Five hard-won business lessons after scaling to $250 million

Use the ICE score (Impact × Confidence ÷ Effort) to rank ideas.

List every growth idea, score each on 1-10 for Impact, Confidence, and Ease, then sort by total ICE to pick the next experiment.

Why: Quantifies gut feelings and prevents low-yet-costly initiatives from diluting focus.

"Impact, confidence, ease. And so we actually do this. Like, I have this massive sheet that it's called growth, and I have it in all caps."

Every change incurs an immediate ~20 % performance tax.

Manual processes drop ~20 % in effectiveness when altered; automated systems behave differently. If the upside of a tweak is <20 %, the guaranteed loss outweighs the speculative gain.

Why: Human learning curves and temporary confusion create friction that compounds if changes are stacked.

"I get about a 20% decrease in effectiveness across any function that I'm going to change, especially if it's manual."

Doing nothing yields a natural 5 % annual improvement.

Teams organically get better through repetition, compounding like GDP growth without intervention.

Why: Experience and micro-optimizations accumulate over time when processes remain stable.

"If you change nothing, you get about a 5% guaranteed improvement."

Revenue retention is the universal gold standard, not virality.

Measure what % of Jan-1 revenue is still being paid by the same customers 12 months later; ignore logo count if dollars shrink.

Why: Retention compounds cash flow and de-risks growth; virality depends on product-market communication density that may not exist.

"The gold standard should absolutely be revenue retention. Bar none, full stop, and I can say that with absolute confidence no matter what business you're in."

Virality requires low sharing disincentive and high customer-to-prospect communication frequency.

B2B clients often compete with prospects, suppressing referrals; consumer products with social overlap spread naturally.

Why: Word-of-mouth is a network effect; sparse or adversarial networks kill the coefficient.

"some companies, like especially B2B, there's sometimes disincentives for a customer to bring in a competitor of theirs"

Target LTV:CAC on gross profit, not revenue.

Subtract COGS from lifetime revenue to get lifetime gross profit, then divide by acquisition cost; software margins allow 3:1, manual service models need 20-30:1.

Why: Low-margin or high-touch businesses incur real variable costs and training lumpiness that erode thin ratios when scaling.

"make sure you're doing it off gross profit, not off of revenue... if we're low leverage, we want to be at 20 plus"

Required LTV:CAC scales inversely with leverage across attraction, conversion, and delivery.

Paid ads + automated checkout + software delivery → 3:1 is fine; manual outbound + 1-to-1 sales + concierge service → demand 20-30:1.

Why: Manual processes create compounding inefficiency and recruiting lumpiness that magnify CAC faster than LTV.

"the three-to-one only matters at the companies that have 100% scale, 100% leverage across all three components"

The $1-3 M revenue band is a cash-flow swamp.

At ~$2 M revenue and 20 % profit, hiring a $250 k star consumes >50 % of annual profit, making the risk of a bad hire existential.

Why: Infrastructure demands rise faster than retained earnings, forcing founders to choose between overwork or a high-stakes personnel bet.

"You need the help, but you don't have the money to afford the help."