Under $10M? Stop optimizing. Seriously.

#035

Read Time: 3 minutes

The short version: There are 2 kinds of tech debt. One is free. One is a migration project waiting for you. Below is the 3 question test that tells you which one you're about to build.

Last week I told you technical debt is a tax nobody budgets for.

Then I recorded a podcast where my guest told me to go build some on purpose.

Haris runs a RevOps firm in Amsterdam. He wrote a book on this. He works with B2B SaaS companies from mid-market to enterprise.

His advice for anyone under $10M in revenue:

"Don't make it perfect. Don't over automate. It's okay to break a few things. It's even okay to build a little bit of tech debt."

I sat with that for a week.

He's right. And so was last week's issue.

Because there are 2 kinds of debt, and almost nobody separates them...

You can undo most of it by Tuesday. Some of it, never.

Some decisions you reverse in an afternoon.

Some decisions require a migration, a budget, and a consultant with a statement of work.

That's the whole line. Reversible or not.

Haris's example was using the Product object for something the Quote object should be doing.

Tiny decision in year 1.

Get this backwards and you'll spend 6 weeks arguing about field names while quietly breaking the one thing you can't take back.

3 questions. Run them before you approve anything.

1. If we reverse this in 2 years, do we have to move data?
→ Yes = expensive. Stop and decide properly.

2. Will reporting history or an integration depend on it?
→ Yes = expensive. Stop and decide properly.

3. Can 1 person undo it in an afternoon?
→ Yes = cheap. Ship it messy. Today.

Most of what's on your roadmap is cheap. That's the good news.

Ship those fast and stop treating them like architecture.

The expensive list is short. In my experience it's under 10 decisions for the entire life of a company.

Those 10 are the only ones that deserve a real meeting.

This is also why your next hire is probably wrong

Most companies at this stage hire a Salesforce admin. $70K to $80K, 1 to 2 years of experience.

That person can execute the cheap list all day long. Genuinely useful.

But they've seen 2 orgs. Maybe 3.

They've never watched a data model decision go wrong 4 years later, because they haven't been doing this for 4 years.

So you bought execution when what you needed was 6 hours of judgment.

Haris's answer was to get someone senior on a fractional basis. 1 day a week. Sometimes less.

We do that for a living, so weigh my agreement accordingly.

But the math doesn't care who says it. A few hundred dollars of architect time on the 10 expensive decisions beats a full-time salary spent on the cheap ones.

Do this today

Open your roadmap. Run the 3 questions against every item on it.

Anything that flags expensive, pull it out of the build queue and put a decision meeting on it this month.

Everything else, stop debating and ship it.

You're allowed to have a messy CRM.

You're not allowed to have a messy data model.

Two different problems. Two very different price tags. Last week's issue was about the second one. This week is about not confusing them.

Full conversation with Haris is live now. We get into single source of truth, why AI can't rescue a broken CRM, and the 8 hours a week your reps are losing to admin.

Listen here