A corporate venture capital portfolio can contain startups in different industries while remaining concentrated in the same risk. Factory computer vision, energy management and software supporting the financing of transport companies' receivables appear distinct. Yet all three may depend on industrial investment, overlapping buyers and the sponsoring corporation's willingness to open its operations. If that willingness disappears, product variety offers limited protection.
Diversification starts with the cause of a loss
Diversifying by sources of risk means identifying events that could damage several investees simultaneously. Analysis extends beyond sector labels to budgets, distribution, infrastructure, regulation, data, people and exit conditions. Ask a concrete question: if one variable changes, which revenues, pilots and valuations change together? That reveals relationships a sector classification may hide, including dependencies that run through customers' customers.
CVC introduces a particular connection: the corporation may be an investor, customer, data supplier and the startup's principal commercial reference. This proximity can provide knowledge and access while concentrating four dependencies in one organization. Investors must distinguish an advantage that reduces technical uncertainty from an arrangement that relocates uncertainty into internal decisions they cannot easily observe or control.
The purpose is to give every material dependency a reason, a limit and an accountable owner. Some concentrations may be deliberate because they support the program's competitive advantage. Record them as explicit choices, with consequences and alternatives. Several companies with different Brazilian corporate registration numbers, or CNPJs, do not establish diversification when the same decision could undermine them all.
Separate equity returns from strategic benefits
Before comparing startups, the mandate must distinguish appreciation of an ownership interest from benefits to the corporation's operations. A solution can reduce waste while the equity investment loses value. An ownership stake can also produce a successful sale without internal adoption. State which outcomes the program seeks and which tradeoffs it accepts, so that every eventual result cannot be reclassified as success.
An operating saving belongs to the business that achieved it; it does not automatically enter the investment vehicle's cash balance. Factory savings from a pilot may support a purchasing decision, but they are not a distribution to the startup's shareholders. Record the strategic benefit separately, identifying its calculation, implementation costs and validating owner. Adding it to financial returns without explaining the measure misleads.
Learning, access to technology and supplier development are legitimate objectives when defined beforehand and supported by proportionate evidence. Learning might appear in a rejected hypothesis, an architecture decision or a capability absorbed by the team. An interesting presentation or several meetings does not establish that the program acquired a valuable strategic option. Explain what decision became possible and what uncertainty remains.
The mandate should specify the investment horizon, follow-on capacity, operating budget and stopping criteria. Accepting lower financial returns for technological access requires authorization and a measurable rationale. If competitive equity returns are the objective, the committee must be able to reject an operationally useful startup when its price, investor rights or commercial dependence make the investment difficult to justify.
Build a dependency map that can be checked
Start with a record for each investee: who pays, who uses the product, who approves purchases, what resources enable delivery and where the next financing is expected to come from. Identify critical dependencies such as a cloud provider, software license, dataset, specialist or corporate systems integration. Include enough detail to describe how a loss could occur, rather than merely completing a checklist.
Distinguish market, execution, financing, legal and corporate relationship risks. One startup may have diverse customers but rely on a single third-party technology. Another may control its technology while selling to buyers with the same budget cycle. A third may report recurring revenue yet need additional capital before collections cover expenses. Categories organize the analysis; concrete dependencies explain how trouble would spread.
Label the evidence supporting each statement. Signed contracts and collection records establish bounded facts; forecasts, interviews and letters of intent support hypotheses with different confidence levels. Contracts also require interpretation: duration, termination rights, acceptance conditions and permitted volume reductions can change their economic significance. Treating every contracted revenue stream as equally stable removes the variation the map is intended to expose.
Consolidate the records by common cause. Several investees may depend on the same executive sponsor despite different customers and technologies. Shared platforms require examination of substitutes, migration time and cost. The output should be a short list of material concentrations, each linked to evidence, unanswered questions and possible decisions. Readers should be able to trace a conclusion back to what was actually checked.
Internal sponsorship is an investment exposure
An internal sponsor turns corporate interest into access to people, data, facilities and budgets. This can accelerate learning by connecting a startup with a real operational problem. Personal enthusiasm, however, is not an institutional commitment. Leadership changes, reorganizations or spending reductions can end a pilot even when the technology works. Assess that vulnerability before investing, including the effect of delays on the startup's cash.
Sound sponsorship identifies the problem owner, budget owner and person authorized to decide on adoption. These may be different people. The executive introducing an opportunity may not control procurement, while users may want the solution without having integration resources. Documenting responsibilities helps investors distinguish access to senior leadership from the practical ability to implement and pay for a solution.
Test continuity: who takes over if the sponsor leaves, and can another department understand the economic case? Can the work resume from existing documentation? Additional contacts help only when they have relevant authority. A recognized problem, identified resources and a transferable sequence of decisions reduce dependence more effectively than a long contact list built around one person's preferences.
Portfolio limits by sponsor should reflect operational workload as well as invested capital. Five small pilots may compete for the same security, engineering or procurement team. If only two can be delivered properly, financing five creates a queue that consumes startup cash. Possible responses include staggering projects, funding specific adaptations or pursuing external customers. More investments require implementation capacity as well as committee attention.
Specialists reduce uncertainty when they can disagree
Specialist access is a potential CVC advantage, particularly when technology must work under specific industrial, commercial or regulatory conditions. Its value depends on the questions asked. An opinion about an elegant solution cannot replace analysis of maintenance, availability, integration and total cost. Specialists should investigate barriers to repeated use, including products that perform well in controlled demonstrations but struggle in everyday operations.
Separate current capabilities from promised development. If an algorithm needs retraining for every customer, measure labor requirements and data availability. If equipment requires particular components, assess replacement lead times and alternatives. Determine which activities scale through software and which still require people, facilities or individual negotiation. The distinction affects margins, financing requirements and expansion speed, even when the underlying technology performs as intended.
Specialists bring conflicts and preferences. One department may resist a solution because it changes established routines; another may support it to preserve its own project. Record the reviewer's role, assumptions and observed tests. When opinions differ, seek an experiment that can distinguish competing explanations. Choosing the most senior person's view without testing the disagreement leaves the underlying uncertainty unresolved.
Governance should preserve three separate conclusions: the technology failed, the test was inconclusive, or the technology passed but the business remains unattractive. An inconclusive test may justify a limited additional check; it does not establish validation. Technical success can coexist with expensive customer acquisition, excessive corporate dependence or insufficient investor rights. Each conclusion supports a different decision and should be reported accurately.
Pilots, commercial contracts, debt and equity serve different purposes
A pilot purchases evidence about a defined hypothesis. A commercial contract establishes supply, delivery and payment obligations. Equity exposes investors to the business's residual value and the conditions governing distributions or an exit. Debt creates payment obligations under its contractual terms. These relationships can coexist, but none automatically substitutes for another. Their budgets, approval processes and measures of success should reflect their different purposes.
A paid pilot can support startup cash flow and reduce product uncertainty without establishing renewal, scalable sales or equity appreciation. Recurring purchases may be profitable for a supplier yet small relative to expansion costs. A loan can specify scheduled payments while remaining exposed to default, renegotiation and enforcement difficulties. Naming the instrument does not remove its economic risk or establish repayment capacity.
Existing eligible receivables differ from hoped-for sales: an expectation of winning customers does not itself create an established receivable. Convertibles require examination of conversion conditions, maturity, liquidity events and later fundraising effects. In Brazil, a business registered as a MEI cannot admit equity partners while remaining a MEI. The proposed ownership arrangement must therefore be compatible with the business's legal structure.
Keep pilot spending separate from equity capital, even within one program. Debt needs its own rationale and risk approval. A donation to a nonprofit has a different legal and economic purpose; it is neither an equity interest nor a promise of financial return. Separating instruments makes the true cost of learning visible and avoids presenting expenditure as a recoverable asset without a defensible basis.
A portfolio that looks more diversified than it is
Consider a hypothetical program with a total budget of R$ 12 million. It allocates R$ 8 million to four initial investments of R$ 2 million, reserves R$ 2 million for follow-on investments and sets aside R$ 2 million for diligence, pilots and integration. All figures are arithmetic examples, not market prices, rates or projected returns. This separation supports assessment of both equity performance and total program cost.
Investee A develops industrial visual inspection; B manages energy use at production facilities; C provides receivables management software for transport companies; D organizes clinic appointments and capacity. Their products and immediate buyers differ. Diligence nevertheless shows substantial industrial exposure in A, B and C: factories buy A and B, while transport companies using C move freight for those same economic groups.
Suppose industrial activity supports 80% of A's revenue, 70% of B's, 60% of C's and 10% of D's. Applying these percentages to each R$ 2 million investment gives R$ 1.6 million, R$ 1.4 million, R$ 1.2 million and R$ 0.2 million. Together, R$ 4.4 million represents 55% of the initial R$ 8 million. This is an approximate dependency indicator, not expected loss.
Revenue declines do not translate proportionally into equity losses. Margins, cash, debt and adaptability change the outcome, and the calculation assigns no probability to a downturn. A and B also share the corporation's pilot implementation team, adding operational concentration. The committee may accept both dependencies, reduce an investment or seek another opportunity. Four apparent sectors cannot establish four independent sources of risk.
Use limits without manufacturing statistical precision
Startup portfolios often provide few comparable observations, infrequent valuations and rapidly changing business models. A numerical correlation matrix can therefore imply more confidence than the evidence supports. Two investments without recorded losses may still share a cause of failure. Exposure limits, causal scenarios and documented judgment provide a useful starting point, with uncertainty and missing information made explicit.
In the example, the committee could set one internal limit for capital linked to industrial activity and another for projects relying on the same implementation team. These limits need different denominators: invested capital for financial exposure, available hours and simultaneous projects for operational capacity. Combining them into one percentage would conceal the distinction and could suggest spare capacity where none actually exists.
Treat revenue percentages as estimates that can change. A report can provide a range and identify the customers supporting its classification. If C expands among food transport companies whose demand is less dependent on industrial production, its exposure may shift. If D concentrates sales in clinics belonging to one network, another dependency emerges. The map must follow the actual portfolio rather than preserve its investment-day description.
An exception may be defensible when supported by evidence and capacity to absorb loss. Record the decision-maker, expected benefit and trigger for reconsideration. A limit loses value if repeatedly rewritten to fit an already preferred deal. Equally, selling a strong investee merely to improve a chart may be costly: assess the achievable price, liquidity, contractual rights and effect on the investment thesis.
Cash schedules: learning also consumes the budget
In the same hypothetical program, initial outflows consist of R$ 8 million in equity and R$ 400,000 in diligence. During months one through six, pilots cost R$ 1.2 million, or R$ 300,000 per investee, and integration costs R$ 400,000. Cumulative spending reaches R$ 10 million: R$ 8 million in equity and R$ 2 million in program expenses. R$ 2 million remains reserved for follow-ons.
These amounts are not interchangeable without a decision. Covering an integration overrun from the equity reserve reduces capacity to participate in a later round. Committing to pilots beyond available cash may force startups to finance corporate delays. Track payments already made, contracted obligations and estimated expenses separately. Cash in the account may already be committed and should not automatically be treated as available funding capacity.
In month twelve, the committee invests another R$ 1 million in A and R$ 1 million in D. Equity exposure becomes R$ 3 million in A, R$ 2 million in B, R$ 2 million in C and R$ 3 million in D. Equity investment totals R$ 10 million, using the complete R$ 12 million budget once expenses are included. A's additional funding still requires justification despite increasing a known concentration.
The schedule explains why a reserve is not absolute protection. Once invested, it is no longer available cash. If two investees need funding the following month, the program must decline, obtain new authorization or accept dilution and possible losses. Test those alternatives before exhausting the reserve. The corporation's future ability or willingness to replenish capital should remain an assumption requiring evidence, rather than an implicit backstop.
Dilution and exit: complete the equity calculation
For D, assume the initial R$ 2 million investment occurs at a R$ 20 million post-money valuation, giving the program 10%. In month twelve, D raises R$ 10 million at a R$ 30 million pre-money valuation. The program contributes R$ 1 million of that round, consistent with the cash schedule. Post-money valuation becomes R$ 40 million: the pre-money valuation plus the new capital.
The original holding falls to 7.5%: 10% multiplied by R$ 30 million divided by R$ 40 million. The new R$ 1 million purchases 2.5%, restoring total ownership to 10%. A subsequent R$ 20 million round at a R$ 60 million pre-money valuation proceeds without the program. Its stake becomes 7.5%, calculated as 10% multiplied by R$ 60 million divided by R$ 80 million.
Assume a sale in month sixty makes R$ 120 million available to D's shareholders. With proportional distributions and no debt, preferences, costs or taxes in this simplified calculation, 7.5% produces R$ 9 million for the program. Actual distributions require the payment waterfall and each share class's rights to be calculated. An announced company sale price may differ substantially from the amount available to the CVC holding.
Complete the example with no recovery from A, R$ 1 million from B in month thirty-six and R$ 2 million from C in month forty-eight. Total receipts are R$ 12 million, including D's R$ 9 million. The R$ 2 million nominal gross gain over R$ 10 million in equity merely covers R$ 2 million in program expenses. The nominal balance is zero before unmodeled costs and taxes; delayed receipts further reduce economic value.
Adverse scenarios must reach shared causes
The example shows D's importance without proving portfolio protection. If D's final distribution were R$ 6 million instead of R$ 9 million, total receipts would fall to R$ 9 million. The nominal loss would be R$ 1 million against R$ 10 million of equity capital and R$ 3 million against total program cost. This measures sensitivity to one outcome without assigning a probability or predicting its likelihood.
A fuller causal scenario begins with lower industrial investment. A and B face longer sales cycles, C's transport customers carry less freight and demand lower prices, and the corporation reduces its pilot team. One change affects demand, execution and financing together. Modeling a small, isolated revenue decline for each startup would miss the possibility that several need additional cash simultaneously.
Another scenario is a corporate leadership change without an external crisis. Pilots lose priority even though procurement remains formally open. Delays prevent the startup from providing references to other buyers and weaken its next fundraising effort. Transferable documentation, a replacement sponsor, payment schedules and freedom to pursue external customers address these mechanisms. They can improve options without assuring survival, adoption or continued financing.
Also test failure of a shared technology supplier, a data incident and a shortage of exit buyers. Identify affected investees, potential additional cash needs and decisions within the program's control. The purpose is to know what would stop, which obligations would continue and when further funding would enlarge a loss without sufficient supporting evidence. A scenario should inform a decision rather than merely describe a catastrophe.
Intellectual property can create value or hide dependence
Begin intellectual property diligence with the chain of rights: who developed the code, who assigned their work and which components belong to third parties? The startup must demonstrate its ability to commercialize the solution on the proposed terms. Repository access or a registered trademark does not settle every issue. Founder, employee, contractor and partner agreements may contain rights incompatible with the expected expansion.
Within a pilot, distinguish pre-existing technology, specific adaptations and joint development. A corporation demanding ownership of everything could reduce the value of the investee it has financed. A startup claiming unrestricted use of all results could reach information or solutions the corporation cannot share. Define each party's rights to use, reproduce, modify, develop and commercially exploit the relevant parts of the work.
Personal data, confidential information and intellectual property are related but distinct. Permission to test a model on a dataset does not establish unrestricted permission to reuse that data or train other products. Clarify purpose, access, retention, deletion and responsibilities. Guidance from Brazil's ANPD is an entry point for study; its application requires examining the actual documents and activities, including the limits of any permission.
Diversification suffers when several investees depend on the same license, exclusive data or corporate permission to sell externally. Similar clauses can constrain an entire portfolio despite different technologies. Map substitutes, adaptation time and cost. A digital or blockchain record does not prove truth, legal title or freedom from third-party rights. Blockchain integrity also establishes neither reserve solvency nor a payment guarantee.
Commercial independence preserves growth options
Commercial independence means the startup retains viable ways to sell, develop its product and negotiate with other customers. Cooperation with the corporation can support that independence when it builds transferable capabilities. The danger is combining dependence on one buyer with restrictions on alternatives. Access to a first customer then becomes a weakness at the next stage, especially when future financing assumes broader demand.
Exclusivity may be reasonable with proportionate compensation, a precise scope and a justified duration. A restriction covering one application or segment differs from a broad ban on serving competitors. Assess the market made unavailable and whether restrictions continue if purchases fail to materialize. Exclusivity without minimum revenue or an effective termination route can block growth while leaving the startup with insufficient compensation.
Preferential rights over a startup sale also require care. Their design may deter buyers unwilling to incur diligence costs while the corporation retains an advantage in the transaction. The corporation may nevertheless have a legitimate interest in protecting strategic integration. Negotiate procedures and deadlines that consider the investee and other shareholders without making an eventual transaction impractical or assuming a buyer will remain interested.
Operational independence also requires controls against using shareholder information to pressure commercial prices or conditioning a necessary purchase on acceptance of investment terms. Procurement, investment and governance decisions should remain identifiable, with conflicts recorded. A startup should be able to reject an unsuitable supply agreement without automatic consequences for its shareholder relationship. This separation strengthens the credibility of revenue used in portfolio analysis.
Pilot criteria should support an actionable decision
A useful pilot starts with a question answerable within its time and budget. For A, the hypothesis might be reducing undetected defects at a production stage without increasing false alarms beyond human review capacity. Define the baseline, sample, operating conditions and person responsible for checking results. Selected demonstrations can favor a positive conclusion that may not survive ordinary operating conditions or representative production volumes.
Pair technical criteria with economic criteria. Better inspection may still be uneconomic if it requires additional specialist hours and expensive equipment. Include integration, training, maintenance, security and productivity effects in total cost. A pilot can establish technical capability while supporting a product or pricing redesign. That is useful evidence when it prevents expansion of a solution requiring continuing subsidy to remain attractive.
Before starting, define which findings support a contract, another test, closure or referral for investment analysis. No purchase or investment promise is necessary; a clear procedure is. A successful test may still require commercial negotiation and a separate equity decision. This protects the startup from implied expectations and prevents the investor from treating user satisfaction as complete validation of the business.
Evidence should include failures, outages and modifications during the experiment. If the startup changes the model, identify observations belonging to each version. If users change their process, the baseline comparison may no longer hold. Reassess the design and narrow the conclusion accordingly. Do not conceal unfavorable results or extend a pilot indefinitely to postpone a difficult decision; further testing needs a defined question and budget.
Governance should resolve disagreements and correct mistakes
Allocate responsibilities without obscuring who decides. Investment teams assess price, rights and exit rationale; business units establish the problem and adoption feasibility; specialists examine the solution; legal, security and privacy teams address their respective risks. The committee integrates these assessments and owns the decision. A favorable opinion from one function cannot establish approval of matters it did not examine.
Suppose operations supports another investment in A because its pilot matters, while finance favors D because it is less dependent on the corporation. Translate disagreement into assumptions about contracting, implementation margins, required cash and funding alternatives. Funding A may be reasonable, as in the example. Minutes should explain why the expected benefit justifies greater concentration and which evidence supports that judgment.
A common mistake is investing to recover past pilot expenses. Sunk spending does not improve a new investment; compare additional capital with possible future outcomes, including closure. Another mistake is calling a startup independent because it has several customers when all were introduced by one corporation and rely on the same commercial authorization. Revisit the origins and autonomy of revenue rather than counting customer names.
Team incentives also create conflicts. Compensation based only on deals or pilots started can reward activity without demonstrated results. Quality measures should include evidence produced, price discipline, dependencies resolved and execution of closure decisions. Investigate losses to learn, without presuming individual failure: a consciously accepted risk can materialize after a sound process. Equally, a favorable outcome alone does not prove that controls or reasoning were adequate.
Turn the analysis into an investment routine
Begin with three concise documents: the program mandate, a portfolio dependency map and an opportunity decision record. Before each investment, review capital and operational capacity consumed, concentrations increased and rights that may constrain the investee. Afterward, update material changes. The objective is to preserve decision memory and accountability while keeping investigation central; document volume cannot substitute for evidence or a clear investment rationale.
Periodic reviews should compare assumptions with facts: did sponsorship continue, did integration stay within budget, did external customers buy, and can the startup deliver without extraordinary support? Rebuild the map when pilots close and before new rounds. Commercial expansion may reduce concentration, while losing a customer may increase it. Both developments can require investment decisions, revised limits or changed follow-on priorities.
Organizing project information and documents can support this process. KEZS connects entrepreneurial projects and potential funders through technology and intermediation. Investors retain responsibility for diligence and decisions; KEZS does not assign a credit score. Clear statements of funding use, proposed rights, evidence and dependencies help formulate questions. They do not establish investment quality, regulatory authorization or any guarantee of approval, repayment, profitability, liquidity or capital preservation.
Public securities distributions and regulated financial activities require appropriate structures and providers. Technology, a CNAE economic activity classification or blockchain registration is not permission. Banco Central and CVM references are study entry points, not proof of KEZS authorization. Keep private documents in controlled-access environments, off public blockchains. Before committing more capital, examine the risk's cause, evidence, rights, full cost and realistic exit alternatives.
Frequently asked questions.
How many startups does a CVC portfolio need to be diversified?
No standalone number settles the question. Ten investees may depend on the same corporate budget, while a smaller portfolio may spread its causes of loss more effectively. Examine position sizes, shared dependencies and monitoring capacity alongside the company count. Follow-on reserves also matter: financing many companies without sufficient capital or implementation resources can increase fragility. State which risks remain concentrated and why they are acceptable, rather than assigning a generic diversification label. This is a portfolio assessment method, not a prescription for an individual's asset allocation.
Should a successful pilot be sufficient grounds to invest?
No. A pilot answers its predefined experimental questions, usually covering only part of the technology, use case and economics. Equity investment also requires analysis of price, rights, team, external demand, capital needs and exit conditions. A passing pilot may expose excessive customization costs; a failed pilot may reflect a correctable problem without invalidating the whole business. Interpret the evidence within the investment thesis. Neither technical success nor user satisfaction establishes automatic investment approval, a purchase commitment or future returns.
How can strategic benefits be recorded without inflating financial returns?
Maintain separate records. Equity performance tracks investments, distributions, losses and relevant financial costs. Strategic benefits identify who received the benefit, how it was measured and the resources required. Operations should validate waste reductions against a defensible comparison without double counting. A combined management indicator can be used if its components and limitations are explicit. It must not be described as cash distributed to investors when it includes corporate savings or learning. The distinction also reveals whether a financially weak investment delivered a separately authorized strategic objective.
When can commercial exclusivity be acceptable?
When it protects a defined interest, has proportionate scope and provides compensation that can be assessed. Examine territory, applications, covered customers, duration, contracted volume and termination options. Broad restrictions can reduce future revenue and discourage other investors or potential buyers. Corporate share ownership does not automatically make exclusivity advantageous to the startup. The negotiation should specify what each party receives and what happens if expected purchases do not occur. Compare the compensation with the opportunities surrendered, including restrictions that could survive a reduction in corporate support.
What should a program do when several investees need funding simultaneously?
First identify whether requests share a cause, such as contracting delays, falling demand or unavailable external finance. Then compare each business's needs, recent evidence, investor rights and alternatives. Do not divide reserves mechanically or favor whichever company has already consumed the most resources. Accepting dilution, seeking other investors or refusing follow-on funding may be necessary. Assess what additional capital can realistically enable, including the cost of postponing an unresolved problem. Some losses may remain unavoidable, and the corporation's ability to replenish the reserve cannot be assumed.
How can confidential information be used without weakening startup independence?
Define the purpose of access, authorized recipients and boundaries between equity analysis and commercial negotiation. Information needed for diligence should not circulate indiscriminately to departments competing with the investee's customers or products. Agreements should address confidentiality, permitted use and conflicts, with separate consideration of personal data. Digital records can support traceability, but they do not authorize data processing or make public disclosure appropriate. Keep private documents off public blockchains. Where access is necessary, restrict it to the stated purpose and preserve accountability for how the information is used.
What role can a platform play in diversification by sources of risk?
Project information organized consistently can support comparison, documentation of assumptions and identification of gaps. Its usefulness depends on material quality, updates and independent investor verification. KEZS provides technological connection and intermediation between entrepreneurial projects and potential funders; this does not replace diligence, decisions or applicable regulatory structures. KEZS does not assign credit scores. Investors must still assess costs, conflicts, contractual restrictions, instrument-specific rights and possible losses. A clearer presentation helps investigation, but it does not establish that a project is approved, diversified, liquid or capable of delivering a financial return.
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.
