Jonathan Ingles: Elena, quick question before we start — does it bother you when a number just refuses to make sense?
Elena Marsh: Oh, constantly. Which number?
Jonathan Ingles: A cybersecurity firm. $1.2 million to $4 million revenue in three years. Net margin starts at 4% — already thin — and by the end it's nearly invisible. She tripled the top line and the bottom line basically vanished.
Elena Marsh: And the gut reaction is — that's a management failure, right? Someone made bad decisions.
Jonathan Ingles: That's what I assumed. But the research pushes back on that hard.
Elena Marsh: Because what we're actually looking at is structural margin compression — the idea that this isn't a downturn, it's not a mistake, it's a feature of how businesses mature. And I want to see if we can make that click without any jargon at all.
Jonathan Ingles: Give me the one-sentence version.
Elena Marsh: Growth spends your savings. Early on, you're finding easy wins — cutting waste, automating the obvious stuff. But those wins are finite. Once they're gone, every dollar of new revenue costs almost as much to produce as it earns. CFO Pro Analytics puts it squarely: the operational challenges at $5M look nothing like the ones at $100M, and the gap is almost always overhead that revenue can't outrun.
Jonathan Ingles: And that overhead gap is what I want to pull apart, because — okay, the first savings make sense. You automate the manual invoice process, you cut the supplier you were overpaying, you kill the product line that was dragging cost. That's efficiency depletion, right? One-time gifts. But here's what I don't fully track: why can't you just keep finding the next gift?
Elena Marsh: Because the cheap interventions are gone. What's left — those require expensive, uncertain transformation. Think about it this way: a founder at, say, $3M revenue finds out her team is manually reconciling payables every Friday. She buys accounting software. Margin improves, almost immediately. That intervention cost her twelve thousand dollars. The next inefficiency she finds? It's in how the whole service delivery org is structured. Fixing that means a reorg, probably six months of disruption, and no guarantee.
Jonathan Ingles: So the ROI on fix number one is obvious. Fix number fifty is — I mean, you're almost guessing.
Elena Marsh: Exactly that. And Morgan Stanley and Columbia Business School actually mapped this across the corporate life cycle. ROIC — return on invested capital — rises through growth and into maturity, then it falls. Not because the company is doing something wrong. Because the structural forces compressing margins start to offset whatever organic efficiency gains you can still find. That's the ceiling. That's what the plateau is.
Jonathan Ingles: Hold on. The plateau is empirical? Like, companies don't just keep compressing forever — they stabilize in a band?
Elena Marsh: Observed across mature firms, yes. They settle into margin bands they cannot escape through organic scaling alone. Revenue keeps climbing, margins stop following. Amazon is the — well, it's the most uncomfortable version of this. Retail margins that thin, persistent, despite a revenue scale that should by any intuition produce leverage. But the continuous infrastructure investment — warehouses, transportation, last-mile — it just absorbs the gains.
Jonathan Ingles: Amazon spending its way to flat margins at that scale is actually a damning data point — because if anyone could brute-force past the ceiling, it's them.
Elena Marsh: Which is the uncomfortable version of the structural argument. The ceiling isn't a failure of nerve. It's — well, it's almost architectural. Once the early wins are spent, you're not finding savings anymore. You're buying transformation. And that math is different entirely.
Jonathan Ingles: But here's where it stops feeling abstract — because you said 'architectural,' and I want to push on that. Because architecture implies something that was designed. What if nobody designed this?
Elena Marsh: That's — that's actually the more unsettling version. Picture a compliance officer. Hired on a Tuesday. One specific regulation needs a sign-off, so you hire someone. Six months later, a second rule comes in. She can't handle both alone. You add a junior. Eighteen months after that Tuesday hire, you have three people, a budget line, and a reporting structure. No single leader decided to build a department. But no single leader stopped it either.
Jonathan Ingles: And the revenue that triggered the first hire — it didn't triple the compliance workload. It maybe doubled it.
Elena Marsh: Right — but the *structure* of compliance scales in steps, not proportionally. That's the overhead ratio trap specifically. HR, IT, finance, management layers — they don't grow with revenue, they grow with organizational complexity. And complexity grows faster than revenue almost by definition once you hit a certain size. The Polish Economic Institute tracked this at national scale — net margins across enterprises fell from 4.3% to 3.4%. Worst in a decade. That's not one bad CEO. That's the pattern running through the whole economy.
Jonathan Ingles: Hold on — 4.3% to 3.4% nationally? That's not rounding error.
Elena Marsh: No. And that's what makes it structural rather than anecdotal. Though — and I want to sit with this — there's a debate I don't think we should paper over. Is the overhead growth genuine complexity? Or is it partly managerial self-interest? Headcount signals power in most organizations. A CFO with twelve direct reports has more institutional weight than one with four.
Jonathan Ingles: The incentive is to build the department whether you need it or not.
Elena Marsh: Yes. And here's what's almost frustrating — both explanations, genuine complexity *and* managerial self-interest, produce exactly the same outcome on the margin line. You can't distinguish them from the P&L. Which means the question of whether this is fixable or inherent — well, that depends entirely on which mechanism is actually driving it. And there's supposedly an escape route. We'll get to whether EY's top-ten-percent blueprint is real discipline or just a story companies tell about their starting position.
Jonathan Ingles: That's the part I don't buy yet.
Elena Marsh: Well, the blueprint is real — EY's data on top-10% margin-leading firms is pretty specific about what separates them. It's not operational excellence in the hustle sense. It's three things: defensible competitive advantages, industrialized operating models, disciplined growth. That's the architecture. The thing that nags at me though — you can only build a moat if you have something to build one with.
Jonathan Ingles: Proprietary technology. A trusted brand. Regulatory barriers. You're talking about structural moats.
Elena Marsh: Exactly those. And that's what enables premium pricing even under competitive pressure — not leaning harder on operations. But most firms hitting $50M don't have a proprietary moat. They have a good sales team and a serviceable product. So when EY says 'deliberate structural choices,' I think we have to ask — choices available to whom?
Jonathan Ingles: That's — frankly, that's the part I think gets laundered as wisdom. 'Be disciplined.' Right. But the operational leverage mechanism, the actual math of it — fixed cost base, revenue grows, margin expands — that only works if the cost base *is* fixed. And most firms, the moment they scale, are adding headcount proportionally. So the escape valve exists, but they're actively blocking it.
Elena Marsh: Picture a founder. Revenue just crossed $10M. She could freeze hiring in the support layer and let existing infrastructure absorb the next $3M. Margins expand. Or — she hires the VP of Customer Success her sales lead has been requesting for five months, because the sales lead is her best employee and she's afraid to lose him. One decision. The operational leverage is gone.
Jonathan Ingles: And that's not incompetence. That's a retention decision that kills the margin math.
Elena Marsh: Which brings me to something that genuinely surprised me. The sector that seemed immune — SaaS, high gross margins, low COGS, infinite scaling — that model is actually breaking right now. As of 2026, B2B SaaS companies integrating AI features are seeing per-query compute costs push gross margins into the 60–70% range. Down from the 80%-plus profile that made software the investor darling for two decades.
Jonathan Ingles: Wait — 60 to 70%? The software escape hatch is closing?
Elena Marsh: That's the AI COGS problem. And Alten Capital has a counter-thesis — their 'agentic margin' argument is that AI agents can decouple revenue from labor cost in services firms, which is the mirror image. But the sources don't actually resolve it. Whether AI is primarily a new cost burden on software or a decoupling mechanism for services — honestly, both things might be true simultaneously, in different sectors, right now.
Jonathan Ingles: Both true simultaneously. That's actually where I land on all of it. The three mechanisms — efficiency depletion, overhead ratio, competitive pressure at the top — they don't take turns. They run together. And knowing that doesn't fix it. But it changes what you're deciding.
Elena Marsh: That cybersecurity founder — she didn't fail. She did exactly what growth requires. And the margin followed the structural law anyway. Which is — well, it's a quieter indictment than incompetence. It means you can execute perfectly and still hit the ceiling. The only question is whether you see it coming.
Jonathan Ingles: The companies that escape decided earlier what cost base they were willing to live with. That's it. Not smarter. Just earlier.
Elena Marsh: Three mechanisms. If you know them, you at least see the ceiling before you hit it. That's the thing you carry out.
