Investment Thesis and Selection Criteria for Financial Institutions

Start with the exposure the institution is prepared to hold

The thesis might begin with a specific hypothesis: businesses in a particular sector face a recurring gap between supplier payments and customer receipts, and a project-oriented channel can document that gap effectively. This does not establish financeable demand, repayment capacity, or competitive advantage. Product demand, verifiable documentation, conversion into contracts, and subsequent portfolio performance each require evidence. A convincing explanation of the problem is the beginning of an investigation.

The target market must also state what is excluded. An institution might reject businesses whose controlling owners cannot be adequately identified, projects dependent entirely on another funding round for repayment, or transactions beyond its monitoring capacity. These boundaries maintain consistency. Without them, the thesis can become a generic description of any business seeking money, making pricing, comparison, and accountability harder.

The resulting document should connect four choices: the business problem being addressed, the economic right being acquired, the risks being accepted, and the process supporting selection. A technology channel should be evaluated within that chain. A clear project presentation can improve understanding, but the party committing funds must make an independent decision and determine whether the exposure fits the institution's mandate.

Translate strategy into policy, limits, and approval authority

Policy turns the thesis into repeatable decisions. It defines eligible businesses, required documents, permitted products, maturities, exposure limits, and rejection grounds. It must also address economic groups, related parties, and concentration. Contracts with different companies can carry substantially the same risk when they depend on one buyer, controlling owner, or commercial event. Counting separate Brazilian business registrations can therefore create an appearance of diversification without independent repayment sources.

Approval arrangements should separate application preparation, analysis, approval, and exception handling. Commercial staff may gather information and clarify the applicant's needs, but should not remove a disqualifying condition to meet a target. Whoever approves an exception needs to understand its basis, economic effect, and monitoring responsibility. The explanation must allow another qualified person to reconstruct the decision, including the evidence considered and the limit being overridden.

Distinguish missing documentation, acceptable risk, and policy incompatibility. A missing document may justify a request for additional information. Excessive concentration may require a smaller amount. An activity outside the institution's scope may end the review. Grouping all three under a vague pending status leaves applications circulating and consumes resources. It also invites commercial negotiation over requirements that should be mandatory or over problems that need a clear decision.

Investors can assess this governance through the frequency and performance of exceptions. If exceptional transactions become the majority, the formal policy no longer describes the actual portfolio. The response may require changing the thesis or suspending a channel, rather than adding signatures. Having an independent policy means retaining practical authority to reject business, including when a channel supplies substantial volume or a relationship has strategic value.

Select products according to rights and repayment sources

Products serving the same businesses can create very different exposures. A loan involves the company's repayment obligation and contractual terms. Purchasing or advancing funds against receivables requires examining the credit's existence, eligibility, documentation, transferred rights, and events that could reduce collections. Expected sales are not existing receivables, however plausible the forecast. This distinction changes both the evidence required and the source from which repayment is expected.

Equity interests and convertible instruments introduce governance rights, dilution, conversion events, and uncertain exit opportunities. They require analysis appropriate to those rights. Under Brazil's MEI framework, an individual microentrepreneur cannot admit equity partners while remaining a MEI. If a proposal requires equity participation, the business structure must be compatible before contracting. A matchmaking platform cannot resolve the incompatibility simply by describing the arrangement differently.

Donations to nonprofit organizations must also be distinguished from investments seeking financial returns. Donations may involve objectives, restrictions on use, and reporting obligations, but do not automatically create equity ownership or entitlement to remuneration. Combining these products in one commercial process without explaining their differences increases the risk of unsuitable arrangements, incorrect documentation, and expectations unsupported by the contract. Similar project purposes do not create identical legal rights.

Selection must begin with activities the institution is permitted to undertake and structures appropriate to each transaction. Public securities distribution and other regulated financial activities require suitable arrangements and providers. Technology or a Brazilian CNAE activity classification does not confer permission. Banco Central do Brasil and CVM materials are entry points for specialist analysis; they do not establish authorization for KEZS, an origination channel, or a particular product.

Evaluate a channel by the information it improves

An origination channel deserves attention when it improves something relevant: access to eligible businesses, understanding of their needs, document completeness, analysis time, or cost per contracted transaction. Registration volume alone is insufficient. A large channel may produce little business compatible with policy. A smaller one may deliver more verifiable applications and reduce rework at critical stages. The comparison should follow usable information through to its operational and economic consequences.

KEZS connects entrepreneurial projects and potential funders through technology and intermediation. A project-oriented method can help organize the intended use of funds, execution stages, documents, and economic assumptions. Its usefulness must be assessed in context, without assuming specific integrations or operational features. Investors conduct their own diligence and make their own decisions. KEZS does not assign a credit score, and organized information does not itself establish creditworthiness.

Compare similar applications and record business size, sector, product, requested amount, and documentation stage. If one channel serves larger companies while another receives incomplete submissions from smaller businesses, approval or loss comparisons alone can mislead. Outcomes may reflect the applicant mix rather than channel quality. Reviewers should test alternative explanations before attributing better results to the channel, including differences in product terms and portfolio maturity.

Measure the burden imposed on applicants as well. Repeated document requests, contradictory forms, and unclear responses can cause withdrawal and adverse selection. A transparent record of outstanding requirements can improve the experience and traceability without promising approval. A useful assessment combines information quality, applicant effort, and observed economic results while recognizing that a properly evaluated application may legitimately be rejected.

Banking correspondent arrangements require defined scope and supervision

Acting as a banking correspondent in Brazil depends on an appropriate contractual relationship and scope with the institution, together with applicable requirements. A commercial connection between businesses does not establish that status. Before treating activities as part of a correspondent arrangement, determine which services are covered, who may perform them, what information they may use, and how the contracting institution will supervise execution. The relationship's name cannot substitute for its substance.

The funder retains the responsibilities assigned to it by regulation. Outsourcing information collection, customer service, or application forwarding does not automatically transfer selection, control, and supervision duties. Contracts therefore need working procedures: training, role-based access, review of customer materials, complaint handling, and ways to stop out-of-scope activity. A broad contractual clause is weak evidence unless the institution can show how the controls operate.

Communication should let the business owner identify who presents the proposal, conducts the analysis, enters into the contract, and makes the decision. A simulation must not appear to be final approval. A document merely received or stored must not be described as verified. These distinctions affect real commitments: an entrepreneur might promise payment to a supplier before a financing decision exists, creating costs that clearer communication could have avoided.

Partner selection should examine the ability to follow instructions and record deviations, alongside commercial reach. Investors can review service samples, correction histories, and supervision evidence. Banco Central guidance on correspondents provides a starting point for understanding the subject. The applicable arrangement still requires its own assessment; neither technology nor association with a familiar brand proves that a channel is authorized to perform particular activities.

Establish KYC and financial crime controls before conversion pressure builds

Knowing the customer means understanding who seeks financing, who controls the company, what it does, and whether the proposed transaction makes sense in that context. Checking that a business registration exists is only one step. Recent ownership changes, representatives with unclear authority, inconsistent information, and structures without an evident economic purpose require proportionate review, assigned responsibility, and defined escalation routes.

Know-your-customer procedures and financial crime controls should combine sources, documents, and contextual analysis. No single database check eliminates fraud, money laundering, or identity misuse. The process must specify how discrepancies are handled: pause the application, request clarification, verify through an independent channel, or escalate to the appropriate function. Commercial urgency should not turn an unresolved concern into a harmless-looking note while processing continues.

Consider a request shortly before contracting to release funds into an unrelated third party's account. There may be a legitimate explanation, but it must be documented and assessed under the institution's rules. Confirming the change only with the contact who requested it provides no independent verification. Review should consider account ownership, representative authority, purpose, and applicable procedures before any recipient change is accepted.

Data used for these controls needs a defined purpose, access rules, and retention arrangements consistent with applicable Brazilian data protection guidance. Personal documents and sensitive business information should not circulate indiscriminately among potential funders. More sharing does not necessarily mean better analysis. Effective controls combine detection, confidentiality, and a record explaining why a decision proceeded, required escalation, or was stopped.

Turn project documentation into verifiable evidence

A well-organized project separates documented facts, applicant statements, and execution assumptions. A submitted contract supports a claimed relationship, subject to checks on authenticity, validity, and terms. A revenue projection remains a hypothesis. A supplier quotation indicates proposed pricing and scope, potentially subject to expiry, availability, or adjustment. Treating every attachment as certainty produces an analysis that looks precise while resting on fragile economic assumptions.

For each decisive assumption, record the question, available evidence, limitations, and consequence for the decision. If repayment depends on a major buyer, establish what was contracted, what has been delivered, which conditions permit disputes, and when payment is expected. Documents should bring the analysis closer to the actual cash mechanism. Increasing the number of attachments without answering these questions adds review cost without necessarily reducing uncertainty.

Versions and responsibilities matter. Replacing a quotation after approval may change the funding requirement; changing the buyer may alter risk. The record should identify the version supporting the decision and changes requiring reassessment. Corrections must preserve history and identify what changed. Silently replacing an inconsistent document can erase the reason for an earlier decision and prevent reviewers from understanding whether the correction affects the wider application.

Digital records can help demonstrate file integrity or event sequence. Even blockchain integrity does not establish content truth, legal title, reserve solvency, or guaranteed payment. Private documents should not be placed on public blockchains. The objective is proportionate, controlled verification. A recording technology is useful when it solves a concrete evidentiary problem, with its costs and limitations recognized, rather than becoming a substitute for checking the underlying transaction.

Build an economic example with explicit boundaries

Consider a wholly hypothetical batch of business loans. The channel receives 120 expressions of interest; 80 reach minimum documentation, 40 pass eligibility screening, 20 receive approval, and 10 are contracted and disbursed. Each has BRL 100,000 of principal, for BRL 1,000,000 in total. These numbers illustrate arithmetic only. They are not market rates, observed performance, or an institution's offer.

The batch matures in 90 days with one scheduled payment of BRL 1,045,000: BRL 1,000,000 of principal and BRL 45,000 of contractual income. Assume BRL 24,000 in funding costs for the period, BRL 8,000 in operating expenses allocated to the batch, and BRL 5,000 in channel remuneration. Operating expenses include work on this group's unconverted applications, so discarded commercial effort is not treated as free.

Assume an expected loss of BRL 6,000 against the total contractually receivable, after estimated recoveries. Expected economic profit is BRL 45,000 minus BRL 24,000, minus BRL 8,000, minus BRL 5,000, minus BRL 6,000: BRL 2,000. This is before taxes, capital costs, and other unmodeled items. Expected loss is an estimate of reduced collections, not necessarily a separate cash payment that will occur.

To isolate the mechanism, assume funding covers all loan principal and its remuneration is payable at maturity. This does not describe capital requirements or recommend a funding structure. The limited conclusion is that BRL 45,000 of gross income leaves a thin margin under these assumptions. Investors should identify missing expenses, test allocation methods, and ask how little the assumptions would need to change to eliminate that balance.

Read the funnel and negotiate incentives that support useful origination

In the example, only 10 of 120 expressions of interest become contracts, approximately 8.3% overall. Half of the 20 approvals convert. That second figure needs investigation: unattractive terms, slow responses, project withdrawal, or simultaneous applications elsewhere could explain the gap. Approval is not revenue. Blaming every withdrawal on commercial staff may conceal a product problem or a response time that does not match the business's needs.

Hypothetical channel remuneration of BRL 5,000 equals BRL 500 per disbursed contract and 0.5% of batch principal. These ratios support comparison but do not establish an appropriate price. If remuneration rises to BRL 7,000 with all other assumptions unchanged, expected profit falls from BRL 2,000 to zero before omitted items. A relatively small change in commercial terms can consume the entire modeled margin.

The institution and channel may disagree over who bears the cost of incomplete applications. Prospecting and guiding businesses require work, while the institution faces uncertain conversion. Negotiations can distinguish verified activities, documentation quality, and completed contracting while examining conflicts. Paying only for submissions encourages excess volume. Paying only for approvals can encourage pressure on analysts or concealment of unfavorable information. Each incentive needs a corresponding control.

A verifiable agreement defines the event triggering payment, calculation method, cancellation treatment, and dispute process. Here, the BRL 5,000 is attributed to actual disbursement of ten loans. Any additional payment component belongs in the economics. Investors should examine the behavior the arrangement produces: reliable documents and applicants suited to policy can be more valuable than a large funnel sustained by repeated correction and unnecessary review.

Follow cash movements and liquidity requirements

Cash analysis complements expected profit. At the start, the institution receives BRL 1,000,000 of funding and disburses the same amount to borrowers. Operating expenses of BRL 8,000 and channel remuneration of BRL 5,000 require separate liquidity as they become payable. For simplicity, assume all BRL 13,000 is paid initially. Those payments remain relevant even though loan collections occur later; funding the principal does not also cover these expenses.

At maturity, full payment produces BRL 1,045,000 of receipts against BRL 1,024,000 owed for funding principal and remuneration. BRL 21,000 remains at that point. Subtracting the BRL 13,000 paid initially gives realized profit of BRL 8,000 for the modeled items. This exceeds the BRL 2,000 expectation because the estimated BRL 6,000 loss did not materialize in this particular outcome.

An aggregate representation of the expected outcome instead uses receipts of BRL 1,039,000 after the estimated BRL 6,000 reduction. Deducting BRL 1,024,000 for funding and BRL 13,000 of expenses leaves BRL 2,000. This represents an expectation, not a prediction of an exact receipt. Once the assumed loss has reduced collections, subtracting it again would double-count its effect and misstate the economics.

Now suppose some borrowers pay after the funding obligation matures. Even if all amounts are eventually recovered, the institution needs interim cash and may incur additional costs. It must map maturities, available resources, and contingencies instead of treating contractual income as available money. Estimated profit, capital to absorb losses, and liquidity to meet obligations are related dimensions, but each requires separate measurement and its own assessment.

Stress losses, delays, and concentrations

Keeping the same structure, assume a realized loss of BRL 20,000 against total receivables of BRL 1,045,000. Collections fall to BRL 1,025,000. After BRL 1,024,000 of funding obligations, BRL 1,000 remains at maturity. Including the BRL 13,000 paid initially produces a loss of BRL 12,000. The realized BRL 20,000 replaces the expected BRL 6,000 in this scenario; the two are not added together.

The modeled components break even at a loss of BRL 8,000: contractual income of BRL 45,000 covers BRL 37,000 of funding, operating, and channel costs, leaving BRL 8,000 for losses. This is not a safe limit because taxes, capital, and other costs are omitted. It shows that only BRL 2,000 separates expected loss from modeled break-even, leaving little room for assumption errors or additional expenses.

Separately, assume BRL 100,000 is delayed at maturity while all other scheduled amounts are received in full. Cash receipts then equal BRL 945,000 against funding obligations of BRL 1,024,000, creating a BRL 79,000 shortfall. The initial BRL 13,000 already required separate cash. Later recovery of BRL 100,000 may prevent a principal loss, but does not remove the interim mismatch or its carrying cost. Delay and loss need distinct scenarios.

Risks can also materialize together. Ten businesses relying on the same buyer do not provide ten independent repayment sources. A problem at that buyer could trigger simultaneous delays, renegotiations, and collection costs. Stress analysis should examine common shocks and limits by economic risk factor. Small portfolios and short observation periods can produce widely dispersed results, so a favorable batch provides little basis for assuming stable future performance.

Make settlement and reconciliation part of the thesis

Before contracting, establish who receives funds, who pays, which accounts are used, and which obligation each movement satisfies. Settlement fulfills the payment obligation; reconciliation compares contracts, instructions, records, and actual movements. A customer-supplied payment receipt does not replace appropriate confirmation that funds arrived. The design should identify responsible staff, verification sources, and procedures for investigating differences before they enter portfolio results.

Each transaction needs an identifier connecting proposal, approval, contract, disbursement, and receipt. The ten hypothetical contracts must total BRL 1,000,000 of principal and BRL 45,000 of contractual income. If a collection file shows BRL 1,044,000, the BRL 1,000 difference needs explanation before classification as a discount, loss, or error. Unsupported adjustments distort both financial results and the apparent payment behavior of the portfolio.

Partial, duplicate, and unidentified payments need treatment rules. An unidentified amount should not be assigned wherever it makes an indicator look better. A procedure may hold it temporarily, investigate, and link it to an obligation once evidence is available. Reversals and changes also require controls recording who authorized them, the supporting information, and whether they changed the balance communicated to the customer.

Failures here can consume the example's thin margin without any deterioration in credit. A duplicate BRL 5,000 commission would turn expected profit of BRL 2,000 into a BRL 3,000 loss, with other assumptions unchanged. Detecting and recovering the duplication requires an operational trail and reconciliation. Accurate settlement and verification therefore belong among the selection criteria for both the channel and other providers.

Allocate responsibilities and correct errors with a traceable record

Responsibility starts with the information's source. Applicants are responsible for their statements and documents; the channel performs contracted activities; the institution analyzes and decides under its policy and duties. Legal, risk, compliance, and operations teams participate according to the product and circumstances. Coordination should prevent unassigned tasks and assumptions that another team checked something. Several people viewing a document does not demonstrate that any one of them verified the decisive point.

Suppose a loan term is entered as 90 days, while documented business cash flow shows receipts only after that interval. The error changes delay risk. Correction requires pausing progress, confirming the information, recalculating the transaction, and seeking renewed approval where necessary. Changing the database term after signature, without examining the consequences, does not resolve a contractual inconsistency or establish that the revised exposure remains acceptable.

Another example is an already-paid receivable included in a submitted list. The team should correct the data, check other potentially affected items, and investigate the cause. It could reflect outdated records, a defective process, or intentional conduct, each requiring different treatment. Removing the item alone does not establish that the remaining population is accurate, especially when every item came through the same preparation process.

Error indicators should track occurrence, impact, correction time, and recurrence by cause. Punishing every reported mistake encourages concealment; ignoring repeated failures undermines control. Investors should look for learning that preserves evidence. The institution needs to explain whether the response required training, a form change, access review, channel suspension, or a deeper revision of the thesis, and whether the intervention addressed the underlying cause.

Define the evidence needed to continue or expand

An initial evaluation should state in advance what it aims to learn and what findings justify continuation. It might test whether the channel supplies adequate documents, the institution can analyze applications within the necessary time, and contracting preserves the expected margin. These objectives require different measures. A short review can reveal intake performance but cannot establish final losses on transactions that have not yet matured.

Group transactions by origination period and follow their maturation. A newly formed portfolio can look healthy before payments become due. Combining it with older transactions can artificially dilute arrears indicators. Distinguish contracted, disbursed, due, received, and recovered amounts. State every indicator's denominator so another reviewer can reproduce the calculation and understand whether a change reflects performance, portfolio growth, or a different population.

The example's ten loans illustrate calculations but cannot establish a stable performance pattern. The institution may learn about stage-by-stage costs and documentation failures while remaining highly uncertain about losses. Any expansion should use limits consistent with that uncertainty and operating capacity. An initially positive result does not remove the possibility that the next group has a different borrower mix, risk concentration, or collection experience.

Investor reporting should distinguish hypotheses, observations, and interpretations. Saying a channel reduced rework requires a defined comparison and evidence; saying that reduction could improve margins is an inference to test. Changes in policy, applicant mix, or collection practices must be disclosed because they may explain results. Clear gaps support better decisions than an uninterrupted success narrative built from a small number of observations.

Agree how conflicting interests will be handled

Conflicts are predictable. The channel may want broader eligibility, the institution may require more documents, and the entrepreneur may need a quick response. Contracts and governance should address these tensions without promising approval. Negotiation needs to cover response times, minimum information quality, outstanding-requirement communication, and activity boundaries as well as price. Obligations unsupported by practical execution arrangements are likely to become disputes.

If the channel requests exclusivity, assess genuinely eligible volume, commercial dependence, and the cost of losing other sources. If the institution requests broad document access, define the purpose and limits of sharing for the channel and applicant. Audit rights should support service verification while protecting information. Confidentiality should not obstruct legitimate supervision, and an audit clause should not become permission for indiscriminate access.

Negotiate deterioration procedures before the first problem. Who identifies arrears, contacts the borrower, proposes terms, and approves renegotiation? A channel should not assume collection powers or authority to change contracts unless assigned those responsibilities appropriately. The institution must assess the economics and consequences of renegotiation and preserve records distinguishing actual recovery from an extension that merely postpones recognition of financial difficulty.

Exit arrangements also need to work. The institution should be able to stop new origination, maintain necessary service for existing transactions, and retain documents under applicable obligations. A thesis is weaker if replacing a partner would destroy access to its history. Investors should examine termination and continuity arrangements to understand how much projected value rests on durable processes and how much depends on a fragile relationship.

End selection with a conditional, verifiable decision

The final decision should bring together strategic fit, permitted product structure, channel quality, transaction economics, and execution capacity, with a separate conclusion for each. An opportunity can be commercially attractive yet unready because of documentation or reconciliation weaknesses. The appropriate response may be rejection, redesign, or proceeding within verifiable limits and conditions. Vague approval merely transfers unresolved problems into contracting.

A useful decision memorandum states the eligible population, policy-based maximum exposure, responsibilities, required evidence, and stopping conditions. It also identifies assumptions still awaiting observation. The example's expected BRL 2,000 margin before omitted components warrants questions about full costs, loss sensitivity, and liquidity requirements. It cannot independently establish attractiveness or safety. A conditional decision should make clear which unanswered questions must be resolved before exposure increases.

After the decision, records should support comparison between expectations and outcomes. Higher operating costs, changed commissions, or applicants drifting outside the target population require reassessment. Review should not wait exclusively for defaults. Repeated complaints, identification failures, settlement differences, and growing exceptions may justify intervention before a financial loss appears. Monitoring must connect these signals to someone authorized to change or stop the process.

For investors, the outcome is a clearer understanding of the mechanism they may finance and the conditions under which it can fail. Organizing projects and documents can improve discussion between businesses and funders, but guarantees neither approval, repayment, profitability, liquidity, nor preservation of capital. Decisions remain independent, supported by diligence, explicit limits, and willingness to recognize when evidence contradicts the original hypothesis.

Frequently asked questions.

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.