Young companies often combine incomplete controls with rapid decisions, concentrated customers and overlapping responsibilities. This makes it difficult to distinguish disorganization, commercial optimism and fraud. For a seed fund, the investigation starts with an economic question: will the capital finance a repeatable product, or circulate through relationships that benefit people close to the founders? Useful technology can coexist with contracts that divert margin, conceal obligations or undermine reported metrics.
The central question: who receives the value and who bears the risk?
Fraud involves deliberate deception. Accounting errors, conflicts of interest and related-party transactions are different categories. A relative's business may provide a necessary service at a competitive price; an independent customer may participate in a fabricated sale. The relationship label does not settle the issue. Investors must reconstruct the transaction's substance, identify decision-makers on both sides and compare the product, price, payment and risk with the account presented. Suspicion guides diligence; it is not a finding.
The investigation must also address the fund's investment thesis. A scalable business needs evidence of independent demand, repeatable delivery and a financing strategy consistent with its capital needs. Apparent traction may disappear if purchases depend on the founders' other companies. The aim is to determine how much of the thesis survives corrected information, what conditions an investment would require and which uncertainties make the opportunity unsuitable, even without proof of fraud.
Map related parties beyond the shareholder register
Start with founders, directors, significant investors and those responsible for purchasing, sales and payments. Extend the map to controlled companies, previous businesses, involved relatives and people with practical influence over decisions. Indirect links matter: a consultancy may subcontract a shareholder's company; a distributor may have received a personal loan from a founder; an apparently external customer may depend economically on another group company. The formal ownership chart captures only part of this network.
Proportionate information gathering combines conflict declarations with questions about specific transactions. Who introduced the supplier, negotiated the price and approved payment? Does anyone hold an economic interest, receive a commission, provide a loan or expect an ownership interest outside the main contract? Relevant documents should support the answers. This does not justify indiscriminate collection of personal information: establish the relationship relevant to the investment risk, restrict access and protect information unnecessary to the decision.
The useful output is a transaction register, rather than a list of names alone. Record the counterparty, relationship, purpose, contracted amount, outstanding balance, approval process and evidence of delivery. A declaration that no conflicts exist also needs an accountable person and updates when circumstances change. If an undisclosed relationship emerges, assess its materiality, the explanation and the implications for other representations. Voluntary correction differs from repeatedly changing a story as contradictory documents appear.
Artificial revenue: trace the contract, delivery, billing and money
A Brazilian tax invoice, or nota fiscal, does not by itself establish high-quality recurring revenue. Connect the contractual obligation to delivery, acceptance where applicable, billing and collection. For software, check whether external users actually use the product and have an effective obligation to pay. A channel sale may conceal consignment, broad return rights or unmet conditions. Rising invoiced amounts say little if a transaction presented as final still depends on the buyer's confirmation.
Circular funding is another concern. A startup pays a consultancy connected to a shareholder; the consultancy transfers money to a supposed customer; that customer pays the startup. A bank statement shows a receipt, but not independent demand. Other possibilities include undisclosed discounts, later reimbursements, founder-funded payments and reciprocal purchases without an economic purpose. Within authorized access and the agreed scope, examine the relevant sources and destinations of funds and any obligation to return them.
Related-party revenue should neither be automatically removed from the accounts nor accepted without adjusting the investment thesis. Separate total revenue, independent-customer revenue, substantiated recurring revenue and amounts subject to conditions. Accounting recognition and commercial validation answer different questions: revenue may be properly recognized without demonstrating a broad market. Investment pricing should reflect the ability to attract and retain customers who have no separate interest in supporting the company's reported performance.
Market demand and unit economics: where dependence can hide
Related parties can distort revenue and the cost of serving customers. A friendly customer may pay above the usual price while a founder's company supplies infrastructure free of charge. The resulting margin does not demonstrate repeatable efficiency. Ask what would happen to pricing, conversion, retention and support if independent counterparties replaced those relationships. The exercise need not predict the future precisely; it should expose the conditions on which the reported economics depend.
Customer acquisition cost, or CAC, needs the same scrutiny. Commissions paid by another company, founders' sales work outside payroll and campaigns booked as development can understate acquisition costs. A long contract does not establish retention if customers can cancel easily, fail to pay or never use the product. Compare independent-customer cohorts, channels and customer profiles. Customer lifetime value estimates should disclose limited cohort history and sensitivity to cancellation, rather than treating immature observations as established behavior.
Seed investing necessarily involves hypotheses. Problems arise when projections appear as facts. 'Customers will stay for several years' is a forecast; 'this cohort paid and used the product during the observed periods' is verifiable. Preserve that distinction in the investment memorandum. Include replacement costs if margins rely on a related supplier accepting losses. If sales depend on a founder's contacts, explain how customer acquisition could move to an operation capable of growing beyond that personal network.
A numerical example: from reported revenue to a defensible economic base
Consider Aurora, a fictional management-software company. Every amount in this example is hypothetical arithmetic, not a market benchmark. Its presentation reports R$100,000 in monthly revenue, R$20,000 in direct costs and an 80% gross margin. Diligence divides revenue into R$55,000 of subscriptions from independent customers, R$10,000 of one-off services for independent customers, R$20,000 of contracts with a founder's companies and R$15,000 invoiced to a distributor whose payment obligation remains conditional.
These four amounts total R$100,000, but support different conclusions. Exclude the conditional R$15,000 from the scenario's confirmed revenue until the outstanding conditions are clarified. Show the related-party R$20,000 separately, even if its contracts are valid and paid. This leaves R$65,000 of confirmed independent revenue, of which R$55,000 is recurring. Final accounting treatment requires its own assessment. The investment analysis must keep invoicing, independence and recurrence distinct: none establishes the other two.
Looking only at independent recurring services, the fund identifies R$22,000 in necessary monthly costs, including expenses previously paid by a related company. Gross profit for this base is R$33,000, and R$33,000 divided by R$55,000 gives a 60% margin. This is not Aurora's corrected consolidated margin: the other contracts need their own cost analysis. Comparing the reported 80% with 60% requires explaining the different scope. The revised thesis rests on R$55,000 of independent recurring revenue and its corresponding costs.
Valuation and dilution: revise the price without buying an undisclosed problem
Aurora's initial proposal is a R$3 million investment at a R$9 million pre-money valuation. Post-money value would be R$12 million, giving the fund 25%. Before the round, the fully diluted capitalization comprises 80% for founders, 10% for an angel investor and a 10% incentive reserve. With no other adjustments, these become 60%, 7.5% and 7.5%, alongside the fund's 25%. The calculation assumes no convertibles, additional options or other rights that change the denominator.
After reviewing revenue quality, the committee might instead consider a R$6 million pre-money valuation. The same R$3 million would acquire one-third of the fully diluted capital. Founders would retain approximately 53.33%, with approximately 6.67% each for the angel and incentive reserve, subject to rounding. This is an alternative scenario, not a continuation of the first cap table. A lower price reflects a revised economic thesis; it does not cure fabricated contracts, reveal hidden liabilities or repair an unsustainable relationship of trust.
Separate price negotiation from conditions for investing. An honest disagreement about growth may justify a lower valuation or funding conditional on verifiable milestones. Deliberately concealed flows may justify ending diligence. Equity and convertible instruments require documentation, signing authority and distribution arrangements appropriate to their structure; renaming an instrument does not fix underlying defects. Model valuation caps, discounts, contractual interest, conversion rights and new incentive reserves so that participants understand the dilution they are agreeing to.
Cash and timing: how diligence changes the funding decision
Aurora reports R$800,000 in bank accounts. Reconciliation identifies R$80,000 unavailable for general expenses, leaving R$720,000 of unrestricted cash. At net cash burn of R$180,000 per month, estimated runway is four months. That calculation assumes constant burn and effective availability of the R$720,000. Overdue obligations, customer refunds and unrecorded related-party payments may shorten it. A bank balance is a starting point; restrictions and obligations determine how much can actually fund operations.
Now assume the R$3 million round closes at the projection's start, with R$150,000 in transaction expenses. Available cash becomes R$3.57 million: R$720,000 plus net proceeds of R$2.85 million. Planned post-investment burn of R$210,000 per month gives 17 months of runway. An operating milestone is targeted for month nine, fundraising starts in month ten and closing is expected in month fifteen. At that point, R$420,000 remains, equivalent to two months of burn. The expected closing is an assumption, not another investor's commitment.
Diligence also takes time. If funding arrives one and a half months later while pre-investment burn stays at R$180,000, Aurora spends R$270,000 and unrestricted cash falls to R$450,000. With the same net proceeds and subsequent burn, runway after closing is approximately 15.7 months; the earlier 17-month estimate no longer applies. In a separate downside scenario with immediate funding but R$280,000 monthly burn, R$3.57 million lasts 12.75 months, bringing financing needs forward and reducing time to correct execution problems.
Evidence: build a trail that allows the numbers to be challenged
Reliable diligence lets a reviewer retrace the path from the presentation to source documents. For selected revenue, the trail may include the contract, order, delivery evidence, billing, collection and any refund. For material expenses, examine the supplier, approval, service description, proof of work and payment. Record which question each document answers. An internal report does not become independent evidence because someone exports it as a PDF; a bank statement proves movement of money without necessarily explaining its economic cause.
Sampling should combine materiality and risk. Reviewing only the largest customers may miss numerous small payments to one beneficiary. Selecting only transactions proposed by a founder can produce a convenient picture. Include large transactions, related counterparties, movements near reporting cutoffs and unusual items, plus a portion selected without management direction. Authorized external confirmations should use contacts whose origin can be verified. Responses passed on exclusively by an interested party provide less assurance than appropriately obtained independent responses.
Separate observed documents, interpretation, management's explanation and unresolved work. 'The contract permits returns; management says none are expected; distributor confirmation is outstanding' is more precise than 'false revenue' or 'validated revenue.' Assign a reviewer and record limitations for important conclusions. Keep private data in an environment with controlled access and defined retention. Demonstrating an inconsistency rarely requires copying an entire customer database: sampling, field redaction and restricted access can support many checks.
Warning signs: combine indicators without treating suspicion as a verdict
Signs warranting prompt investigation include resistance to bank reconciliation, incompatible cap tables, suppliers without identifiable deliverables, large unsupported reimbursements and revenue concentrated near period-end. Questionable signatures or signing powers, balances that disappear when confirmation is requested and frequent changes to metric definitions also matter. Each sign admits different explanations. A system migration may produce legitimate discrepancies, but investors still need to test that explanation and determine how much of the problem it accounts for.
Concern grows when independent sources converge. A founder-connected supplier with an unclear scope, out-of-budget payments and charges inconsistent with delivery presents a stronger concern than any one feature alone. Urgency increases if money may continue leaving the business or operations rely on potentially altered documents. A fund may suspend its own disbursement while clarifying the facts. Any action inside the company must respect existing authority, the fund's rules and evidence-preservation requirements.
Define responses before enthusiasm for the opportunity clouds judgment. A limited, explainable discrepancy may require correction and a larger sample. A missing essential document prevents completion of that diligence point. Material deliberate omission may justify rejection even when the amount seems small, because it undermines representations used elsewhere. Conversely, accusing someone solely because of a statistical pattern or family relationship causes harm and weakens the investigation. State what is known, what supports it and what remains uncertain.
Authority, ownership and instruments: verify what the fund would acquire
A functioning product does not establish that the company holds the rights needed to exploit it. Contractor-developed code, a trademark held personally, contracts signed by another entity and databases shared without an appropriate basis can affect the financed asset. Verify relevant ownership and usage rights and the authority of those signing the round. Across related companies, locate customers, staff, intellectual property, debts and collections. The investment must reach the business described in the thesis, rather than an entity missing essential assets.
Brazilian legal structure also limits the alternatives. An MEI cannot admit equity partners while remaining an MEI; an equity investment needs a compatible structure before entry. A convertible still requires review of obligations, signing authority, conversion conditions and effects on other investors. Loans, receivables, equity, convertibles and donations create different rights and risks. Existing eligible receivables differ from hoped-for sales. A donation to a nonprofit is neither an equity investment nor a promise of financial return.
Public securities distribution and regulated financial activities require appropriate structures and providers for the case. Technology, a standardized contract or a CNAE business-activity classification is not permission. Funds must also follow their governing documents, mandate and applicable responsibilities. Resolve these questions before startup cash pressure dominates negotiations. Institutional sources are reference entry points for assessing the structure, not proof of authorization for a particular transaction or platform. Material uncertainty about an instrument requires specialist assessment before signing.
Negotiate related-party arrangements without bringing operations to a halt
Banning every related-party transaction may be impractical and destroy value. Suitable premises may belong to a shareholder, or a group company may provide an integration that is difficult to replace. Make the conflict visible and govern the decision. Contracts should specify the scope, price, term, exit conditions and each side's obligations. Independent alternatives provide useful comparisons, but consider quality, continuity and migration costs. The lowest quoted price may conceal operational dependence or a transfer of responsibility.
A proportionate policy distinguishes ordinary transactions, exceptions and material changes. It may require approval by people without an interest in the deal, abstention by the beneficiary and a documented rationale. If the company is too small for sufficient internal independence, a focused external review may be more useful than a merely formal committee. Cover split transactions and amendments so that dividing a contract does not bypass scrutiny. Internal limits should reflect budget, risk and monitoring capacity, without presenting them as universal legal requirements.
For Aurora, the fund might temporarily accept infrastructure supplied by a founder's company if projections include its real cost, the scope is evidenced and a viable continuity plan exists. Demanding immediate replacement without considering product interruption could create a different loss. Indefinite dependence also needs scrutiny. If the supplier refuses to document pricing or provide necessary information, that behavior changes the risk assessment and may prevent investment, regardless of the savings claimed.
Operational responsibilities: controls that produce usable information
Founders remain responsible for management and disclosure; the fund does not take over daily operations. Nevertheless, the agreement should identify who prepares cash reports, reconciles banks, approves payments and reviews exceptions. Overlapping roles may be unavoidable, but need compensating checks. If one person creates suppliers, authorizes payments and executes them, someone else should promptly review documents and movements. A review arriving months later may be too late to prevent repetition or recover records. Controls must fit actual working conditions.
A monitoring package can cover unrestricted cash, obligations, independent recurring revenue, overdue customer payments, concentration, related-party transactions and cap-table changes. Keep definitions stable and reconcile changes. Aurora's reports should track the initial R$55,000 of independent recurring revenue separately from one-off services and conditional sales. Follow the same customer cohorts and associated service costs over time. Comparing figures calculated under different definitions can suggest improvement without establishing any underlying economic change.
Governance also involves access, continuity and cost. A replacement should be able to perform finance duties without relying on personal passwords or inaccessible files. Grant access by role and revoke it when no longer needed. Accounting, independent review and advice consume cash; prioritize material risks over accumulating unused reports. Negotiate information rights, monitoring and any representation on corporate bodies in line with the company's stage and the investor's capacity to exercise them. Rights written into a contract have limited value if nobody uses them.
Disagreements, errors and corrections: assess the response to evidence
The response to a discrepancy can be as informative as the original figure. Suppose Aurora called all R$100,000 recurring revenue because its dashboard summed every invoice. A proper correction identifies affected periods, separates categories, rebuilds comparable metrics and explains the effects on margin and cash. Assign responsibility and review to prevent recurrence. Renaming a spreadsheet column while leaving the committee's presentation uncorrected preserves the error in the decision, even if the internal file is now accurate.
Founders and investors can disagree in good faith. Management may consider related contracts genuine customers, while the fund finds them insufficient evidence of independent demand. That disagreement need not become an accusation or force everyone into a single metric. Keep documented revenue visible while presenting a growth scenario that excludes it. A disagreement about whether a payment occurred is factual and requires evidence. The memorandum should distinguish interpretive differences from verifiable contradictions and describe both clearly.
If evidence suggests deliberate deception, prioritize document preservation, limit circulation of allegations and obtain suitable advice on investigation and applicable obligations. Do not alter source files to produce an apparently clean record. Corrections must remain traceable and reach recipients of materially inaccurate information. Once a fund has invested, its response depends on its contractual rights and powers. Pressure to recover capital does not authorize improper access, disclosure of private data or movement of company funds without authority.
Follow-on rounds and exits: turn paper values into possible outcomes
Return only to the initial scenario: the fund invests R$3 million for 25% of Aurora. Set aside the lower-valuation alternative. Assume a later R$4.5 million round at R$13.5 million pre-money, giving the new investor 25% of the R$18 million post-money value. Without pro rata participation, new options or other adjustments, the first fund falls to 18.75%. Founders fall from 60% to 45%; the angel and incentive reserve each fall from 7.5% to 5.625%. Including the new investor, ownership still totals 100%.
In a hypothetical R$24 million sale, subtract R$1 million in costs and R$2 million in debt repayment. R$21 million remains for equity holders before any taxes or other adjustments. With proportional distribution, the first fund's 18.75% produces R$3.9375 million. Relative to its R$3 million investment, that is 1.3125 times invested capital, before considering elapsed time, fund expenses or taxation. The previous round's R$18 million valuation was not cash received by that investor; the outcome required an actual exit.
Contract terms can change the downside dramatically. A sale for R$8 million with the same R$3 million of costs and debt leaves R$5 million. Suppose the later investor has a one-times, nonparticipating liquidation preference senior to the first fund: its R$4.5 million consumes almost all the balance. If the first fund also has a one-times preference, it receives only the remaining R$500,000, leaving nothing for common equity. These are specific hypothetical terms. Read and model the distribution order, including priorities benefiting other investors.
Technology-based intermediation and documentation: support analysis without manufacturing trust
Technology's usefulness in connecting entrepreneurial projects with potential funders depends on what it makes clear. Project documentation can distinguish observed facts, projections, the proposed instrument, intended use of funds and disclosed conflicts. KEZS connects projects and potential funders through technology and intermediation; investors conduct their own diligence and make their own decisions. KEZS does not assign a credit score. A project's presence on a platform does not establish governance approval, confirmation of every statement or an indication of returns.
A project-based method is useful when claims remain connected to their support: which revenue supports expansion, which capacity would be financed and which event might provide liquidity. This enables specific questions beyond comparing presentations. Such potential usefulness does not require claims about unverified operating features. Investors need to know what documents exist, who prepared them, which access was authorized and what remains unchecked. A field completed by a project proponent remains a declaration until evidence supports the intended conclusion.
Recordkeeping technology has limits. A blockchain record may help establish that content has not changed since recording; it does not establish truth, legal title, reserve solvency or future payment. Private documents should not be placed on public blockchains. Access controls, data minimization and proportionate retention matter more than making every detail permanent. Regulatory references likewise guide inquiry without proving platform authorization or investment suitability. Neither technology nor documentation guarantees approval, repayment, profitability, liquidity or preservation of capital; investors can lose their investment.
Conclude with verifiable conditions and a willingness to decline
A useful committee conclusion answers difficult questions with sufficient evidence. Does independent revenue support the thesis? Do costs include material dependencies? Will capital enter the entity controlling the necessary assets? Do related transactions have substance, defensible prices and appropriate approval? Have ownership and distributions been modeled? For Aurora, abandon the undifferentiated R$100,000 reference. Explain how R$55,000 of independent recurring revenue, a 60% margin within that scope and unrestricted cash change the exposure the fund would accept.
The decision may be to invest, invest after objective conditions, await more evidence or decline. Useful conditions identify an owner, proof of completion and consequences for nonperformance. Examples include correcting financial presentations, formalizing ownership of an essential asset, resolving an unrecorded obligation or approving a related contract with documented reasons. Staged disbursements can reduce investor exposure while creating financing risk for the startup. State clearly what is committed, what is conditional and what remains an expectation, including any hoped-for follow-on financing.
After closing, apply the same discipline: compare promises with delivery, update relationships, investigate deviations and recognize losses when the thesis fails. A good relationship does not eliminate conflicts, and a lengthy contract cannot prevent every fraud. Make misconduct harder, more detectable and more manageable at a cost proportionate to the business. Seed investors accept uncertainty about markets and execution; they need not accept opacity about existing facts. Declining an opportunity is a legitimate control when the exposure cannot be understood.
Frequently asked questions.
Should a sale to a founder's company be disregarded?
Not automatically. It may involve real delivery, a valid obligation and actual payment. Check its substance, terms and any circular funding, and report it separately. Even a legitimate related-party sale provides limited evidence of independent demand. For a growth thesis, test whether pricing, usage and retention would hold for unrelated customers. Accounting treatment and commercial relevance are connected but distinct questions.
Is a bank receipt enough to validate revenue?
It establishes a cash inflow, not the whole transaction. The money could be an advance, loan, refundable payment, third-party payment or funds circulated through a related party. Link the contract, delivery, payment obligation and source of funds within authorized access. Also separate recurring revenue from one-off services. Real cash can improve liquidity without demonstrating a repeatable product or independent customer demand.
When should a discrepancy interrupt negotiations?
Do not proceed as though an essential issue is resolved when available cash, product ownership, signing authority or capitalization remains unclear. Disbursement can await evidence and correction. Material deliberate omissions, unexplained contradictory documents and persistent resistance to proportionate checks may justify rejection. Record the facts, missing information and limitations. An unresolved suspicion should not be presented as a proven accusation.
Can a lower valuation compensate for fraud risk?
A lower price can reflect weaker economics or less revenue than originally presented. It does not reveal hidden liabilities or automatically restore trust in management's statements. If the information used to calculate the discount is unreliable, the protection may be illusory. First determine whether the problem can be bounded, the correction verified and the company monitored. Some opportunities remain unsuitable even at a large ownership percentage.
Should every seed startup undergo a full audit?
Depth should reflect materiality, complexity, the instrument, warning signs and applicable requirements. A targeted review may answer specific questions at lower cost; complex relationships or widespread discrepancies may require broader work. Define scope and limitations: an accounting review, balance confirmation and fraud investigation answer different questions. No report eliminates every risk, and a lack of findings outside its scope is not validation.
Which controls help most when the finance team is small?
A workable combination includes regular bank reconciliation, documented expenses, another person's review of exceptions and monitoring of unrestricted cash and obligations. Identify related transactions and, where possible, have someone without an economic interest review them. If complete segregation of duties is impractical, periodic independent review can offset part of the risk. Specify frequency, responsibility and how discrepancies will be investigated and resolved.
How should an exit be assessed when liquidation preferences apply?
Start with proceeds actually available after debt, costs and relevant adjustments. Then model contractual seniority, preference multiples, any additional participation and conversion choices. Owning 18.75% does not guarantee receipt of 18.75% of the announced sale price. In lower-value exits, later-round preferences may consume nearly all distributable proceeds. Include favorable and adverse scenarios, and keep valuation distinct from cash realizable through an exit.
Can a platform or technological record replace investor diligence?
No. Organizing projects and documents may improve comparison, questioning and traceability, but does not establish truth, regulatory authorization or ability to pay. KEZS investors conduct their own diligence and decide; KEZS does not assign them a credit score for the project. Technological records do not guarantee ownership or returns. Examine information sources, completed checks and outstanding issues, protect private data and obtain specialist assessment where needed.
References and context
Institutional reference points for checking applicable rules and guidance. Calculations and scenarios are original examples, not quotes, historical results or individualized recommendations.
- Banco Central — correspondentes no País ↗
- CVM — Resolução 88, texto e alterações ↗
- CVM — Resolução 175, fundos de investimento ↗
- Portal do Empreendedor — perguntas sobre MEI ↗
- ANPD — orientações e perguntas frequentes ↗
AI-assisted educational content. Decisions require verification of the information, instruments and providers in each actual case.
