A portfolio containing several businesses can still be concentrated. If they depend on the same sales channel, infrastructure provider or favorable fundraising conditions, much of the capital faces a common cause of loss. For an angel investor, diversifying by sources of risk means investigating why each business might fail and which causes could also affect the others. Counting companies, sectors or cities only begins the analysis.
Diversification starts with what could cause a loss
Businesses still validating their models often face more uncertainty than their available data can resolve. One team may have a working product without evidence that customers will pay. Another may sell successfully but spend more serving each customer than it receives. A third may earn a healthy margin while relying on one founder to sell, deliver and manage everything. Their early stage does not make their dominant weaknesses identical.
The objective is not to eliminate losses or find a combination that ensures profitability. It is to avoid paying several times for essentially the same exposure. Some risks will remain shared: an economic slowdown, difficulty raising further capital and a long wait for liquidity can affect a portfolio that otherwise appears varied. Diversification can change exposure without making private investments readily saleable.
The analysis should produce observable decisions: which dependencies are acceptable, what information is missing, which conditions justify an investment and which events require review. The examples throughout this article are hypothetical exercises in arithmetic and judgment. They indicate neither market prices nor expected returns, and they do not prescribe an allocation suited to anyone's personal assets or financial circumstances.
Separate product risk, business risk and ownership risk
Start by distinguishing problems that require different responses. Product risk concerns whether the solution works and addresses a meaningful need. Commercial risk involves finding buyers, converting interest into payment and sustaining demand. Operational risk concerns delivery, support and growth without declining quality. A persuasive technical demonstration does not establish that customers will buy repeatedly or that the team can serve them economically.
Financial risk includes running out of cash before the company generates enough evidence to continue. Corporate governance risk involves ownership, obligations, conflicts and decision rules. The investment itself brings another question: what right is the investor actually acquiring, and how could it become cash? A company may keep operating while a minority holding has no buyer, makes no distributions and offers no clear exit.
These categories interact. A faulty integration can delay deliveries, increase expenses and undermine a funding round. Separating them nevertheless helps identify the cause and a possible response. Hiring support staff does not solve absent demand. Allowing more time does not resolve a fundamental disagreement between founders. A contractual preference cannot create money when the company has no assets available to satisfy it.
For each project, record a central hypothesis and evidence that could challenge it. 'Repair shops will pay because the system reduces rework' is a hypothesis. Reconciled payments, repeated use and interviews with buyers free to cancel provide evidence. This distinction prevents confidence in a sector, presentation or founder's experience from being treated as proof of every dimension of the investment.
Map dependencies across projects
A simple matrix can place projects in rows and sources of risk in columns: customers, sales channels, essential suppliers, technology, key people, capital needs and liquidity events. Describe each dependency concretely. 'Technology' is too broad; 'the product stops working if a supplier withdraws API access' supports useful questions about alternatives, migration time and cost. Include who owns the response and whether an alternative has actually been tested.
Consider six fictional projects. Alfa sells retail software; Beta serves clinics; Gama coordinates deliveries; Delta develops industrial sensors; Épsilon monitors water consumption in agriculture; Zeta manages repair shops. Their sector labels suggest variety. Yet Alfa, Beta and Gama might obtain almost all their customers through the same commercial partner. A change at that partner could affect three holdings despite differences among their end customers.
Delta and Épsilon might share another dependency: both need an imported component to complete installations and collect payment. Zeta might have an independent sales channel but rely entirely on its technical founder. The map reveals groups of exposures that sector labels conceal. A new project with contracted revenue generated through direct sales could reduce repeated channel exposure while introducing its own delivery and collection risks.
These observations do not need a seemingly scientific score. Descriptions such as 'high dependency,' 'tested substitute' or 'insufficient information' work when accompanied by reasons. Sparse early-stage data rarely support precise correlation estimates. The map helps formulate questions, compare choices and update judgments; it should not imply statistical precision the portfolio cannot support. Unknown dependencies deserve investigation rather than a low-risk label.
Founder experience offers specific protection
Founder experience changes some risks, rarely all of them. Someone familiar with hospital purchasing may avoid commercial mistakes in that setting. This does not establish an ability to develop software, build a team or manage capital. Connect each person's experience to work required for the next milestone, rather than treating professional prestige as a general guarantee of execution or sound judgment.
For a new team, investigate what its members have accomplished together. Impressive individual résumés differ from a history of sharing responsibilities, handling delays and resolving real disagreements. Request verifiable examples: who delivered an integration, negotiated with the first customer, communicated a failure and changed a decision afterward? Professional references help when they describe observed behavior, but they do not replace analysis of the project.
People-related concentration can also span a portfolio. Businesses referred by the same group may share a commercial outlook, supplier network and reluctance to question assumptions. Teams with different backgrounds can broaden the range of approaches, but superficial résumé differences are insufficient. Examine relevant skills, incentives, learning capacity and coverage of essential functions. A varied network is useful only if it changes how dependencies are understood and managed.
During monitoring, avoid allowing an experienced founder to remain indispensable merely because that is convenient. Documentation, appropriately shared access and development of another leader can reduce dependency. If business continuity requires the investor to take over operations whenever trouble arises, the portfolio carries an additional exposure: an obligation to work that may never have been included in the original investment decision or cost estimate.
Understand what early sales actually demonstrate
A first sale shows that someone agreed to pay under particular conditions. It does not, by itself, establish a repeatable market. The buyer may know the founders, receive an exceptional discount or require a bespoke solution. Reconstruct the sale: contact source, stated problem, budget owner, decision period, price, promised delivery and payment received. These details show how much of the transaction another customer might realistically repeat.
Separate signed contracts, invoicing, collection and recurring revenue. An accepted proposal may depend on implementation. An invoice may remain unpaid. An advance payment may fund future obligations rather than represent freely available cash. Use consistent definitions across projects and preserve the sequence of events. Combining these stages can make businesses with very different cash positions appear equivalent and conceal the cost of fulfilling existing commitments.
The distinction also matters for receivables. An existing claim whose origin, ownership and eligibility can be examined differs from sales the founders hope to make. Expected sales remain a commercial hypothesis. Even an existing receivable requires scrutiny of disputes, debtor concentration, payment timing, assignment conditions and any prior pledge or transfer. Its existence does not eliminate late payment, competing claims or loss.
To assess diversification, examine who pays and why. Ten customers exposed to one principal contracting company may represent a single economic dependency. Buyers in different regions may suspend spending for the same reason. Understanding the payment chain and the sensitivity of customer budgets generally reveals more than counting logos in a presentation. The relevant question is which event could interrupt several payment streams together.
A portfolio example with explicit commitments
Consider a hypothetical R$600,000 budget for an angel investment strategy. Of this, R$360,000 funds six initial investments of R$60,000 each; R$180,000 remains reserved for possible follow-on investments; and R$60,000 is budgeted for diligence, documentation and monitoring expenses. The three amounts total R$600,000. The expense budget does not purchase ownership and should not be reported as capital invested in the companies.
This allocation illustrates decisions; it recommends neither a number of projects nor a reserve percentage or investment size. It separates commitments already made, money available for later decisions and the cost of investing. Without that separation, the same funds can mistakenly be counted as a reserve for several companies or advertised as available even though fees and earlier obligations will consume them.
In the fictional portfolio, Alfa, Beta and Gama depend heavily on the same commercial partner. Consequently, R$180,000 of the initial investments is associated with that common source of risk: half of the R$360,000 initially invested. This measures concentration of contributed capital, not a predicted loss. The consequences of disruption would depend on each company's cash, commercial alternatives and ability to adapt.
Now suppose the investor considers adding R$30,000 to Alfa. Total contributions would rise to R$390,000, with R$210,000 invested in the three businesses dependent on that partner. Concentration would increase to approximately 53.8%. The reserve would fall from R$180,000 to R$150,000, while the expense provision remained unchanged. The follow-on investment might be defensible, but its effect on shared exposure must remain visible alongside its company-level rationale.
Cap tables and dilution: track the right acquired
Look more closely at Alfa. Before the initial investment, its founders own the entire company. The investor contributes R$60,000 for newly issued ownership interests in a hypothetical transaction valuing the business at R$540,000 before the new money. The post-investment value is R$600,000. The investor therefore receives 10%, and the founders retain 90%, considering only this simplified ownership structure and the issuance described.
Later, Alfa negotiates a R$300,000 round at a pre-money valuation of R$1.2 million and a post-money valuation of R$1.5 million. The new capital represents 20% of the company after the round. If the initial investor does not participate, the earlier 10% becomes 8%, calculated by multiplying it by 80%. The founders' 90% becomes 72%, and new investors receive 20%. Ownership still totals 100%.
To retain 10%, the initial investor contributes R$30,000, or 10% of the R$300,000 round. The existing holding becomes 8%, while the additional contribution purchases 2%. Other participants contribute R$270,000 and receive 18%. The final structure is 72% for the founders, 10% for the initial investor and 18% for the others. The initial investor has now contributed a total of R$90,000 to Alfa.
This assumes no other issuances, options, convertibles or rights affecting the calculation. An actual cap table must disclose such commitments and when they enter the calculation. Percentage dilution does not automatically mean economic loss, and a higher round valuation does not establish a realizable gain. Preferences, restrictions and illiquidity can substantially change the value of the rights acquired. Compare economic terms as well as ownership percentages.
Equity, convertibles and debt change the exposure
Diversifying instruments is meaningful only when their rights are understood. Equity exposes the investor to the business's residual value under its legal structure and documents. It may carry specific economic and governance rights without a general repurchase obligation or an exit market. A minority position depends on execution and rules allowing oversight of important decisions. A successful product alone does not establish how shareholders will receive money.
A convertible can postpone the ownership calculation without removing the economic negotiation. Examine conversion triggers, discounts, valuation caps, maturity and what happens if a funding round never occurs. Instruments for identical amounts can create different ownership outcomes and conflicts. Model both conversion and non-conversion scenarios alongside existing instruments. A maturity provision must also be understood in light of the company's actual resources and the applicable contractual terms.
Debt creates a repayment obligation that can strain cash. A short term does not turn an uncertain business into safe credit. If repayment depends on fundraising, access to capital remains central despite the contract's form. Guarantees and security require examination of existence, validity, priority and enforceability; mentioning them does not establish full recovery. Equity, loans, receivables and convertibles create different rights, costs and loss scenarios.
In Brazil, a MEI, or individual microentrepreneur, cannot admit equity partners while remaining a MEI. An equity proposal must resolve the business structure before being treated as a completed ownership investment. Other arrangements also require legal suitability. A donation to a nonprofit buys no equity and promises no financial return. Renaming a contract neither makes different rights equivalent nor authorizes a transaction incompatible with its substance.
Cash and milestones: how much time is available to learn?
Return to Alfa using a cash scenario separate from its valuation. Immediately after the first investment, the company holds R$180,000: R$120,000 previously provided by the founders and R$60,000 from the investor. Assume monthly cash expenses of R$40,000 and customer collections of R$10,000. Net cash consumption is R$30,000 a month. With no changes, the available cash covers six full months of operations.
At the end of month three, R$90,000 remains; after month four, R$60,000; after month five, R$30,000; and after month six, zero. This simplified calculation assumes timely receipts and no exceptional expenses. It demonstrates why starting round negotiations only when the balance runs out leaves no room for diligence, contractual adjustments or a delayed closing. An expected investment is not available cash before it arrives.
Suppose the R$300,000 round closes at the end of month four, after that month's operating cash movements. The balance rises from R$60,000 to R$360,000. At unchanged net consumption of R$30,000 monthly, this would support another twelve months from that point. This is arithmetic, not a forecast: hiring, customer defaults or increased costs could shorten the period. A plan using that balance must identify changes in its assumptions.
A milestone by month three might require eight implemented pilots, five paying customers and three renewals among customers already eligible to renew. Define payment, usage and renewal eligibility in advance. The investor can then review evidence while cash remains. New funding and demonstrated demand are separate events: receiving the former does not supply the evidence needed to justify continuing the business or making another investment.
Reserves preserve a choice, not an obligation to rescue
A reserve preserves the ability to act after learning more, but its usefulness depends on discipline. Automatically funding the project asking most urgently for help can concentrate exposure in the weakest business. Choosing only the company announcing the highest valuation is also inadequate. The headline price may conceal cash pressure, preferential economics for new participants or limited evidence that customers will buy and remain.
For Alfa, compare the R$30,000 needed to retain 10% with the rights acquired and other available uses of the funds. A contractual right to join a round provides an option; it does not establish that exercising it is attractive. Consider new evidence, round terms, dependence on the common commercial partner and the remaining reserve. The earlier investment does not determine whether the next contribution is justified.
Write down beforehand what would justify reconsideration: better retention, reduced channel dependency, implementations that no longer require exceptional work or progress in hiring an essential person. These criteria need not become a rigid algorithm. They help limit the temptation to reinterpret every outcome as confirmation of the original decision. A missed criterion should prompt an explanation and review rather than an automatic change in its meaning.
Distinguish an internal reserve from a contractual commitment. Money set aside internally may remain available for redirection under the chosen strategy. A signed obligation to contribute under specified conditions may restrict that freedom. For staged funding, documents should address milestone evidence, review periods and consequences of nonfulfillment. Clear terms prevent an ambiguous expectation of permanent financing and show how much of the reserve remains genuinely available.
Negotiate information rights that support decisions
Information rights form part of the ability to monitor risk. A general access clause can generate disputes over content, frequency and deadlines. For a business validating its model, a concise package may be more useful: cash position, summarized cash movements, material obligations, sales separated from collections, milestone progress, cap table changes and events threatening continuity. Each item should answer a decision-relevant question rather than merely fill a report.
Name a responsible person and agree on a schedule suited to the business's stage. Monthly reporting can be supplemented by exceptional notices of a major customer loss, founder departure, data incident or unexpected obligation. Investors should be able to reconcile essential figures to supporting records without requiring a small team to operate a reporting department. Define the underlying measures consistently so comparisons do not change with each report.
Access must account for confidentiality, purpose and proportionality. Commercial information and personal data should not circulate freely among all investors or their businesses. Controlled folders, permissions based on need and version records help preserve useful documentation. Depending on the circumstances, revenue verification may require only evidence with personal data minimized. Wider circulation can create privacy risks, commercial harm and conflicts unrelated to the original monitoring purpose.
Agree how delays and inconsistencies will be corrected. A late report with an explanation and reconciliation differs from deliberate concealment, although repeated delays weaken confidence in oversight. Information cannot guarantee capital recovery; it can reveal problems earlier and improve decisions. Refusing meaningful visibility while requesting close investor involvement creates a mismatch that should be resolved before investment, including the practical cost of producing reliable information.
Test shared shocks without pretending to precision
A stress test starts with a concrete question: what happens if the commercial partner stops referring new customers to Alfa, Beta and Gama? Do not assume all three immediately fail. Examine cash, existing contracts, direct sales and the cost of replacing the channel. A shared shock can produce different consequences. Understanding those differences is more useful than applying the same arbitrary reduction to every holding.
To examine only initial exposure, consider an extreme scenario in which Alfa and Beta each lose the entire R$60,000 invested. Assume an economic loss of half the Gama investment, or R$30,000. The hypothetical total loss is R$150,000, approximately 41.7% of the R$360,000 initially invested. This excludes follow-on contributions, expenses and losses in the other three projects. The denominator matters when communicating the result.
The half-loss assumption for Gama is an exercise input, not an observed quotation or exit price. Private holdings may have no buyer willing to pay an estimated value. Distinguish realized loss, monitoring valuation and assumed recovery. Combining them can suggest stability that does not exist. An estimated remaining value is not cash available to meet obligations or replenish the reserve.
Also test a funding shock in which no company raises money within its planned window. Even businesses serving different customers might then need support from the R$180,000 reserve simultaneously. Ask which can reduce spending, achieve a smaller milestone or operate on customer receipts. The reserve cannot automatically meet every need. Stress testing should expose competing demands and difficult choices rather than assume additional money will resolve them.
Use evidence and disagreements to update hypotheses
Not every disagreement about risk is a conflict of interest. A founder may see an unpaid pilot as a commercial opening, while an investor sees insufficient evidence of demand. Discussion improves when each identifies the disputed conclusion and what observation would change it. 'We need traction' is vague. 'We need three new sales paid without exceptional discounts' is specific enough to debate and verify.
Organize evidence by both strength and limitation. A reconciled bank statement helps demonstrate collection but does not explain satisfaction. An interview reveals motives but may be influenced by a personal relationship. A contract establishes obligations but may contain unmet conditions. No document answers every question. Different kinds of evidence addressing the same hypothesis reduce dependence on one narrative without removing the need to assess consistency and provenance.
Suppose Alfa reaches five paying customers, but four are personal contacts of the founders. The numerical milestone is met, while repeatable acquisition remains unproven. Record that limitation and design a test involving independent buyers within the available cash and time. Quietly redefining 'customer' to preserve the appearance of progress damages everyone's decisions. The useful response is to distinguish what was learned from what remains unresolved.
Keep a short record of hypothesis, evidence, decision and review condition. Include material disagreements and responsibility for resolving them. If new information contradicts the original thesis, reducing exposure or declining a follow-on investment may be consistent. Money already spent does not make the next decision attractive. Learning requires allowing conclusions to change while preserving the history and assumptions that produced the earlier judgment.
Negotiate protections without obstructing operations
Connect each negotiated protection to the risk it addresses. Information rights respond to unequal access. Rules on new issuances and participation in rounds address dilution and financing. Consent requirements may constrain decisions that materially change exposure. A clause should not be included merely because it appeared in another agreement: its effect depends on the company's legal form, the transaction and the full set of rights.
Overly broad protections can create new risk. If every hire or commercial discount requires approval from several investors, the company may lose speed while nobody takes responsibility for the delay. Distinguish ordinary decisions managed by the team from structural changes. Procedures should identify responsible parties, response periods and treatment of urgent matters, using wording appropriate to the corporate structure and practical demands of the business.
Conflicts arise when an investor also sells services, recommends suppliers or holds a position in a competitor. A referral may be useful, but compensation, relationships and alternatives should be disclosed. Buying the investor's services should not become an informal funding condition. Information moving between portfolio companies requires care so that one holding does not benefit at another's expense. Access and commercial influence deserve scrutiny alongside formal voting rights.
Define the operational role. Advice, introductions, governance participation and contracted work are different functions, with different hours, costs and limits. Founders remain responsible for execution within their duties. An investor intervening daily without sufficient availability or competence can create dependency and confused authority. Clarifying responsibilities helps identify whether the arrangement adds expertise or creates an unpriced workload and a new source of operational disruption.
Technology and intermediation require sound documentation
KEZS connects entrepreneurial projects and potential funders through technology and intermediation. A project-based approach can help organize the problem, team, intended use of funds, instrument and verifiable milestones. Its usefulness depends on the quality and transparency of the information supplied. Investors perform their own diligence and make their own decisions; KEZS does not assign a credit score. Organization supports inquiry without replacing independent judgment.
Documentation should distinguish the entrepreneur's statements, submitted evidence and the investor's analysis. An attached document does not make every claim true. Clear records of versions and origin can help identify changes, unresolved questions and inconsistencies. This may reduce repeated work, but it does not replace checks of identity, authority to represent the business, offered rights and economic consistency. Maintaining reliable records also takes time and resources.
How an opportunity is communicated and intermediated matters. Public securities distribution and regulated financial activities require appropriate structures and providers. Technology or a CNAE business activity classification is not permission to conduct them. Banco Central do Brasil and CVM materials can serve as reference entry points for study; they do not prove KEZS authorization or establish that a particular offer complies with applicable requirements.
If a solution uses blockchain, record integrity does not establish factual truth, legal title, reserve solvency or payment. Private documents should not be placed on public blockchains. Access controls, data protection and maintenance costs still matter. A technical layer should address an identified documentation or process problem. It cannot remove commercial or legal uncertainty, guarantee repayment or liquidity, or turn an unverified claim into an established fact.
End each review with an actionable decision
Prepare a one-page decision record for each project. Identify the right to be acquired, contribution and expected total cost, central business hypothesis, dominant sources of loss and dependencies shared with other holdings. Add available cash, the next milestone and the window for obtaining evidence. Separate established facts, representations and assumptions. This makes the basis of the decision reviewable without suggesting that unresolved questions have disappeared.
Then assess the portfolio together. In the example, Alfa's follow-on investment preserves 10% ownership, consumes R$30,000 of the reserve and increases concentration in the shared commercial partner. All three consequences coexist. An informed decision could be to invest, renegotiate, await evidence or decline. Maintaining a percentage should not obscure the other effects, including additional capital at risk and reduced flexibility for decisions elsewhere.
Correct process errors when identified. If the dependency map omitted a shared supplier, update every affected holding. If reports confused invoicing with cash, reconcile the history before comparing growth. If contracts have committed part of the reserve, remove that amount from funds available for new decisions. Corrections should reach the calculations and documents that relied on the incorrect information, preserving an understandable record of what changed.
Assign responsibilities and link the next review to an event: period-end reporting, pilot completion, customer loss or a funding proposal. Diversification by sources of risk means maintaining the ability to compare and respond as evidence changes. Exposure to loss, illiquidity and uncertainty remains. A clearer process helps investors understand which risks they chose, what they paid to assume them and what would justify the next decision.
Frequently asked questions.
Is investing in different sectors enough to diversify?
Different sectors may reduce some exposures, but they do not reveal every dependency. Clinic software and retail software may rely on the same commercial partner, infrastructure supplier or investor group. Examine what could interrupt sales, delivery and financing in each project, then identify events that could affect several holdings together. Sector variety is useful within that map. It does not replace investigation of the underlying causes of loss or establish that the businesses will respond differently to a shared shock.
How can projects be compared with almost no financial history?
Compare the quality of their hypotheses, evidence already produced and cost of testing unresolved questions. A project with few sales may have reconciled payments, repeatable delivery and independent customers. Another may have many interested people without financial commitments. Set milestones appropriate to each stage and estimate cash consumption before reaching them. Avoid filling gaps with excessively precise projections. Missing history means uncertainty to investigate, rather than evidence of low risk, and comparisons should preserve differences in what has actually been demonstrated.
Should maintaining an ownership percentage be the priority in every round?
Maintaining a percentage is an economic choice subject to the rights agreed. In Alfa's example, another R$30,000 preserves 10%; not participating reduces the holding to 8%. The additional contribution also reduces the reserve and increases concentration in the shared sales channel. Consider price, new evidence, round conditions and alternatives together. Accepting dilution can be reasonable. Preserving ownership on unfavorable terms can increase exposure without sufficient compensation, and neither percentage establishes a realizable investment return.
Does a convertible remove the need to analyze the cap table?
No. Future ownership depends on the instrument's rules and other existing commitments. Model conversion under different conditions, examine how discounts and valuation caps interact, and check the treatment of options and other convertibles. Also establish what happens if no round occurs or the company closes. Postponing part of the ownership calculation does not remove the need to understand rights, obligations, potential dilution and competing claims. The same contribution amount can produce different outcomes under different documents.
Which information deserves close attention without overburdening the team?
Prioritize decision-useful information: cash and net consumption, material obligations, collections separated from sales, milestone progress, critical dependencies and ownership changes. Agree on frequency, responsibility and supporting evidence. Events threatening continuity require communication outside the regular schedule. Reporting depth can grow with the business, but length does not compensate for figures that cannot be reconciled. Access should reflect confidentiality, data protection and each recipient's actual need, while reporting costs remain proportionate to the company's resources.
Do debt and receivables automatically make an angel portfolio safer?
No. They change rights and exposure without eliminating limited payment capacity. Debt can strain cash before validation. Existing receivables require checks of origin, ownership, disputes, transfer conditions and prior commitments; hoped-for sales are not existing claims. Apparently different instruments may still depend on the same customers or future fundraising. Follow the underlying economics and realistic recovery possibilities, rather than relying on a contract's name, a repayment schedule or an unexamined reference to security.
How should results be assessed before a portfolio provides liquidity?
Separate contributed capital, expenses, money actually received and estimated values. A funding-round valuation provides information but may not represent a price at which the entire holding could be sold. Record assumptions and restrictions behind estimates, and disclose known losses without offsetting them against hypothetical gains treated as certain. Before an exit or distribution, emphasize the development of evidence, cumulative exposure and financial capacity to continue. Estimated value cannot automatically pay expenses or replace cash available for future decisions.
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.
