E31: The Cost of Building Fast

On this episode of The SaaS Operators, Rishabh, Jeremiah, and I talked about what it actually looks like to operate a SaaS company in 2025, when markets are tightening, AI is reshaping budgets, and founders are being forced to make hard calls.

‍

We dug into Stamped’s journey through tech debt, layoffs, and restructuring, and used it as a lens to explore how software companies adapt when their old playbooks stop working.

‍

There are seasons in a company when you feel the push and the pull at the same time. Opportunity on one side. Pressure on the other. That is where a lot of us are operating right now.

Canada, incentives, and the push pull

We joked about it on the podcast, but there is a serious undercurrent here. Canada keeps nudging private enterprise in the wrong direction. Exit taxes. Policy drift. If that continues, owners will leave out of necessity, not just for opportunity. The only thing stopping many is the immediate pain of leaving.

‍

On the flip side, if you are based in Canada you end up leaning overseas for labor anyway. Fewer visas to worry about. Fewer hidden costs. Every system has tradeoffs. Pick the ones you can live with.

Capital is loud again

Andreessen raising a ten billion dollar set of funds is a reminder that there is plenty of money to deploy. The question is where it lands. Right now the gravity is obvious. New software spend is AI software spend. If you sell anything that is not clearly AI or clearly saving money, you are competing for a shrinking slice of wallet. That shows up as consolidation everywhere else.

A concrete case: Stamped

Jeremiah stepped into Stamped as CEO this year. Quick backdrop.

‍

Stamped was a rocket starting around 2015. Reviews first. Loyalty later. Acquired by Tiny in 2021. The product had been built fast. Customers asked and got what they wanted. It worked, until scale made the tech debt bill come due.

‍

Legacy stack in C Sharp and dot net. Microsoft servers and licenses. Patches on patches. Then a big lift to migrate toward Postgres and Node with services instead of a monolith. During the transition you carry two burdens at once. You still pay to keep the old thing alive while you fund the new thing. Infra costs go up. Headcount goes up. Meanwhile revenue pressure rises because the market moved.

‍

Competition tightened. Judge.me nailed the reviews use case and anchored price expectations at the low end. Revo did the same in loyalty on Shopify. When a competitor sets the mental model that reviews are worth fifteen dollars a month, you can still charge more if you deliver differentiated value, like syndication into Sephora, but you must be meaningfully better. Otherwise you bleed.

‍

Layoffs happen when the math turns. Expenses up. Revenues down. No outside funding. You cut. That is not theory. That is a P&L.

The strategic fork

People romanticize rebuilds. Why not take five great engineers and rewrite the product from scratch, then migrate customers. Sometimes you do. Often you cannot do it fast enough without breaking trust or bankrupting the company during the overlap. The cost structure kills you before the new ship floats.

‍

So you pick. Keep servicing everyone and move slow. Or tell a non trivial number of customers no and move fast.

‍

There are examples of companies that cut deep to refocus. There are also Netflix style separations. Let the old thing run on a minimal team while the future gets the attention. Both paths exist. What you cannot do is pretend nothing changed.

Startups have the edge right now

If you have zero customers you can ship in minutes. No backward compatibility. No thousand edge cases. That is why small teams are moving scary fast with AI. Incumbents can still win, but they must stomach angry customers and sunsets or carry the ballast of every legacy promise. Neither is fun.

‍

Also true: most cool features are not a business. The fundamentals still matter. Gross margin. Payback. Retention. Cost to serve. The companies that survive the next decade will pair AI velocity with operator discipline.

What buyers actually want

AI is a label. Buyers want the thing the label promises. Fewer inputs. More outputs. Profit. If you can tie your product to revenue or cost reduction in a way that is obvious on a statement, you can grow. If you cannot, the AI budget will eat your lunch.

My take on strategy right now

1. Stop doing work that is only fixing yesterday. If a feature does not map to how brands win in the next five years, consider divesting it.
2. Make the current core cheaper to run. Infrastructure, data flows, support load. Every dollar you save becomes fuel for the bet.
3. Place a clear bet that aligns with how the market is actually buying. Not a deck. Not vibes. A product that removes steps, saves time, or opens a revenue route customers feel immediately.
4. Be willing to say no. Some customers will not come along. That is part of the cost of moving.

A note on the multi product fantasy

We brought up Peter Levels. Incredible operator. Solo portfolio, many apps, media friendly lifestyle. Also rare. There are more unicorns than there are people who can run six profitable products alone. Do not mistake outliers for a plan. For most of us, focus beats breadth.

The unglamorous part

At some point every business hits a wall. You will be disliked for choices you did not really have. You still own them. Do it clean. Communicate like an adult. Optimize for being respected in five years, not loved this week.

Where this leaves me

I want to build for the long term. That means hard calls. It means biasing toward things that make customers money with fewer inputs. It means keeping cost to serve low enough that price pressure does not crush us. And it means shipping where the market is going, not where it used to be.

‍

I would not trade this game for anything. But let’s be honest about the rules.