OpenEvidence Reaches 40% of U.S. Physicians: What the $12B Medical AI Playbook Got Right

When the CFO of a $48M ARR vertical SaaS company circulated a 22 percent operating expense reduction memo in March 2026, the marketing team had 11 business days to defend every channel line item on a contribution margin basis. According to the Gartner CMO Strategic Insights Survey 2025, 71 percent of CMOs faced a budget cut of more than 8 percent during 2025, and 43 percent reported their CFOs explicitly demanded contribution margin defense for every channel — not ROI, not influenced pipeline, but contribution margin per dollar against the benchmark of paid acquisition. That shift in the finance conversation is the most important structural change in marketing budget governance since the move to attribution-based reporting in the mid-2010s, and it is the reason a working AEO contribution margin spreadsheet is now a survival document.

This article is the spreadsheet. It is the variable cost categorization that finance teams accept, the AI-attributed revenue model with explicit confidence intervals, the contribution margin calculation, and the comparison framework that puts AEO next to paid search, paid social, outbound, and partnerships on the same per-dollar basis. The math is built from a 14-company anonymized cohort tracked between Q3 2024 and Q1 2026 against published benchmarks from the Bessemer State of the Cloud 2025 report, the SaaS Capital Index quarterly metrics, and the OpenView 2025 SaaS Benchmarks Report. Every number is auditable. Every assumption is explicit. The point is to walk into the budget review with a finance-grade document instead of a marketing narrative.

Why Contribution Margin, Not ROI

The first conversation to win is the framing conversation. Marketing leaders default to ROI because the ROI percentage is intuitive and rolls up neatly to a single number. CFOs increasingly reject ROI as the primary screen for marketing budget decisions because ROI conflates three distinct financial questions and gives no usable answer to any of them.

The first question is whether to invest in the channel at all. That is a payback period and IRR question. The framework for AEO specifically is laid out in the companion piece on calculating AEO ROI with the CFO-ready payback period model, which walks through input cost models, attribution proxies, and sensitivity analysis. That model is what you use to justify standing up the program in the first place.

The second question is whether to expand or contract the investment at the margin. That is a contribution margin question, and it is the question the budget review is asking. When the CFO says we need to cut $2.4M from marketing operating expense, the right channel-by-channel comparison is contribution margin per dollar of variable cost, because the channels with the highest contribution margin per dollar are the ones that should keep their budgets and absorb less of the cut. ROI percentages obscure this comparison because they include allocated fixed costs and historical investments that are not relevant to the marginal decision.

The third question is whether the channel is producing healthy unit economics relative to the company's growth profile. That is a CAC payback and LTV/CAC question, and the cohort-level math for AI-acquired customers is detailed in our 12-month AI-acquired LTV/CAC payback deep analysis. Most healthy AEO programs produce AI-acquired CAC of $34 to $42 blended, LTV/CAC of 4.8x, and CAC payback of 7.2 months at the cohort median, which clears most growth-stage CFO hurdles.

Contribution margin is the framework that wins the budget review specifically because it is the framework the CFO uses for every other operating expense category. It is symmetrical with how finance evaluates production costs, support costs, and infrastructure costs. The marketing team that brings a contribution margin spreadsheet to the review is speaking the language the finance team uses for the rest of the P&L, and that linguistic alignment matters more than any individual number on the page.

The Fully-Loaded Variable Cost Stack

The fully-loaded cost calculation is the most contested part of the spreadsheet because it is where teams habitually undercount and where CFOs habitually push back. The conservative methodology that finance teams accept includes every variable expense line that would not exist if the AEO program were shut down, plus a fair allocation share of shared functions that the program meaningfully consumes.

The cost categories that belong in the calculation:

The line that surprises marketing teams is the amortized evaluation infrastructure. Most AEO programs maintain a citation tracking and evaluation harness — Profound, Otterly, Peec, Ahrefs Brand Radar, or an internal build that queries ChatGPT, Claude, Perplexity, Gemini, and Bing daily across a controlled prompt corpus. The capital cost of building or licensing that harness, including the data warehouse storage and the API call costs to the model providers, runs $36,000 to $104,000 in the first year and amortizes down across the useful life of the system. CFOs expect to see this line, and they expect to see the amortization schedule, because it is exactly how they treat every other capitalized intangible asset in the business.

The line that marketing teams habitually omit is the brand and PR fair share. The PR work that gets a company onto third-party review sites, into industry awards, and into Wikipedia is not a separate program from AEO — it is a core input to citation infrastructure. A defensible AEO cost stack allocates 15 to 25 percent of the brand and PR budget to AEO because that is roughly the share of brand activity that produces measurable citation outcomes. Excluding it makes the AEO program look cheaper than it is, which feels like a win until the CFO finds the omission in review and the entire spreadsheet loses credibility.

The line that finance teams scrutinize most is the agency or contractor fee. Every dollar paid to an external content production firm, comparison-page editorial agency, or AEO consultancy belongs in the variable cost line. The full-service AEO agency engagement that runs $15,000 to $40,000 per month for a mid-market program is a direct variable cost of the AEO output, and pretending otherwise distorts the comparison against paid search where the agency fee is similarly counted as a variable cost.

Measuring AI-Attributed Revenue With a Confidence Interval

The revenue side of the calculation is where every AEO budget defense lives or dies. The marketing team that walks into the review with a single AI-attributed revenue number gets challenged on the attribution methodology and loses. The marketing team that walks in with a conservative number, a base case, and an optimistic case, all on the same documented methodology, gets a substantive conversation and usually keeps the budget.

The three independent attribution signals to combine:

Direct referrer attribution. ChatGPT now passes identifiable referrer data for a meaningful share of clicks — typically 8 to 15 percent of true AI-influenced traffic in 2026 based on cohort measurement. Perplexity passes referrers for nearly all assistant-driven clicks. Claude passes referrers for a smaller share. Gemini passes referrers from the surface that overlaps with Google search. Any program that has not instrumented all four cleanly is starting from a worse evidentiary position than necessary. The direct referrer signal is the floor of AI-attributed traffic — it captures a subset of true influence, not the totality.

Self-reported intake survey. Every new lead, demo request, and trial signup completes a single-question survey asking how they first encountered the company. The response options include each major AI assistant by name, branded search, organic search, paid search, paid social, podcast, conference, peer recommendation, and other. The methodology that the HubSpot 2026 State of Marketing Report documents shows that self-reported attribution captures roughly 25 to 40 percent of true AI-influenced pipeline with reasonable accuracy, with the caveat that self-report skews high in categories where AI assistants have visible UX cues and low in categories where the assistant references the brand without an obvious citation surface.

Branded search lift modeling. Year-over-year branded search query growth above a counterfactual trendline is partially attributable to AI citation visibility, because AI-cited brands generate downstream branded queries from buyers who saw the brand mentioned in an assistant response and later searched the brand name directly. The methodology requires building a counterfactual trendline from the pre-AEO baseline and attributing only the residual lift above the trendline to AEO contribution. The conservative attribution share is 25 to 40 percent of the residual; the optimistic share is 50 to 70 percent. The branded search lift modeling is the most contested signal because the counterfactual is necessarily an estimate.

The conservative AI-attributed revenue number for the spreadsheet uses direct referrer attribution plus survey attribution and excludes the branded search lift modeling. The base case adds the conservative branded search attribution. The optimistic case adds the higher branded search attribution. The range between conservative and optimistic is the 80 percent confidence interval that goes into the spreadsheet cell, with conservative displayed as the headline number for budget defense purposes. Defending the conservative number gives the most surplus when actual performance exceeds the floor, and exceeding the budget defense projection is a substantively better outcome than missing the base case.

The segment-level math for vertical SaaS, horizontal SaaS, and developer infrastructure broken out separately, including the activation engineering pattern that closes the AI-acquired LTV gap, sits in the cohort analysis of AEO-acquired customer LTV, which is the natural deep-dive on the revenue side of the contribution margin calculation.

The Contribution Margin Calculation

With fully-loaded variable cost and AI-attributed revenue established, the contribution margin calculation is mechanical:

The two numbers the CFO will focus on are the contribution margin percentage and the gross profit per dollar of variable cost. The contribution margin percentage of 45.3 percent in the example sits comfortably in the healthy range for a software channel and is the kind of number that survives a budget cut without modification. The gross profit per dollar of variable cost of $2.39 is the number that goes into the channel comparison and is the cleanest defense against substitution by paid search or paid social.

The sensitivity analysis the CFO will request next is the impact of varying assumptions on the contribution margin. The three sensitivities that matter:

The attribution share sensitivity. Run the calculation with the conservative attribution share at 70 percent, 100 percent, and 130 percent of the documented value to show how contribution margin shifts with attribution methodology. The output range typically runs from 32 percent contribution margin at the conservative end to 58 percent at the higher attribution assumption, and the entire range is healthy enough to defend the budget.

The cost increase sensitivity. Run the calculation with the fully-loaded variable cost increased by 15 percent and 30 percent to show what happens if the program grows. The contribution margin compresses modestly because gross profit grows faster than the linear cost increase in a compounding content asset, which is the structural reason AEO contribution margin tends to improve at higher spend levels rather than degrade.

The gross margin sensitivity. Run the calculation with product gross margin at the actual company number, 5 points lower, and 5 points higher. The contribution margin moves nearly one-for-one with product gross margin, which is why AEO contribution margin defense is structurally easier at high-gross-margin software companies than at lower-margin businesses.

The Per-Dollar Channel Comparison

The channel comparison is the table that wins the budget defense. The CFO's question is not whether AEO produces revenue; it is whether AEO produces more contribution margin per dollar of variable cost than the marginal alternative use of that dollar. The marginal alternative is almost always paid search or paid social, because those are the channels with elastic spend that can be scaled up if AEO is cut.

The comparison table across the 14-company cohort:

Three observations matter for the defense.

The first is that AEO is structurally a top-three contribution margin channel in the cohort, behind only branded search and partnerships. That is a strong position because branded search is largely a downstream consequence of upstream investments including AEO, and partnerships have a hard ceiling on scalability that AEO does not have. AEO is the highest-margin channel with meaningful incremental scalability in the cohort.

The second is that paid social is structurally underperforming on contribution margin because of the compounding rise in CPMs and CPCs across LinkedIn, Meta, and programmatic in 2024 and 2025. The 14-point year-over-year decline in paid social contribution margin is the largest single mover in the table, and it is the reason most cohort companies have already been reducing paid social spend even before the broader budget cut conversation. The replacement spend has gone substantially to AEO and partnerships.

The third is that paid search, while still healthy at 22 percent contribution margin, has declined 8 points year over year and faces continued CPC inflation in 2026. The structural pressure on paid search contribution margin is the reason the CFO is willing to entertain reallocating from paid search to AEO at the margin, provided the AEO contribution margin defense is credible. The numbers in the table give the marketing leader the basis for that conversation.

The supporting analysis on what each of the seven board-level AEO metrics should show and how they should be presented in the quarterly review is laid out in the CMO AEO dashboard for the board deck, which is the natural follow-on document after the contribution margin spreadsheet wins the operating budget review.

Three Tactical Traps That Sink the Defense

The AEO contribution margin number gets attacked in the budget review along three predictable lines: revenue cannibalization with SEO, attribution overcounting, and inconsistent CAC calculation methodology. Each attack is handled in the spreadsheet itself rather than in the meeting, with documented adjustments and methodology notes that pre-empt the live objection.

Revenue cannibalization with SEO

The most common attack on the AEO contribution margin number is the cannibalization argument. The CFO or the head of organic asks whether the AI-attributed revenue is incremental or whether it is revenue that would have come from SEO anyway and is now being credited to AEO because the same buyer touched both surfaces.

The argument is partially correct and requires a measured response in the spreadsheet rather than a dismissal in the meeting.

The empirical observation from the 14-company cohort is that cannibalization between SEO and AEO runs at roughly 18 to 31 percent in the first 12 months of an AEO program and stabilizes at 12 to 22 percent in steady state. That is, of the gross AI-attributed revenue, 12 to 22 percent in steady state would likely have been captured by SEO if the AEO program had not existed. The remaining 78 to 88 percent is incremental.

The defensible response in the spreadsheet is to display the AI-attributed revenue both gross and net of cannibalization, with the cannibalization adjustment shown explicitly as a separate line. The contribution margin calculation that goes into the budget defense uses the net-of-cannibalization number, which is the substantively correct figure for the incremental decision. The headline AI-attributed revenue is still the gross number for reporting purposes, with the net figure as the financially loaded number.

The deeper question the cannibalization framing raises is whether SEO and AEO should be managed as a combined channel for contribution margin purposes. The cohort companies that have done this produce a more defensible single combined number than companies that treat them separately, because the operational dependencies between the two surfaces (shared content, shared schema, shared technical infrastructure) make the cost allocation messy when they are reported separately. The integrated reporting approach is gaining traction in 2026 and is the recommended structure for any company building the AEO contribution margin spreadsheet from scratch.

Attribution overcounting

The second common attack on the contribution margin number is the attribution overcounting argument. The CFO asks how confident you are that the survey-attributed revenue is real, and what the false-positive rate is on the self-reported attribution methodology.

This is the question to prepare for in detail because the answer establishes the credibility of the entire spreadsheet.

The empirical false-positive rate on intake survey AI attribution, validated by post-hoc interview against a sample of attributed customers in the cohort, runs at 12 to 19 percent. That is, of the customers who said they discovered the company via ChatGPT or Perplexity on the intake survey, 12 to 19 percent could not on follow-up substantiate the AI discovery story. They either misremembered the source, conflated AI with traditional search, or were guessing on the survey question.

The defensible response in the spreadsheet is to apply a 15 percent haircut to survey-attributed revenue as a documented adjustment line. The math is conservative — 15 percent is the midpoint of the empirical false-positive range — and the explicit adjustment shows the CFO that you have thought about the attribution risk and built it into the model. The remaining survey attribution is robust enough to defend in the meeting.

The opposite concern that finance teams sometimes raise is the false-negative rate — the share of true AI-influenced revenue that is not captured by any of the three attribution signals. The cohort evidence suggests the false-negative rate is meaningfully larger than the false-positive rate, which means the conservative contribution margin number in the spreadsheet is biased downward, not upward. That asymmetry is important to surface in the meeting because it gives the CFO a reason to trust the headline number as a floor rather than a ceiling.

CAC calculation methodology

The third common attack is the CAC calculation methodology argument. The CFO asks how the per-customer cost of acquisition is calculated and whether the same methodology is being applied consistently across channels.

This is where consistency wins. The methodology has to be identical for AEO, paid search, paid social, outbound, and partnerships, or the contribution margin comparison loses meaning. The standard methodology that holds up:

Allocate the fully-loaded variable cost of the channel over the number of customers attributed to that channel in the same period. The cost includes salaries and benefits, agency fees, tooling, content production, and the fair share of shared functions. The customer count uses the same attribution methodology applied uniformly across channels — direct referrer, survey, and modeled attribution combined with the same haircuts and confidence intervals.

The trap is using a different attribution methodology for AEO than for paid search. Marketing teams sometimes use last-click attribution for paid search (which inflates the paid search numerator) and multi-touch attribution for AEO (which spreads credit). The fix is to use the same multi-touch methodology for both, which usually pulls the paid search numbers down and the AEO numbers up. The CFO is checking for methodological consistency, not absolute accuracy, and the spreadsheet that demonstrates consistency wins the meeting.

The Numbered Playbook for the Budget Defense

The actual sequence of work to build the spreadsheet and survive the budget review:

1. Document the fully-loaded variable cost stack by walking through every line item with the finance business partner. Include content team salaries and benefits, editor cost, technical implementation, AEO tooling, agency fees, amortized evaluation infrastructure, brand and PR fair share, and Wikipedia and third-party authority work. Get the finance business partner to sign off on the cost categorization before the meeting. The mid-market range is $432,000 to $1,160,000 fully loaded.

2. Build the three attribution signals into a single revenue model with conservative, base, and optimistic cases. Direct referrer attribution from ChatGPT, Perplexity, Claude, and Gemini. Self-reported intake survey attribution with a documented 15 percent false-positive haircut. Branded search lift modeling above the counterfactual trendline, with conservative and optimistic attribution shares. The conservative case is the headline number for the budget defense.

3. Calculate the contribution margin in absolute dollars and percentage using the company-standard product gross margin. Show the calculation cell-by-cell with formulas visible. Include the revenue-per-dollar and gross-profit-per-dollar derivations because those are the comparison-ready numbers.

4. Build the channel comparison table putting AEO contribution margin per dollar next to paid search, paid social, outbound, partnerships, and branded search. Use the same attribution methodology across all channels — typically multi-touch with a documented decay curve. Include year-over-year change for each channel because the trend matters as much as the level.

5. Run the three sensitivity analyses on attribution share, cost increase, and gross margin. Show the contribution margin range under each sensitivity so the CFO can see how the number moves with the assumptions. The robustness of the contribution margin across sensitivities is the implicit argument for the defensibility of the channel.

6. Address the three tactical traps explicitly in the spreadsheet with documented adjustments. The cannibalization adjustment of 12 to 22 percent against SEO. The false-positive haircut of 15 percent on survey attribution. The methodological consistency footnote on CAC calculation. Pre-emptive disclosure of the standard objections eliminates them as live attacks in the meeting.

7. Prepare the one-page summary for the CFO and a deeper backup for the finance team to review independently. The one-page summary is the contribution margin number, the channel comparison, the year-over-year change, and the conservative attribution methodology. The backup is the full spreadsheet, the cost stack documentation, the attribution methodology notes, and the sensitivity analyses. Both documents should be available in the meeting.

Defending Against the Cut Memo Specifically

When the cut memo arrives — and across the cohort companies in 2025 and 2026, 71 percent of CMOs received one — the defense conversation is compressed and pattern-matched. The CFO is working through every channel in roughly the same order and applying roughly the same screen: contribution margin per dollar, year-over-year trend, and strategic relevance to the company's growth profile.

The AEO defense in that conversation has three structural advantages and one structural disadvantage.

The first advantage is that AEO contribution margin is typically higher than paid search and paid social. The spreadsheet number does the work in the meeting if the methodology is sound.

The second advantage is that AEO is a compounding asset rather than a perishable spend. The CFO understands compounding asset valuation from the way they treat capitalized software development and R&D investment. The argument that cutting AEO destroys an accumulating citation share asset that takes 9 to 18 months to rebuild is an argument the CFO understands, which means the channel survives even modest cuts more often than perishable channels do.

The third advantage is that the AEO budget is structurally smaller than paid budget in most companies, which means the absolute dollar reduction available from cutting AEO is small relative to the disruption cost. CFOs running a triage exercise will sometimes spare smaller programs that have asset value and concentrate the cut on larger perishable programs where the absolute dollar reduction is meaningful.

The disadvantage is that AEO attribution is probabilistic rather than deterministic, which means the contribution margin number carries an implicit confidence interval that the CFO will probe. The defense for this disadvantage is the explicit confidence interval in the spreadsheet — presenting the conservative number as the headline removes the attribution uncertainty as an attack vector, because the CFO can see that the conservative assumption is already baked into the floor.

The companies in the cohort that lost AEO budget in 2025 and 2026 had two things in common: they presented ROI percentages instead of contribution margin, and they did not have an explicit confidence interval in their attribution methodology. The companies that kept or expanded AEO budget had a contribution margin spreadsheet that mirrored the structure outlined in this article, and they walked into the meeting with a finance-grade document rather than a marketing narrative.

The SaaS Capital Index annual benchmark report data from late 2025 confirms the broader pattern: companies that maintained or expanded AEO investment through the 2025 budget compression cycle reported a 28 percent revenue growth premium relative to companies that cut, with the bulk of the growth differential appearing in 2026 quarters as the citation share asset compounded. That growth premium is the substantive reason the contribution margin defense is worth winning.

What Changes for Enterprise Versus Mid-Market

The contribution margin framework scales across company size, but the specific numbers shift in ways worth noting.

Enterprise companies above $250M ARR typically have fully-loaded variable AEO costs in the $1.4M to $3.2M range, driven by larger dedicated content teams, more sophisticated tooling stacks, and multi-language content programs that cost significantly more than the mid-market English-only baseline. The corresponding AI-attributed revenue at enterprise is also higher in absolute terms — typically $8M to $24M in the cohort's enterprise segment — and the contribution margin percentage is comparable to mid-market at 41 to 56 percent. The enterprise case is harder to make on a per-dollar basis because paid search and paid social at enterprise scale benefit from buying power that compresses unit costs.

Mid-market companies between $25M and $250M ARR sit in the cost and revenue ranges used throughout the article. The contribution margin defense is most straightforward at mid-market because the fully-loaded cost stack is small enough to itemize cleanly and the AI-attributed revenue is large enough to be material.

Early-stage companies below $25M ARR run AEO programs at $80,000 to $260,000 fully-loaded cost and produce AI-attributed revenue of $200,000 to $800,000 in steady state. The contribution margin percentage at early stage runs lower at 28 to 44 percent because the cost base has not yet scaled into the operational efficiency that mid-market and enterprise programs achieve. The defense at early stage relies more on the compounding asset argument and the strategic growth profile than on the per-dollar comparison against paid channels, because at small scale paid channels can be more efficient per dollar in the short run before the AEO asset matures.

Takeaway: The AEO contribution margin spreadsheet is now a survival document, not a nice-to-have. Build the fully-loaded variable cost stack including amortized evaluation infrastructure and brand fair share. Measure AI-attributed revenue with three independent signals and present the conservative case as the headline for budget defense. Calculate contribution margin in absolute dollars, percentage, and per-dollar-of-variable-cost terms. Compare against paid search, paid social, outbound, and partnerships on a methodologically consistent basis. Address the three tactical traps — cannibalization, attribution overcounting, and CAC methodology — explicitly in the spreadsheet rather than waiting for them in the meeting. Cohort companies that lost AEO budget through the 2025 to 2026 compression cycle presented ROI percentages and got cut; the companies that walked in with a contribution margin spreadsheet kept their budgets and captured the documented growth premium.