30 years operating: what stays true.

I started in financial services in the early 90s. Spent years at moneycorp on the FX side. Operated and chaired across BPO and telecoms — Crewit Resourcing, Itelecoms, Data Fusion International. Now CEO of a climate-tech company that runs across 195 countries. The categories look unrelated. The operating problems are mostly the same.

This is a short piece on the patterns that keep showing up.

1. The unit economics tell you who you are

Every business I've operated at scale has the same first question: what's the gross margin per transaction, and how does it move with volume? In FX, it was the spread. In BPO, it was utilisation against fully-loaded headcount. In telecoms, it was minute-cost arbitrage. In carbon-baked travel, it's the commission spread net of payment fees and the offset retirement.

If you can't write the unit economics down on one page, you don't understand your business. If you can write them down but they don't get better with scale, you have a marketing problem dressed up as a business.

The IMPT unit economics work because hotel commission spread compresses slowly, carbon retirement cost compresses faster, and the payment fees compress fastest of all. The scissor opens. That's the bet.

2. Distribution is the moat. Always.

I've watched dozens of products with superior tech lose to inferior products with better distribution. This is the most consistent pattern in commerce. The 2026 version: AI-native consumer apps with no distribution channel are losing to LLM features inside existing distribution channels. The lesson hasn't changed in three decades.

For IMPT, distribution means three things: the hotel-network coverage (8M properties), the carbon-project supply (six categories, multiple jurisdictions), and the partner-OTA relationships that carry our inventory into other surfaces. The product is good. The distribution is what makes the product matter.

3. Compliance is a feature, not a tax

Coming out of regulated industries — FX, financial services — I've never bought the founder mantra of "move fast and break things". The things that break in regulated industries are people's pension contributions. So you don't break them.

This shows up in carbon. The integrity story I've written about elsewhere is exactly this point: if you treat verification, retirement-tracking, and audit as bureaucratic friction to design around, you build a product that can't be sold to institutional buyers. If you treat them as features that buyers actively pay for, you build a product that wins in the long market. We've gone the second way.

4. The team you hire in the first 24 months is the team that decides the next decade

I've made every variation of this hiring mistake. Hired too senior. Hired too junior. Hired for skills the business needed in 2018 but not in 2024. Hired great operators with bad cultural fit. Hired great cultural fit with weak operating output.

The version that works: hire people who can operate two levels above where you currently sit, but who are willing to do the work two levels below their pay grade until the company is ready for them. Hire for ownership, not for talent. The team that built IMPT into what it is in 2026 is composed almost entirely of people who joined when there were fewer than fifteen of us.

5. Capital cycles end faster than founders think

I've operated through 2000-01, 2008-09, 2020-22 and the 2024 reset. The pattern is identical: founders raise into the peak, build cost structures into the peak, and then have to cut hard when the market turns. The ones who survive are the ones who modelled the next 36 months not on the last 12.

IMPT raised $30M in March 2025. We've operated as if the next round is two cycles away, not two quarters. The discipline of running like capital is finite is where most of the operating leverage lives.

6. Software eats services until the services become software

This is the BPO observation. The work that BPO firms used to handle — back-office processing, claims handling, contact centres — has been steadily compressed by software automation since I was first in the industry. The firms that survived built the software themselves. The firms that didn't, fought the software vendors and lost.

I see the same pattern in carbon services today. The traditional offset broker is being compressed by API-native carbon platforms. The traditional travel agent was compressed by OTA software two decades ago. The pattern repeats. The operator who reads the pattern correctly invests in the software-replacement before the customer asks for it.

7. The CEO's job is to remove the next bottleneck

Not to be the smartest person in the room. Not to be the visionary. Not to do the talking on stage. The job is to identify the single biggest constraint on the business this quarter, fix it, and move on to the next one.

For most of 2024, the bottleneck at IMPT was hotel-supply integration. Then it became conversion on the booking surface. Then it became carbon-supply diversification. Then it became fundraising to scale the team. Each of those was a months-of-attention problem at the time. Each looks small in retrospect.

If you can't name your current bottleneck, you don't have one — you have several, and you're not paying attention to any of them.

8. Cross-industry pattern-matching is the highest-leverage skill

The thing that keeps me operating across industries — financial services, BPO, telecoms, climate — is that the operating problems repeat. Distribution dynamics, churn mechanics, unit economics, regulatory structure, sales-cycle length, pricing power. The names change. The shapes don't.

The operator who's spent twenty years in a single category often loses to the operator who's spent twenty years across five. Pattern-matching wins. Specialised intuition is a fragile asset.

The honest part

None of this is a guarantee. I've made calls that worked and calls that didn't. The bet IMPT is making — that integrity-grade carbon retirement attached to consumer commerce is a category big enough to matter — could still be wrong. The market could decide consumers don't care about climate at this scale. The regulatory shape could turn against on-chain settlement.

What I have, after thirty years, is a set of pattern-matching priors that say: this looks like categories that worked, when they worked. Big consumer surface. Strong unit economics that improve with scale. Compliance as a feature. Software replacing services. Distribution as the moat. Verifiable receipts as the differentiator.

That's enough for me to commit the next decade to it. Whether it works is a different question — and one we'll find out the way you always find out: by operating, hard, every day, until the answer is obvious one way or another.

← Back to analysis