s a surge.
Most owners are spread across ten problems. Drop nine and go all in on one, and it gets solved.
Here’s how I run one.
1. Name it precisely. In the talent example, “we don’t have enough people to hire” sounded like the problem. The real bottleneck was “we don’t have enough high-quality applicants.” The name decides where the effort goes.
2. Pull capacity onto it. If not enough people are booking, maybe the marketing team stops marketing for a bit and gets on the phones. It sounds crazy, but everyone else is busy while the constraint starves. Have everyone do a little less of their own work so the constraint gets the resources.
3. Swarm it with the 4 Cs of leverage. A mentor of mine in San Francisco taught me that there are only four ways to create leverage on a problem:
- Capital. Is the problem funded? The laptop was a couple thousand bucks.
- Content. Is the process written down? If people are calling leads with no script, the system is broken. Record a training video once and it pays forever.
- Collaboration. Friends, partners, other owners you know. Don’t limit yourself to your own team.
- Code. Software, automation, AI. It’s my favorite, and it goes last on purpose. Too many people are burning tokens on AI aimed at the wrong problem, or in the wrong order.
4. Attack until it’s obliterated. Then give it a DRI, a directly responsible individual, who watches the metric so it never comes back. Nothing’s worse than fixing a bottleneck, moving on, and watching it fall apart because nobody owned it.
Pro tip: when someone tells me a fix will take two or three hours, I ask, “What are you doing for the next two or three hours?” If it isn’t more important than this, go do it now and we’ll review in three hours to see if it actually moved anything.
A try is not a test
You can surge for weeks and still not fix anything, because of how you measure.
When a team tells me, “We tested it,” I usually hear “we guessed.” If you didn’t write down the one variable you changed, it was a try. Let seven people make changes at once and nobody knows what worked, and the problem quietly slides back a few months later.
Here’s the method.
1. Make an observation. If we double the leads, will revenue go up? Everyone on the team can throw ideas in here.
2. Form a hypothesis. In my world, that’s a guess you’re willing to write down.
3. Run a controlled experiment. Change one variable at a time and put enough volume through it to mean something. Flip a coin three times and get three heads, and you haven’t learned it always lands heads. Flip it 50 times and you’ll see 50-50. A lot of “I tried it and it didn’t work” comes down to this. How many times did you try?
4. Analyze it every day. During a surge the team meets first thing every morning to look at the data. On a big problem we do it twice, morning and right after lunch.
5. Decide. Enough data, and do we agree? Deploy it. If it didn’t improve anything, reverse it so the next change is isolated. Leave every failed change in the system and you’ll wake up to a ball of complexity. In marketing, the winner becomes the new control, and everybody plays the game of beating it.
Then ask the bigger question. Did it make more money, raise gross margin, get more people on calls? Clearing a bottleneck only counts if the outcome you were after actually moved.
You might be the bottleneck
You can have great people, but if they don’t know how to find and solve problems themselves, you’ll always be the one who has to. That’s where most owners get stuck, and it’s why I wrote a whole book about learning to let go.
My media team knows this process. Last year, pushing toward 400 or 500 pieces of content a week, they realized sourcing the right B-roll was a bottleneck. So they built custom software called Turbo B-roll. It indexes a few hundred terabytes of our footage, and anyone can search “Dan with wife on beach” and get the clip. It doubled their production throughput with no new hires. I was zero involved. I found out when they showed me.
Early on, I managed by telling people what to do, checking it got done, and telling them what was next. Then a mentor who’d built a 5,000-person organization told me, “That’s the bottleneck. You.”
Here’s how I fixed it.
1. Define the outcome. What does done look like? I describe it visually and emotionally. Before a big event I’ll tell the team how an attendee will feel, what they’ll learn, and what they’ll tell their own team afterward. That picture becomes the North Star.
2. Hand over spending decisions. I use a financial trust ladder. Depending on the role, people can spend $50 to fix a problem, $500 for team leads, $5,000 for directors, and up to $50,000 for my executives, all without my approval. They just tell their leader afterward.
Whatever people spend money on is usually the real bottleneck. One of my editors didn’t want to wait on the designer, so he spent $50 to get a design done himself. Turns out the designer was on contract at 10 hours a week and didn’t have the capacity. We gave him 20. Bottleneck solved, and I only found it because the frontline person was allowed to act.
3. Measure it. Every metric on my dashboards has the owner’s name next to it, plus their cell number. One walk past a dashboard tells me where to ask better questions.
4. Coach. Someone once taught you the thing you’re great at today. Do that for your team. “I saw you do this. I used to do the same thing, until someone pulled me aside and shared this. Does that resonate? Can you commit to doing it next time?”
It feels slow. Then one day you hear them coaching the newer people with your words, and you stop having to repeat yourself. Your real job was never the business you’re in. It’s building the people, because the people build the business.
Find it, fix it, repeat
Your business will always have something broken. That’s what being an entrepreneur means. There will always be a next bottleneck.
The win is getting faster at the loop: find the one constraint, surge on it, prove the fix with a real test, hand it to an owner, and move to the next one.
Pick yours this week. Then go all in on it.
-DM