In 2017, Wistia’s founders took out a $17.3M loan to buy out their investors.

Not to grow or launch a new product… but to control their costs.

This seemed insane to everyone, especially since the company had $17M ARR, 82 employees & was steadily growing. Sure, they had a $4M annual loss but for SaaS, this is not abnormal.

The founders, however, felt differently.

This growing net negative was fostering a toxic cycle for them. With each new tranche of revenue, the burn grew larger and so did the pressure to cover the gap. It was forcing sloppy execution & short-term tactics.

They realized that their cash burn was sabotaging their core value to customers.

It was making them more vulnerable to competition and external events, in a market they otherwise knew they could dominate.

So they took back control of the company & started controlling their costs. Within a year, gross margins increased by 4% & they had a $10M cash swing. They went from $4M loss to $6M profit.

How is this possible?

How can we increase revenue this drastically by investing less?

It’s because ultimately, cash control is not just about balancing the books. It is about exploiting the 2nd compounding lever in our business.

People only see compounding as a growth function.

It is also very much a loss function as well.

Both are the same equation. The 2nd law of thermodynamics works here as well: the faster & farther we go, the more sh*t is going to break down.

Every new phase of growth brings in a new wave of complexity & chaos. We can obviously overcome this by patching up the big holes & go faster than the rate at which we're being dragged down.

But it devours our fuel, makes us bloated & all it takes is one Black Swan event for it to all come crashing down.

That is why the ability to create predictable operations is a true compounding lever.

Because a startup is not just a Customer Value Hypothesis being tested out in the open, it’s also a Value Stability Hypothesis (VSH) being worked out behind closed doors.

The VSH asks, What's it going to take right now in here to guarantee our value out there?

Once we build both levers simultaneously, we now have a system where for every pound of effort we put in, we get more bang for the buck each time. Only when the drag percentage decreases with each step-function of expansion do we unlock true compounding growth.

And to do all this, we don’t need to fix everything all at once.

We can start with our C-3PO & push for 1 of these OKRs:

Most founders obsess over MRR while treating cash as their frumpy parents - they check in once a quarter & rarely introduce them to their friends. Founders also feel more pressure to reinvest in hiring & other costs, and so they treat the growing monthly expenses as investments & move on.

The problem is that MRR is a lagging & smoothed signal.

Cash is more real & urgent.

And the gap between them for pre-$10M ARR startups is so big that they’re often one curveball away from defaulting payroll.

But the real truth of why founders don’t focus on monthly cashflows is coz they’re terrified.

The idea that they have to calculate the # of days to their extinction & see if it’s gotten shorter seems mortifying.

It is a lonely & depressing exercise. But it doesn’t have to be.

Cash balance is not a shameful reminder of a founder’s failure, it’s a lifeline & a privilege whose burden should be shared with the whole company.

That’s not easy at all but it is doable.

Buffer is the best example of this.

In 2016, Buffer had a cashflow crisis.

And co-founder Joel Gascoigne didn’t hide it. Everything from their revenue, expenses, runways… down to the last executive paycheck & contractor line item was shared with the team. He showed them the $1.3M bank balance with a negative trend & shared how they could potentially get to 3.5 months of cash.

The employees immediately got it.

They had some deep & honest discussions but by the end, everyone started thinking about how to do their best work with cost savings.

They canceled their annual retreat. Their team initiated an audit of each of their 70 vendors to see who was critical. Everyone kept scouring the financials to find inefficiencies in their work and brought ideas to each company meeting.

In 2 months, expenses went down 11% & they hit their highest MRR growth.

Within a year, they were profitable & held it ever since.

Build one simple cash forecast: the balance, the trend & the # of months of runway we have. Napkin math will suffice for this.

Then share it with the whole team.

The key is to show them what's in the bank today, what comes in, what goes out, and how much time it buys us. These first numbers will feel scary & shameful.

Don’t worry, this is the hardest part.

It’s now everyone’s problem & most importantly, we have a baseline.

There’s 2 ways to extend runway: Money leaves slower or arrives faster.

Cost side: Each function names its 2-3 single biggest expenses & brainstorms ways to cut them or delay them.

For example, they can renegotiate a contract with a top credit card provider for better rates, run a vendor/tool audit to find who/what’s critical or even ideate creative ways to delay hiring someone.

Cash timing side: Teams can also think up creative ways to increase the speed at which money hits the bank.

For example, charge on signature not delivery, dedicate a teammate to chase AR every 2 weeks or even bill annually upfront & include services.

After picking everyone’s favorite ideas in both categories, have the team commit to a 90 day target with a clear list of actionable items to hit. Each category should have an owner.

Every Friday, each owner reports their number from the past week (10 min to prep) and one thing they learned. No slides, no theatrics.

And ask, did anyone's cost cut create a downstream problem?

For example, if we cut off a vendor only to find that the Support team actually needed that tool or an annual-billing push scared off a customer cohort, that’s not good.

A saving only counts if it doesn't fail somewhere else.

The key is to find the balance between efficiency and following through on our core brand promise.

Run the same napkin calculation. Same inputs. Watch the months-of-cash number climb.

Trust me, it will.

…

Getting to a 3 month runway is a very hard goal.

And it will fluctuate in different seasons, but if done right, it can enforce discipline, seed bold revenue ideas and energize the team.

Because once they see the number go up, it becomes addicting.

A fun money-pinching OKR can unlock an entire culture of transparency, efficiency and ownership that grows alongside our bank account.

What does a Starbucks coffee, a Louis Vuitton bag and your latest AI tool all have in common?

They all take home 50 cents on every $1 they make.

AI compute costs are eating alive the best SaaS companies out there, especially the smaller ones. The new SaaS gross margin floor is no longer 70-85%. It's 50-65% & not getting any better.

Even with token prices falling 60% in 2025, startup P&Ls didn’t see any of the savings.

Because with every new model comes newer, sophisticated & more expensive compute features. Capabilities like retrieval and self-critique are using tokens per compute faster than per-token prices are falling.

Mix that in with a small but rabidly hungry AI user group within our customer base & we will be witnessing true compounding loss in action as we grow.

Unless you nip it in the bud like the Project Discovery team did.

When they first launched Neo, their autonomous security testing platform, it was a real hit.

Enterprise IT security teams loved their new multi-agent workflow.

But just as the team was getting ready to scale, the developers got the bill and their jaws dropped: “a single complex task with Opus 4.5 could consume 60 million tokens.”

They weren’t using the wrong models. It was just that each time their agent pinged the LLM, it was re-shipping the same bloated 20k token system prompt each time, up to 20 - 40 times per task.

So they started caching & routing.

They restructured the prompt so the stable parts could be cached, and the dynamic parts (working memory, per-user variables) got pulled out and relocated.

As a result, their LLM costs dropped 70%.

Have the team yell out the most common & irritating customer requests/issues that they’re seeing, including when they demo.

These are the commands that collectively eat up 60-80% of their compute.

Then group those requests by what they’re trying to actually do, not just what they typed.

This smart mapping is what makes caching and routing economically meaningful.

Over the next 2 weeks, have engineering rebuild prompts so the stable parts (instructions, context, examples) come first and the variable parts (user input) come last.

This single change can unlock 90%+ savings on cached tokens — and protects your margin from LLM price changes.

Simultaneously, have GTM create templates & FAQs to handle the other parts.

Call out the current AI compute spend in your weekly standups & track its progress as a team.

Have the team refine the architecture as they aim for a 50% cost reduction in 90 days.

…

If we do this OKR right, we are not only building an AI foundation for the company that can scale with its growing complexity & skyrocketing model prices,

But we’re also seeding a culture that learns to actively contribute to the gross margin.

Because AI scaling isn’t just an engineering problem,

It’s about making COGS every team’s responsibility.

Every founder is told to lower their CAC. That’s a bit bullsh*t.

CAC by itself doesn’t mean much. It’s a vanity metric dished out by self-important VCs to make themselves feel good and lowering it means Marketing has to cut their budget.

The real metric is CAC Payback.

For example, a 24 month payback means you have the privilege of paying your customer for 2 years, before they start returning the favor. Multiply that times 100 new customers each month & it’s easy to see how a growing company can risk defaulting payroll.

When growth is rented out by our bank balance, that’s compounding loss function in plain sight.

Series A founders usually don’t track CAC Payback coz it’s messy, takes a lot of time to track & is a lagging indicator that doesn’t seem worth all the effort.

But there are 4 easy principles & leading indicators we can apply each week to drastically cut payback time.

Each of Adam Robinson’s new startups would grow faster than the one before, not because of more funding. They were all bootstrapped.

But because with each one, Adam got better at 1 thing:

Finding the uncontested acquisition channel & going all-in on it.

For RB2B, that channel was a LinkedIn audience he'd built before the product even launched — by launch day, 3,000 people were already on the waitlist.

And the reason this works is because long before those customers bought RB2B, they trusted it.

This is the same principle that let Klaviyo win e-commerce email. They went all-in on the Shopify ecosystem & nothing else.

Pick 1 channel & own it.

Pull our last 20 closed-won deals. Split by who closed them.

If the founder closes in 22 days with 30% success rate and the AEs close in 45 & 15%, the goal of the playbook is to close this gap. And it doesn’t have to be a 50 page document either.

Start by recording & stalking the founder’s greatest closes. Do the same with the AE’s most disappointing losses & try to find the 3 biggest bottlenecks where they drop the ball.

Have the founder create a Sales for Dummies process there to overcome those valleys of death.

The key here is to operationalize the founder’s secret sauce in a way that any AE can replicate it.

In every software, there’s the one sticky moment which converts the user from a passive tester to a very motivated user.

For Figma, it was when a new user sketched a design & shared it with a friend. The moment they saw their teammate's cursor go live & explore their design in real-time, they were hooked.

Every Onboarding team should only slap hi-fives when users pass this magical UX moment.

When ConvertKit had 14 people serving 2,300 customers, their founder Nathan Barry set an impossible goal: 2x the customer base without adding headcount.

This forced his team to find unsexy ways to boost productivity while maintaining high quality.

They did things like fix the bugs & UX flows that generated the most tickets, craft pre-made scripts to make the team reply faster and run live trainings to help any new CS rep tackle the toughest tickets.

All these small little efficiency wins pressurized by this hard goal worked.

The average ticket response time fell from 24 hours to 8 hours, while supporting more than twice the customers.

Using 6 months of Sales, Marketing & Support dollars spent against a cohort of customers, we can calculate a CAC Payback that’s accurate enough.

As long as it’s within the ballpark & easily replicable each quarter, we’re good to go.

Keep in mind, this CAC Payback will be ugly but at least we’ll have a baseline.

Next, each GTM team picks 1 KPI to track weekly

Each team should set their own 90 day target & own it fully.

If they want to be ambitious, they can aim to cut their KPI in half. They’ll most likely miss it, but at least they’ll be making moves.

BTW, it might take a few weeks to clean these numbers up & get into a rhythm, but once you do, that’s half the battle. Also in startups, 2 of these legs often have the same owner.

The point is to have 4 distinct numbers to scrutinize each week.

Every Friday. Each owner reports their number from the past week (10 min to prep) & 1 thing they learned. That's it. No slides or theatrics.

And then ask: Did anyone’s win create a downstream problem?

Most OKRs quietly die because one team’s win screws over another’s. The only solve here is to tie all their wins together in a relay OKR.

A metric can only count if it doesn’t fail downstream.

For example, let’s say Sales drops close-time from 45 to 32 days but then CS flags that 9 of those were small contractors who won’t expand. Now, not only does Sales have to exclude these bad fits from their KPI, they also have to tighten their process to ensure this doesn’t happen again.

This may also force them to have a deeper conversation with Marketing on their tactics as well & that’s a good thing.

This baton-relay check-in helps us overcome the temptation to game the KPI & re-centers us on our true goal.

Run the same back-of-napkin baseline calculation from Step 1. Same inputs.

The number will drop — not a lot but if we can lower CAC payback by 2-4 months, that’s a really good sign here.

More importantly, it’ll give the teams hope and now that they’ve seen the full loop, they’ll get a renewed sense of energy & ideas on how to push it further down.

…

The beauty of running a great CAC Payback goal is that it makes unit economics not just a depressing Marketing problem,

But an exciting full-on company challenge that gets the growth to fund more of itself each time.

Guillaume Moubeche was the king of the world at the end of 2021.

His startup Lemlist was the breakout star that year. In just a few years with $0 funding, they exploded onto the most competitive market in tech: cold outbound sales & grabbed $10M ARR.

He & his co-founders were ready to seriously grow so they took some funding and doubled the company to 60 employees.

But 6 months later, chaos struck… internally.

First, Guillaume lost both his CTO co-founders. Soon, his Head of Growth & Head of Sales also left.

Finally, Guillaume himself had to let go of a third of his employees. There was a lot of tension in the company and employees were feeling scared.

As a result, growth itself slowed.

Even with a unicorn business, Guillaume ran into the 2 lies & 2 tectonic shifts of early stage hiring.

The 1st growth stall is a founder with control issues.

Even after bringing in help, they keep finding fires they need to personally put out & details that require their vigilance.

So the founder is now back in the muck, except with more fingers in more pies with less time & less energy. Yet they persist, believing that if they can apply this pressure long enough, they’ll soon get out alive.

Most often, they end up becoming the bottleneck & get burnt out.

In the worst-case, they become one of the walking dead CEOs, with just enough brain cells to answer yes/no questions from their team.

Everything else is white noise inside of them.

Investors often pressure founders to hire against forecasts. They make us believe that higher headcount = higher growth.

But the problem is that human relationships take time to build & there’s a communication tax that multiplies with new recruits.

With every new employee added in a short period of time, chaos slowly increases in the company. And if it passes a certain threshold, the startup tips into a state of dysfunction that it cannot scale past. That is very hard to undo.

The faster we try to hire people, the longer it takes to build our business.

Up until around 20-25 people, startups feel very personal and the team feels like a tight-knit family. But past two dozen employees, it’s now a village.

Conflicts don’t get resolved with a simple chat, they require the CEO to referee them & clear rules/processes to prevent them in the future.

Hiring starts happening from outside our personal networks since those are maxed out by this point. We cross from known people into trusting strangers.

To get to this stage successfully, you need to have hired the right people to protect a founder’s weaknesses. The new specialists execute what the founder struggles the most with.

The next terraforming happens around 40 to 50 employees. The Village becomes a Town and the founder becomes a mayor with official department heads.

It’s not about group size or trust at this stage. It’s about clear teams, early HR structures & SOPs.

The founder is no longer directly running most functions and they pivot to being an orchestra conductor.

Getting here successfully needs fantastic senior leaders who are better than the founder at their core strengths. These should be 0-to-1 systems builders who’ve done this before.

The hardest part here is that these positions are often filled with Day 1 leaders who’ve hit their ceilings.

TLDR:

In a great Village, a founder has hired to make all their pains go away.

In a great Town, a founder has hired to make themselves optional in all roles.

If we don’t account for these 2 lies & 2 ground shifts, the founder will easily trip up & be struggling through a swamp of their own success, all the time wondering why.

…

Here’s 3 steps to escape this hiring quagmire:

Clarity leads headcount.

To generate consistent value while growing, we must only hire when the unit of work is stable, teachable & it hurts not to.

Instead of trying to force your hiring plans to match a VC’s growth org chart, identify the top 3 places where the work is breaking or overflowing.

Before hiring, confirm the work recurs predictably. This shouldn’t be a one-off problem & has to have been around for at least 90 days.

Next, try to see if AI or a better tool/process can fix it. Most often, it can.

For example, Zapier’s early team was drowning in customer support. Just before posting a role, someone found that 40% of their inbox was just auto-response emails. They got their engineers to build a filter to weed those out & it worked.

Don’t staff a problem you can skip or delete.

This is the most important step that founders skip.

Create a Job one-pager that covers the role’s 3-5 recurring tasks, the #1 KPI that measures progress, alongside a playbook that has repeatable main steps & the founder’s judgement baked in. Test it if needed with a contractor.

Finally, do a gut check on the cost of drag vs. the pain of hiring.

Is the ongoing lost time, process inconsistency, or slowed growth more painful than the time & effort it takes to find & train someone new for the next 6 months?

If yes, then pull the trigger.

Hiring is an extremely emotional and intuitive practice that’s more of an art than a science, often leaving us conflicted & unsure within a huge group of qualified applicants.

Here’s 3 questions to immediately clarify our thoughts on each candidate.

Every founder gets excited when they encounter candidates with stellar logos & padded resumes. This is especially true of executives from other successful startups.

But often these people are good at helping scale companies with more established processes & stink at building from scratch.

That’s why the most important trait for the first 100 employees is to find hungry & optimistic builders, who’ve always been tinkering & taking risks in some capacity.

This characteristic far outweighs any specific domain experience they may be lacking.

Once we’re about the Town size, we’ll also add a systems-level thinking to this trait.

The best way to answer this question is to insert a meaty real-life case-study or a sample project with road bumps & curveballs to see how they think & move through it.

To get to $10M and beyond, we want really smart people who are already destined for great things given who they are. Pick them selfishly, so we can benefit from their genius & learn a lot from being tied to their trajectory.

We also want to work with people who’d be a lot of fun burning the midnight oil with.

Imagine it’s a really rough ride for the next 10 years & see if this is who we want to be stuck to for that.

Pay 20-50% more than the market to help you find the 5-10X profile.

Increasing the pay not only helps us attract better candidates but more importantly, it forces us to find the fastest path to clarity in a high cognitive tax decision.

When we raise the stakes on picking someone to the point that it hurts, it simplifies our internal debate & helps us have a clear binary choice to the question:

Is this person worth this much money that we don’t have?

This might be the most controversial point but most entrepreneurs flounder with a bad hire for a long time. They sit on it for almost a year, wasting precious time and money, before making a call on someone they already decided on in the first 8 weeks.

Also if we’ve done the first 2 steps really well, we should have set up the employee to succeed.

But just in case they can’t step up, we also need to protect our downside. Set the expectation up-front in the job posting, that all new hires, exec or employee, have to earn their keep by hitting their first 90 day KPI & outcome goals.

The reality of executing this won’t be as clean & most often, we’ll keep them around longer, giving them more time. But by drawing a line in the sand this early on, we attract candidates with high grit & deep self-confidence in their abilities.

And because these are employees who are getting paid generously, this also forces us to assess them sharper & make gut-wrenching calls on their role-fits faster.

The key to this OKR is that no hire gets approved without a named 90-day productivity ratio against an ARR per employee benchmark.

All of this seems like a lot of ground work to do in a busy startup but like Claire Hughes Johnson who scaled Stripe from 160 employees to 80k says,

“Real scale is repeatable excellence, not headcount.”