Ask any B2B company what it costs to acquire a customer and you will get a number in about four seconds. Try digging into how that number was built and the room gets quieter.

People quote the smallest number available, and not because anyone is hiding anything. Paid CAC is the only version a single function can build on its own. It lives in marketing’s systems and takes a minute to pull. Every larger version needs finance, sales comp, and RevOps to assemble it, and no single function can do that alone.

The gap between that number and the real one is not an accounting curiosity. It is spend that produced nothing. The money evaporates, and no report you run will ever call it a loss.

CAC gets counted three ways

The first three are arithmetic. The fourth is the one worth your attention, because it is the only one still in front of you.

You did not waste the money. You created value and never realized it.

The value leaks at the seams, not inside the functions

The instinct is to look for the loss inside marketing, or inside sales, or inside customer success. I have rarely found it there. Each function can run well on its own metrics and the business can still bleed, because the leak is in the pass, not the play.

There are more than four. These four are the handoffs every B2B revenue motion has by construction, and nobody publishes a measurement of what happens in any of them.

Engagement to next action. Every touch signals where a buyer is. A booth visit, a partner intro, two hundred website sessions, a stalled trial. Nobody reconciles them into a live read, so the company acts on a stale guess. The booth visitor gets a cold call. The high-intent account sits unknown. You paid to create the engagement, then let it go cold.

SDR to AE, and to partners. A qualified lead or booked meeting sits and goes stale. Context is lost in the pass. There is no clear next best action, so whoever owns the next touch acts late or not at all. Speed to act is the leak.

Won to onboarding. The deal closes and the context does not transfer, so onboarding starts cold and early churn risk creeps in. Note that this ledger is net revenue retention, not CAC, which is exactly why it goes unwatched.

Adoption to expansion. Expansion is left to chance. Telemetry showing who is adopting, stalling, or hitting limits goes unread, so upsell moments pass. Nobody maps the other buyers inside the account who would benefit. Meanwhile at-risk accounts still receive upsell blasts.

Fix one function in isolation and you feed cleaner inputs into a system that still leaks at the seams.

What the leak is worth

Take a representative $120M ARR B2B SaaS business at a 20 percent EBITDA margin. That describes a company still growing, not a mature take-private where margins run considerably higher. Fully-loaded sales and marketing runs around 25 percent of ARR. Benchmarkit’s 2025 survey of 583 private SaaS companies puts the median at 33 percent for PE-backed companies, so 25 percent is the conservative end. A conservative estimate puts 12 to 15 percent of that spend in the fourth column, creating value that never gets captured.

Conservative is the right word. Salesforce’s 2026 State of Sales finds the average seller spends just 40 percent of their time selling, so most of the capacity you are paying for is already going somewhere other than the customer. And no new money is arriving to fix it. Gartner finds 56 percent of CMOs lack the budget to deliver their own 2026 strategy. Duke finds firms cutting investment outnumber those increasing it by nearly four to one. The only money available is the value you are already creating and failing to capture.

Representative B2B SaaS. Indicative figures.

Realize it and it drops toward EBITDA, because sales and marketing sit above the line.

If you hold the company. Every dollar of EBITDA is worth your multiple in enterprise value, so the enterprise value lift equals the EBITDA lift. You never have to win an argument about the multiple to get paid for this. And because these same four seams recur across the portfolio, it is not a one-off fix. It is a repeatable value-creation play you run company by company across the hold. And if you hold the view that the margin above is low, run it at your own: the recovered dollars do not change, only the percentage they represent. At a 40 percent margin the same recovery moves enterprise value roughly 7 to 9 percent instead of 15 to 19. The money is identical either way.

If you run the company. Realized, it shows up as margin you did not have to grow revenue to earn, and it compounds. Each fixed handoff feeds the next, so the gains build quarter over quarter instead of fading. Faster pipeline, higher net revenue retention, and a better quality of earnings underneath the growth rather than strain on top of it. If the company is public, the multiple is not yours to argue. Margin earned from the operating model rather than from spend is the kind the market re-rates.

Capacity is the asset. Compound it, do not cut it.

The fastest EBITDA is a headcount cut. The durable EBITDA is capacity redeployed.

This is where AI earns its place, and where most deployments go wrong. Agents are genuinely good at the work that takes a person days: reconciling scattered signals into a live read, preserving context across a pass, watching telemetry continuously, drafting the recommendation. Put them there, keep clear controls on the decisions, and keep your people on the judgment.

What breaks is sequencing. An agent inherits whatever definitions already exist. If marketing, sales, and customer success each hold a different reading of the same account, automation does not resolve the conflict. It executes all three, faster, at machine speed. Definitions and decision rights first. System of record second. Orchestration third. Agents fourth.

And the part most efforts skip entirely: the team has to understand how the work changes and buy into it. Without that, the wiring sits unused.

The ratio underneath all of this

CAC is half of a fraction. The other half has the same defect.

Look again at the four seams. The first two, engagement to next action and SDR to AE, inflate what a customer costs. The second two, won to onboarding and adoption to expansion, suppress what a customer is worth. Cold onboarding seeds early churn. Unread telemetry means expansion moments pass unclaimed.

So LTV over CAC is wrong twice, and both times in the direction that feels good. The numerator assumes retention the handoffs are quietly forfeiting. The denominator leaves out the unrealized value entirely. Boards underwrite on that ratio. Investment committees price off it.

Which is also why the gate matters. Does the Shape of Revenue Strengthen the Business? tests every deal against four sides: margin, LTV, CAC, and EBITDA. A deal fits on the CAC side when it lands in line with the segment and pays back inside the window. But if the CAC in that test is the smallest number you have, the gate is measuring against a short ruler. Deals booked as clean fits may have forced the gate, and nothing in the review would show it.

Fix the seams and both halves of the ratio move. The gate starts measuring what it was built to measure.

Start with a number, not a slide

This does not need a transformation program. It needs one seam, sized and proven, before the next.

Pick the handoff you already suspect. Pull ninety days of it: what came in, what got acted on, how fast, and what happened next. Nobody needs a model to see the shape of it.

You probably already know which seam is yours. The open question is what it costs you, and whether anyone has ever put a number on it.

Sources. Salesforce State of Sales, 7th edition, published 3 February 2026, double-anonymous survey of 4,050 sales professionals across 22 countries, fielded August to September 2025. Gartner CMO Spend Survey 2026, fielded January to March 2026, 401 CMOs and marketing leaders across North America, the UK and Europe, most at companies above $1B revenue, released 11 May 2026. Duke Fuqua CMO Survey, 35th edition, fielded 7 to 29 January 2026, 308 U.S. marketing leaders, 97 percent VP-level or above. Benchmarkit 2025 B2B SaaS Performance Metrics Benchmarks, published May 2025, 583 private SaaS companies, sales and marketing as a percentage of revenue reported by 157. Figures for the representative business are indicative and follow the model set out above, not survey data.

Revenue Integrity Architecture™ is a framework for AI-ready revenue systems that compound quarter over quarter, and for the human and agent guardrails that keep the data and the numbers trustworthy.