Investment Banking AI Report, November 2026

The decision in brief

Reconcile AI usage to contractual entitlement, accepted delivery, invoicing, recognized revenue, collection and sustaining cost before using it in a transaction narrative. Each step answers a different question. A rising activity count or signed spending ceiling can support a commercial discussion, but it does not by itself establish recurring revenue, earnings quality or cash conversion.

Executive summary

AI-related revenue claims can move through a transaction process faster than their definitions. A company may describe usage growth as demand, a signed agreement as revenue and a temporary improvement in delivery cost as a durable margin advantage. Each statement may contain useful information, but the investment-banking team needs to understand the relationship between the measures before presenting the business to buyers, investors or lenders.

OFF's founder pricing instrument illustrates why that relationship matters. Among 148 respondents, 63 selected usage-based pricing, 36 flat subscription, 35 outcome-based and 29 per-seat; multiple selections were permitted. These are reported components, not audited contract structures or a revenue mix. A company may combine fixed and variable charges, minimum commitments, credits and acceptance conditions in a single customer relationship. [PRICE]

The broader planned-spend and return-timing questions should not be used to infer supplier revenue. The operator spending bands describe intended spending over the next twelve months, and the buying question asks when respondents expect measurable return. Neither establishes contracted demand for a named issuer, recognized revenue or realized customer economics. [SPEND; PAY]

November's contribution is a revenue-quality reconciliation with explicit units and time periods. It separates technical activity from the contractual event that creates a charge, then follows the charge through adjustments, accounting recognition and collection. A second analysis examines the cost of delivering the accepted service. A third addresses renewal and concentration, where headline usage can conceal a dependency on a small set of customers or unusual commercial terms.

The worked example is deliberately hypothetical. It shows how 1.2 million logged events can lead to $90,000 of net invoicing, $82,000 of recognized revenue under stipulated accounting conclusions and $75,000 of cash collected. The purpose is to make the distinctions visible, not to prescribe accounting treatment or imply that this conversion is typical of AI companies.

Transaction marketing requires a concise story whose core measures reconcile. The team need not put every operating detail in the main presentation, but it should be able to trace each material claim to a defined record and explain the exceptions. Precision in the supporting work allows the external narrative to remain clear.

What November adds

Earlier OFF dealmaker analysis emphasized operating evidence and the need to distinguish AI adoption from value. This edition focuses on the commercial chain between a technical event and a financial statement. That chain becomes especially important when the billing unit differs from the unit of delivery cost, when acceptance affects payment or when a fixed commitment coexists with variable consumption.

The first addition is a sequence of distinct questions. Did activity occur? Was it eligible under the contract? Did it create an enforceable charge under the agreed terms? Was the promised service delivered for accounting purposes? Was an invoice issued, adjusted or disputed? Was cash collected? Those questions should be answered with the relevant records rather than collapsed into a single label such as AI revenue.

The second addition is a treatment of commercial exceptions. Retries, test activity, credits, promotional usage, minimum commitments and customer acceptance can affect the relationship between activity and payment. The diligence team should understand their frequency and economic consequence. It should also know whether the rules changed during the period used to support growth or retention claims.

The third addition is a recurring-revenue assessment that preserves the terms of the customer relationship. A contract ceiling is not a minimum purchase. A minimum purchase is not necessarily the same as consumption in a particular period. An annualized recent month is a calculation, not automatically a recurring-revenue measure. The report supplies questions for examining those distinctions without imposing one metric definition on every business.

The edition does not claim a new investment-banking respondent cohort or a market valuation effect. It applies the OFF pricing evidence and verified primary guidance to a transaction-diligence problem. It also keeps its worked example separate from the Private Equity report's benefit bridge: this report follows customer revenue quality, while PE follows the transferability of operating benefits through ownership change.

Evidence and sample boundaries

The pricing question has 148 eligible unique respondents and allows multiple selections. It records how respondents describe charging for their products, not the detailed terms of a contract. The data cannot establish the proportion of revenue earned under each model, the profitability of a model or its effect on valuation. Recent founder-related rosters do not add new answers to this instrument.

The operator spending question has 76 responses. It records bands of planned next-twelve-month spending and includes an unknown category. The bands are not actual expenditure, signed commitments or supplier sales. Their open-ended upper category also prevents a defensible average spending estimate without additional assumptions, which this report does not make.

The return-timing question has 389 responses and permits multiple selections. It asks when a measurable return is expected, not when the entire investment will have paid back. It should not be used to forecast a supplier's collection period, customer retention or full investment break-even. [PAY]

All OFF measures are selective self-reports from registration data. No issuer-level financial information is linked to them. The revenue reconciliation below is a hypothetical teaching example with stipulated accounting conclusions. Actual recognition, presentation and disclosure require the applicable reporting framework, contract analysis and professional review.

Exhibit 1. Pricing evidence and its diligence implication

Exhibit note: PRICE, carried forward; multiple selections. These are respondent counts, not a revenue-weighted market distribution.

1. Define the commercial unit before presenting growth

A usage chart should identify the event it counts. A prompt, attempted task, completed workflow and accepted customer outcome are different units. Their growth can have different commercial meanings. The banking team should establish the definition and its relationship to the contract before using the chart as evidence of demand or revenue momentum.

The event record should distinguish customer activity from internal testing, retries and operational overhead. A product may generate more technical events because it performs more work per customer request. That can improve quality, increase cost or simply reflect a different implementation. It does not necessarily mean that customers are buying more. The relationship between raw activity and billable activity should be reproducible.

Changes in instrumentation can create false growth. A new release may log events that were previously invisible or split one event into several. The company should identify those changes and either restate the historical series on a comparable basis or explain the break. A transaction narrative should not rely on a continuous trend whose units changed without disclosure.

The eligible customer population also matters. Trial users, paying customers, internal users and partners may all generate activity. A total can be useful for product operations while being unsuitable as a measure of commercial demand. The team should preserve the segments relevant to the claim, including the treatment of promotional access and customer-specific concessions.

The contract may charge for a different unit from the product's main telemetry. An outcome-based product might record many attempts before one accepted result becomes billable. A subscription might permit substantial use without a variable charge. The diligence team should map those relationships rather than assume a direct price-times-usage formula. That mapping is the foundation for understanding both growth and cost sensitivity.

A clear commercial unit makes the external narrative easier to explain. The company can state what grew, which customers contributed and how the growth relates to payment. It can also explain why some activity is valuable without creating immediate revenue. The goal is not to eliminate operating metrics, but to use each metric for the claim it can actually support.

2. Read the agreement as a set of obligations

A signed customer agreement can contain several economically different elements. There may be a fixed service fee, a usage component, a minimum commitment, a maximum spending authorization and separate implementation work. The team should identify those elements and the conditions attached to each before treating the headline contract value as a single measure of recurring demand.

A ceiling and a minimum should never be confused. A customer authorized to spend up to an amount may have no obligation to use the full amount. A minimum commitment can create an obligation under specified terms, but the timing and nature of service delivery still matter. Cancellation, acceptance, renewal and adjustment provisions can materially affect the commercial interpretation.

Implementation work deserves particular attention. A company may earn an initial fee for integration or configuration while expecting a longer service relationship. The fee can be valuable, but its repeatability differs from an ongoing subscription or consumption charge. The diligence team should understand whether the work is standardized, customer-specific or necessary to make the promised service usable.

Credits and concessions should be visible in the contract analysis. Promotional usage can help acquire customers, while service credits may compensate for delivery problems. Both reduce the relationship between gross activity and realized commercial value. The company should distinguish the reason for the adjustment and explain whether it is expected to recur under normal terms.

The buyer's obligations can also affect delivery. The service may depend on timely data access, customer review or an integration maintained by the customer. If those prerequisites are absent, the supplier may be unable to deliver the intended result. The team should understand the practical dependency and its treatment in the agreement, with legal review of the actual terms where needed.

The output should be a contract schedule that preserves the relevant distinctions. It should identify what is promised, what creates entitlement to payment, what can change the amount and what happens at renewal or termination. This schedule supports financial diligence and the commercial story. It should not substitute a simplified sales-system field for the obligations that govern the relationship.

3. Reconcile billing, recognition and collection

Billing is an operational and contractual process. Revenue recognition is an accounting conclusion under the applicable framework. Collection is the receipt of cash. The three may occur in different periods and can differ in amount. A company with strong billing growth may still face recognition timing issues, disputes or weak collection. Each measure should be presented with its own definition.

The IFRS Foundation's overview of IFRS 15 describes a framework based on customer contracts, performance obligations, transaction price, allocation and satisfaction of obligations. It is a useful reference for understanding why a signed contract or invoice is not automatically identical to recognized revenue. The overview is not a complete accounting analysis and does not replace the requirements of another applicable framework, including US GAAP. [E-IFRS15]

The diligence team should ask the company's accounting advisers to explain the policy applied to the relevant contract types. The explanation should connect the policy to the actual promised service and supporting records. A general statement that the business is software does not resolve questions about implementation, variable consideration, acceptance or the timing of delivery.

Reconciliations should identify opening balances and adjustments where relevant. A period's cash collection may relate to earlier invoices, while current invoices may remain outstanding. Contract assets, liabilities, receivables and credits may affect the bridge. A simplified illustration can teach the distinctions, but an actual diligence schedule must account for the company's full balance movements rather than assume they are zero.

Disputes and credits deserve a commercial interpretation as well as an accounting treatment. A growing adjustment rate may indicate unclear billing definitions, delivery problems or aggressive initial terms. The team should examine the reason and customer concentration before drawing a conclusion. A one-time administrative correction has different implications from recurring disagreement about whether the service achieved the promised result.

The final transaction materials should use labels that match the underlying measure. Contracted amount, billings, recognized revenue and cash collected can all be relevant. The problem arises when one is presented as another or when a growth calculation silently changes its basis. A concise reconciliation can prevent that ambiguity and make the supporting diligence more efficient.

4. Examine the cost of accepted delivery

Revenue quality cannot be assessed in isolation from the service required to earn it. An AI company may incur costs for unsuccessful attempts, review, customer support and integration maintenance even when only accepted outcomes are billable. The diligence team should understand which activities drive cost and whether they scale with the commercial unit used in the revenue model.

The company should distinguish direct customer delivery costs from shared platform and company expenses, with a clear allocation method. This report does not prescribe accounting presentation. Its practical point is that a transaction model should not omit economically necessary work simply because it is recorded outside a convenient product-cost field. The relevant advisers should reconcile management's operating view to the financial statements.

Customer cohorts can have different economics. An early customer may receive extensive founder support, a large customer may negotiate a lower unit price and a complex customer may require more review. An average margin can conceal those differences. The team should examine the segments that materially affect the forecast and explain how the mix is expected to change.

Supplier pricing changes should be separated from operating improvements. A lower model price can improve economics, but the benefit may be offset by greater use, longer tasks or a shift to more expensive capabilities. A temporary discount should not be treated as a permanent advantage without examining the terms. The cost model should show which assumptions management controls and which depend on external relationships.

The accepted-outcome denominator helps connect quality and economics. A system can become cheaper per attempt while requiring more attempts or more human correction per accepted result. The team should preserve both measures where they matter. This allows an engineering improvement to be evaluated without automatically translating it into a financial improvement that the records do not support.

The sensitivity should include a change in the work needed to earn the same revenue. Using the hypothetical monthly example below, an additional $12,000 of necessary review would raise modeled delivery cost from $39,000 to $51,000. With recognized revenue held at the stipulated $82,000, the management contribution would fall from $43,000 to $31,000. This is a $12,000 reduction without any change in headline usage or invoicing. It is not an accounting gross-margin conclusion. The commercial question is whether the company can reduce that review burden, recover the cost through pricing or limit the workload that creates it. If the forecast assumes the burden disappears, the supporting evidence and timing should be visible.

Cost sensitivity should be tested against the proposed growth case. What happens if the next customer cohort is more demanding, if review effort persists or if the service must maintain a higher level of availability? A forecast can still be attractive under those conditions, but the transaction team should know which assumption drives the result. The external narrative should reflect the supported operating model rather than an idealized version of delivery.

5. Test recurrence at the point of customer choice

A recurring-revenue claim should explain why the customer is expected to continue paying and under what terms. A long agreement can support visibility while still containing volume variability, termination rights or service conditions. A recent month annualized may be a useful run-rate calculation, but it does not establish that the same activity will recur for twelve months. The definition should be explicit.

Renewal evidence should distinguish continued use from a renewed commercial commitment. A customer may remain active during an existing term while planning to reduce spending later. Another may use less because the product has become more efficient, while still valuing the relationship. The team should examine the contract event and the business reason for the change rather than infer retention from activity alone.

Expansion should be separated from price, volume and mix effects. A customer can spend more because it submits more work, moves to a higher-priced service or exhausts a promotional allowance. Those mechanisms have different implications for future growth. A useful retention schedule should preserve the calculation method and identify material changes in terms or classification.

Customer concentration can make headline growth less predictable. One large deployment may account for much of the increase, especially in an early company. The team should understand the customer's use case, remaining rollout, procurement conditions and renewal timing. It should also examine whether delivery knowledge or infrastructure built for that customer can be reused elsewhere.

Channel and platform dependencies deserve similar treatment. A company may acquire customers through a partner or depend on a platform for distribution. The commercial relationship can be valuable, but its terms and durability should be understood. The transaction story should identify what the company controls and what could change if the partner modifies access, economics or product scope.

The point is not to reject every variable revenue model. Consumption and outcome-based businesses can have attractive characteristics. The team should describe those characteristics using evidence appropriate to the model, rather than forcing them into a familiar metric without preserving the underlying meaning. A credible account of recurrence is specific about the customer choice and obligation that support it.

6. Keep the transaction narrative synchronized with the records

Transaction materials are often prepared by several teams using different source dates. A commercial presentation may contain a recent usage figure while the financial model uses an earlier contract schedule. A management interview may introduce a claim that has not passed through the same review. The team needs a controlled record of material claims so that the story remains consistent as the process evolves.

Each claim should have an owner, definition, source period and approved wording. The record should distinguish an observed result from management's forecast or aspiration. If the evidence changes, the team should identify which materials need updating. This is especially important when a simple phrase such as AI-driven growth carries several underlying assumptions about product use, customer payment and delivery economics.

The review should include the strongest reasonable alternative explanation. Usage growth may reflect a logging change; revenue growth may reflect a price adjustment; margin improvement may reflect a temporary subsidy. The team should determine whether the records support the preferred interpretation and describe any material limitation. The purpose is to improve the accuracy of the story before another party tests it.

Legal and accounting review should be attached to the actual claim and context. A statement suitable for an internal operating discussion may require different treatment in investor materials or a public filing. The applicable obligations depend on the transaction, jurisdiction and audience. The banking team should not assume that a general disclaimer cures an unsupported material statement.

The SEC's 2024 adviser cases concerning misleading AI-use statements illustrate a narrow but relevant point: claims about how AI is used can themselves require substantiation. The cases do not establish a universal rule for every transaction statement. They reinforce the practical need to verify the actual mechanism behind the label and involve the relevant advisers in the intended communication. [E-SEC-AIWASH]

A strong final narrative can remain concise. It should explain the customer problem, the commercial model, the quality of the supporting revenue and the assumptions in the forecast. The detailed reconciliations can sit behind it. The benefit of rigorous preparation is that management can answer follow-up questions consistently without changing definitions to preserve the headline.

External evidence and its limits

The IFRS Foundation's official IFRS 15 overview supplies the accounting distinction between contractual promises and recognition as obligations are satisfied. This report uses that distinction to frame diligence questions. It does not interpret the full standard for a particular contract or equate IFRS requirements with every other reporting framework. [E-IFRS15]

Stripe's current usage-billing documentation describes infrastructure for recording and charging for consumption. It is useful evidence that billing requires a defined usage model and implementation. It does not determine customer value, legal entitlement or revenue recognition. Product documentation should not be used as a substitute for the company's contract and accounting analysis. [E-STRIPE]

FinOps Foundation guidance on unit economics connects technical measures with business outcomes. That distinction is relevant when delivery cost is incurred per attempt but payment depends on an accepted result. The report's reconciliation and sensitivity questions are original applications. No FinOps benchmark margin or standard AI cost ratio is asserted. [E-FINOPS]

The SEC's March 2024 announcement concerns settled charges against two investment advisers over misleading descriptions of AI use. It is a specific enforcement example, not a finding about the companies represented in OFF's data or a general assessment of AI markets. The report uses it to support careful substantiation of the mechanism behind an AI claim. [E-SEC-AIWASH]

These sources address different layers of the commercial chain. Technical billing guidance can explain how activity becomes an invoice input; accounting guidance addresses recognition; cost guidance helps select a useful denominator; the enforcement example concerns the truthfulness of claims. A transaction team should preserve those boundaries rather than present a collection of sources as proof that its own revenue story is correct.

A live transaction may not allow reconstruction of every historical event. The team should prioritize claims that materially affect the model and narrative, identify which periods can be reconciled reliably and label remaining limitations. If a number cannot be substantiated in time, a narrower supported statement is preferable to a precise claim whose basis changes under questioning. That is an editorial and diligence recommendation, not a conclusion about any specific legal duty.

The source comparison should change the diligence request list. Billing documentation suggests asking for the meter specification and credit rules. Accounting guidance suggests asking how the company identifies its obligations and allocates the transaction price. Unit-economics guidance suggests reconciling attempts, accepted units and delivery expense over the same period. The enforcement example suggests checking the factual mechanism behind statements about AI use. Each request addresses a different failure in the revenue narrative. A clean usage export cannot answer all four, and neither can a management presentation. The adviser should preserve the unresolved difference in the working papers until the responsible technical, commercial or accounting owner supplies the missing explanation.

Evidence comparison

An illustrative usage-to-cash reconciliation

Assume a hypothetical company logs 1.2 million events during a month. Its contract charges $0.10 for each eligible event. The log includes 120,000 nonbillable retries and 80,000 internal test events. Removing those 200,000 events leaves one million eligible events and $100,000 of gross charges. The raw event count would have implied $120,000 if multiplied by the price without applying the contract definition.

Assume the company grants $10,000 of valid credits and adjustments, leaving $90,000 of net invoicing. Its accounting advisers conclude, under the applicable policies and contract facts stipulated for this example, that $82,000 is recognized in the current period and $8,000 relates to service to be provided later. The accounting conclusion is an assumption of the illustration, not a rule inferred from usage or invoicing.

Assume $75,000 is collected against this invoice by the review date, leaving $15,000 outstanding. There are no opening balances or other invoice movements in this simplified example. The cash amount is therefore different from both net invoicing and recognized revenue. An actual company reconciliation would need to include opening receivables, other collections, write-offs, contract balances and the relevant period movements.

Exhibit 2. Illustrative monthly reconciliation

Exhibit note: Hypothetical arithmetic, not an observed issuer result or accounting policy. Contract entitlement and recognition conclusions are stipulated. Actual balances and treatment require professional review.

Now consider delivery economics. Suppose the company attributes $39,000 of current-period delivery cost to the recognized service under a consistent management method. Subtracting that from $82,000 leaves $43,000 before other expenses omitted from this illustration. That is a management contribution calculation, not an audited gross-margin or profit measure. If the cost allocation excludes review or support necessary to deliver the service, the calculation must be revised before it is used to support a margin claim.

Exhibit 3. Questions that change the interpretation

Exhibit note: Proposed diligence questions; no universal metric definition or market valuation relationship is implied.

The transaction team should use the example as a model of the questions to ask, not as a conversion ratio. A subscription company, an outcome-priced service and a consumption platform can have very different relationships between events, invoices and revenue. The appropriate reconciliation begins with the actual contract and operating records. Its value is that it lets the team explain why the measures differ without treating the differences themselves as evidence of a problem.

The resulting narrative might emphasize strong customer adoption, a well-defined billing model and a growing base of accepted service, while separately describing the remaining uncertainty in recurrence or cost. It should avoid claiming that every logged event represents paid demand or that every signed amount is current revenue. The supporting record makes a concise and credible account possible.

The next diligence request

Ask management for one reconciled period connecting the product's main activity measure to contractual eligibility, adjustments, accounting recognition and collection. Choose a period and customer set material to the transaction story. Record the definitions and source dates before discussing growth rates. This will reveal whether the headline measures describe the same underlying business activity.

Then examine the cost of delivering the accepted service and the customer decision that supports recurrence. A company can have strong adoption and still need to improve billing clarity, cost allocation or renewal evidence. Those findings should affect the forecast and the wording of the transaction narrative in proportion to their significance.

November's recommendation is to preserve the meaning of each commercial and financial measure as it moves through the process. Usage, contracts, revenue and cash can all tell a useful story. The banking team's work is to explain their relationship accurately enough that a buyer or investor can assess the business without relying on the AI label to fill the gaps.

Dealmaker questions

Is this a survey of investment bankers?

No. It is a revenue-quality synthesis using broader OFF evidence and primary guidance.

Does planned AI spending equal supplier revenue?

No. SPEND records intended next-twelve-month bands, not actual expenditure or contracted supplier sales.

Does measurable return mean full investment payback?

No. PAY asks about expected measurable return, which is a different concept.

Is a signed spending ceiling recurring revenue?

A ceiling alone does not establish a minimum purchase, delivery or recognition. The actual contract and accounting facts must be assessed.

Does the worked example prescribe IFRS or US GAAP treatment?

No. Recognition is a stipulated assumption. The example illustrates distinctions and arithmetic, not a contract-specific accounting conclusion.

Are the pricing selections a revenue mix?

No. They are multiple-selection respondent counts, not revenue-weighted company categories. [PRICE]

Citable facts

- Usage-based pricing selected: 63/148, 43%, carried forward. [PRICE]

- Outcome-based pricing selected: 35/148, 24%, carried forward. [PRICE]

- Planned next-twelve-month spending base: 76; not a measure of actual expenditure. [SPEND]

Methods and limitations

Calculation-v1.1 uses authorized 1 November exports through the stated cutoff. Byte-identical exports and invited-only records are excluded. Latest eligible nonblank answers are selected per normalized respondent identifier per question. The relevant pricing, spending and return-timing measures reconcile to the published baseline and are carried forward.

The sample is selective and self-reported. No issuer financial statements, transaction results or investment-banking subgroup are linked to these answers. No revenue forecast, valuation premium or comparative pricing performance is inferred. The worked reconciliation is hypothetical and its accounting conclusions are expressly stipulated.

References

[OFF-DATA] Open Future Forum. Executive AI Statistics, November 2026. PRICE, SPEND and PAY.

[E-IFRS15] IFRS Foundation. IFRS 15 Revenue from Contracts with Customers, official overview consulted 2 November 2026. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-15-revenue-from-contracts-with-customers/

[E-STRIPE] Stripe. Basic usage-based billing, current documentation consulted 2 November 2026. https://docs.stripe.com/billing/subscriptions/usage-based

[E-FINOPS] FinOps Foundation. Unit Economics capability. https://www.finops.org/framework/capabilities/unit-economics/

[E-SEC-AIWASH] US Securities and Exchange Commission. SEC Charges Two Investment Advisers with Making False and Misleading Statements About Their Use of Artificial Intelligence. 18 March 2024. https://www.sec.gov/newsroom/press-releases/2024-36

Related research and citation

The November YC Founders report examines pricing design, CFO qualifies buyer benefits, Private Equity follows benefits through ownership change, and General Counsel examines authorization and contractual allocation.

Suggested citation: Newlands, Murray. Investment Banking AI Report, November 2026, Revenue-Quality and Transaction Synthesis. Open Future Forum. Evidence cutoff 31 October 2026.

Open Future Forum research: https://openfutureforum.com/research/

Editable working tool

Usage-to-cash reconciliation: download the editable CSV worksheet. The worksheet implements the report's recommendation; it is not a validated score, mandatory standard or completed assessment.

November report collection

• Executive AI Leverage Report
• CFO AI Leverage Report
• CEO AI Leverage Report
• CISO AI Leverage Report
• CMO AI Leverage Report
• AI Transformation Report
• AI Leaders AI Leverage Report
• Board Director AI Governance Report
• YC Founder AI Report
• VC and CVC AI Investment Report
• Investor AI Report
• Private Equity AI Report
• General Counsel AI Report
• Executive AI Statistics

Share LinkedIn Twitter / X