Blog main article:
Understanding Optima: Why Organizations Get Stuck Between Local Optima—and Fail to See the Global Optimum
Executive Summary: Why a Good Decision Is Far from Being the Best Decision
What happens when a company carefully prioritizes its investment projects, allocates budgets efficiently, and yet fails to capitalize on millions of euros in potential value?
In nearly every executive board meeting, investment decisions are discussed based on financial metrics, strategic priorities, and available resources.
Projects are evaluated. Business cases are compared. Investment budgets are allocated. Compromises are reached.
The result is a portfolio that appears plausible, economically sound, and organizationally feasible.
Yet one crucial question often remains unanswered:
Is our chosen combination of projects truly the best economic solution—or merely a good solution among the alternatives we’ve considered?
This is precisely where the difference between local and global optima lies.
A local optimum cannot be further improved within a defined neighborhood. A global optimum, on the other hand, achieves the best value of the objective function within the entire permissible decision space.
In strategic investment planning, this difference can have significant economic implications.
This is because companies do not just decide on individual projects. They decide on combinations of projects that compete simultaneously for capital, resources, and capacity.
StratePlan combines mathematical optimization methods from operations research with AI-driven approaches to systematically analyze such complex decision spaces.
The crucial management question is therefore not: Have we optimized our projects?
But rather: Have we selected the economically best combination of our projects?
1. What Does “Optimum” Mean? The Mathematical Definition
In mathematics, the term “optimum” refers to a solution that maximizes or minimizes a specific objective value within a defined decision space.
The plural form is “optima.”
In optimization theory, a distinction is made between local and global optima.
Local optimum
A local optimum is a solution that cannot be improved by any admissible change within a defined neighborhood.
The solution may be optimal within this limited region, even though better results are possible elsewhere in the decision space.
Global Optimum
A global optimum is a solution that achieves the best value of the defined objective function within the entire feasible decision space.
In a maximization problem, no other feasible solution has a higher objective function value.
In a minimization problem, no other feasible solution has a lower objective function value.
Important: A global optimum does not have to be unique. Several different project combinations can achieve the same optimal objective value.
For companies, this distinction is of central importance.
An investment decision may be economically optimal within a business unit while simultaneously generating a lower total contribution compared to a cross-unit alternative.
The quality of an optimization therefore depends not only on the calculation method used, but also on the decision space and objective function chosen.
2. Why Organizations Often Remain Stuck in Local Optima
Companies are organized into departments, business units, locations, and lines of responsibility.
This organizational structure enables the efficient management of complex business processes.
However, it often leads to investment decisions being made within limited spheres of responsibility.
A production site optimizes its machinery investments.
The IT department prioritizes its digitalization projects.
The sales department evaluates measures to increase revenue.
Real estate management decides on renovations and site developments.
Each unit can make economically sound decisions within its area of responsibility.
However, the company often has a shared—or at least limited—overall budget.
As a result, different business units compete for the same financial resources.
Typical causes of locally limited optimizations include:
- Separate investment budgets: Capital is allocated within individual business units without considering all investment alternatives collectively.
- Different KPIs: Departments pursue individual goals that do not fully align with the company’s overall objective.
- Historical decisions: Projects that have already been approved are given priority, even though other combinations might be more economically attractive.
- Resource conflicts: Multiple projects require the same personnel or technical capacity.
- Limited evaluation of alternatives: Management committees compare only a manageable number of prepared scenarios.
These factors can lead to an organization successfully optimizing its individual divisions without achieving the best possible overall result.
The company optimizes its individual components—but not necessarily their economically optimal combination.
3. The Invisible Decision Space: Why 2^N Changes Management Logic
As soon as multiple projects are available for selection at the same time, a combinatorial decision space emerges.
In a simplified binary model, each project can either be implemented or not implemented.
With N independently selectable projects, there are theoretically 2^N possible project combinations.
The number grows exponentially:
- 7 projects: 2^7 = 128 combinations
- 10 projects: 2^10 = 1,024 combinations
- 20 projects: 2^20 = 1,048,576 combinations
- 30 projects: 2^30 = 1,073,741,824 combinations
- 40 projects: 2^40 = 1,099,511,627,776 combinations
- 50 projects: 2^50 = 1,125,899,906,842,624 combinations
- 100 projects: 2^100 = approximately 1.27 × 10^30 combinations
With 50 planned investment projects , there are thus approximately 1.126 quadrillion theoretically possible project combinations.
Not every combination is permissible due to budget limits, resource constraints, or project dependencies.
Nevertheless, this order of magnitude illustrates why traditional management processes can only consider a limited subset of the mathematical decision space.
An executive board can evaluate 50 projects individually.
However, it cannot fully compare more than one quadrillion project combinations with one another.
The real challenge is not in understanding individual projects. It lies in determining the economically optimal solution from among the possible combinations of these projects.
4. Why Discussions and Management Workshops Do Not Guarantee Global Optimization
Management bodies are indispensable for defining strategic goals, evaluating economic assumptions, and taking responsibility for decisions.
However, they are not mathematical optimization algorithms.
In practice, a few prepared scenarios are often compared with one another.
For example:
- A base-case scenario with the originally planned investment budget
- A conservative scenario with reduced spending
- A growth scenario with additional investments
- A strategic scenario with preferred projects
These variants can be carefully developed and economically plausible.
However, they often represent only a small portion of the entire mathematical decision space.
Choosing between four predefined scenarios does not automatically answer the question of whether a more economically viable solution exists among millions of other combinations.
The best-discussed option is not necessarily the best mathematically possible project combination.
This is precisely why the strategic discussion should be supplemented by appropriate mathematical optimization methods.
5. A Calculation Example: Why the Best Individual Investment Does Not Lead to the Best Portfolio
A company has an investment budget of 10 million euros.
There are three projects to choose from.
| Project | CAPEX | Net Value Added |
|---|---|---|
| Project A – Production Facility | €7 million | €4 million |
| Project B – Automation | €5 million | €3 million |
| Project C – Energy Efficiency | 5 million € | €3 million |
Project A has the highest single net value contribution.
If this project is selected, 3 million euros will remain in the budget.
This is not enough to implement any of the other projects as well.
The total net value contribution is therefore 4 million euros.
However, a mathematical analysis of the possible project combinations reveals a more economically viable alternative.
Projects B and C together require 10 million euros and generate a combined net value contribution of 6 million euros.
Assuming the projects are independent and with no further restrictions, B + C is the optimal combination for maximizing the net value contribution.
The optimized portfolio achieves a 50% higher net value contribution compared to the original selection.
The company does not need any additional investment capital for this.
It simply needs to select a different combination of its existing projects.
This example illustrates a suboptimal portfolio decision. Whether the original selection is also a local optimum in the strict mathematical sense depends on the definition of permissible neighborhood changes.
The key insight remains: The economically best individual investment is not necessarily part of the economically best combination of projects.
6. Why local optima can be so mathematically stable
Local optima often arise from iterative optimization processes.
An algorithm or a management process starts with an existing solution and attempts to improve it through minor changes.
For example, a project is removed from the portfolio and replaced with another.
If the overall economic contribution improves, the new combination is adopted.
This process can be repeated until no further improvement can be found within the neighborhood under consideration.
The result is then a local optimum.
However, a significantly better solution may lie outside this neighborhood.
To achieve it, it might be necessary to replace several projects simultaneously.
Individual intermediate steps could initially lead to worse results.
A simple local search algorithm might not implement such changes under certain circumstances.
A similar effect can occur in organizations.
A business unit gradually improves its existing investment planning without examining fundamentally different capital allocations.
This can result in an economically plausible and stable combination of projects, even though other portfolios would allow for a higher overall contribution.
The stability of an existing solution is therefore no proof of its global optimality.
7. Global Optimization: The Best Solution Within the Entire Permissible Decision Space
Global optimization aims to determine the best solution for a defined objective function within the entire permissible decision space.
In strategic capital allocation, this objective function can, for example, be the maximization of the expected net value contribution.
Mathematical optimization takes into account the available investment alternatives and the defined constraints.
Typical constraints include:
- Maximum investment budget
- Available human resources
- Technical capacity limits
- Project dependencies
- Time constraints
- Minimum strategic requirements
- Regulatory obligations
Appropriate optimization algorithms explore the mathematical solution space and search for a combination of projects that best satisfies the objective function.
Various methods can be used for this purpose:
- Integer and mixed-integer optimization
- Branch-and-bound method
- Dynamic programming
- Metaheuristics
- Multi-start methods and combined search strategies
- AI-based evaluation and search methods
Under suitable conditions, some of these methods can mathematically prove the existence of a global optimum.
Others provide approximately optimal solutions without providing a complete proof of optimality.
A complete individual calculation of all 2^N combinations is not strictly necessary to mathematically optimize large decision spaces.
Further background is explained in the article “New AI Computation Method in Project Portfolio Management for CAPEX Optimization.”
8. Why Excel, scoring models, and prioritization matrices do not guarantee global optimality
Excel, scoring models, and prioritization matrices are established tools for strategic investment planning.
They enable a structured evaluation of individual projects and the creation of economic scenarios.
Their limitations become particularly apparent when the ranking of individual projects is equated with a mathematically optimal portfolio selection.
Excel: Individual evaluations do not constitute portfolio optimization
Excel is excellent for calculating business cases, returns on investment, and investment scenarios.
With appropriate Solver functions, mathematical optimization problems can also be addressed.
However, a simple Excel list in which projects are sorted by ROI or net present value does not automatically account for all possible project combinations.
For large combinatorial decision spaces, specialized optimization methods may be necessary or more efficient.
Scoring Models: The Best Ranking Is Not Necessarily the Best Portfolio
Scoring models evaluate projects based on defined criteria.
The projects with the highest scores are prioritized for implementation.
This approach provides transparency regarding individual priorities.
However, it does not guarantee that the selected combination will achieve the highest overall economic or strategic contribution.
Prioritization Matrices: Simplification Rather Than Complete Optimization
Prioritization matrices help classify projects based on criteria such as impact, effort, urgency, or strategic importance.
They are useful for presenting complex information in an understandable way.
However, this does not automatically result in a mathematically optimal combination.
The crucial question is therefore: Do we evaluate individual projects—or do we calculate the economically optimal combination of all available projects?
9. StratePlan: Mathematical Portfolio Optimization Instead of Limited Variant Analysis
StratePlan was developed to mathematically optimize complex investment and project portfolios.
The technology combines methods from operations research with AI-based approaches to efficiently analyze large combinatorial decision spaces.
For a basic economic optimization model, only three pieces of information per project are initially required:
- Project ID: Unique identification of the project
- Expenses: Planned expenditures or investments
- Income: Expected revenue or economic returns
In addition, the available investment budget is defined.
The project data can, for example, be imported from existing Excel investment plans.
StratePlan analyzes possible project combinations and calculates a portfolio based on the selected objective function.
With “Maximize Net Contribution,” the focus is on maximizing the total expected net value contribution.
The portfolio ROI can also be reported as an economic metric.
Advanced models can account for additional constraints and evaluation parameters:
- Project dependencies
- Strategic weightings
- Resource and capacity constraints
- Multi-year investment budgets
- Time constraints
- Risk and sustainability criteria
The result is not merely a ranking of individual projects, but a mathematically calculated combination of projects within the defined parameters.
StratePlan thus adds a layer of mathematical optimization to the strategic decision-making process.
10. Live in the Board Meeting: Which project combination is actually the best?
Mathematical portfolio optimization becomes particularly valuable when it is directly integrated into strategic management decisions.
Let’s imagine an industrial conglomerate with 100 planned CAPEX projects.
The total of all proposed investments exceeds the available budget.
Several business units have already prioritized and evaluated their projects from a business perspective.
The executive board now wants to know which combination of all projects yields the highest expected net value contribution.
Scenario A: Cross-Divisional Optimization
The CFO asks:
“What changes if we no longer optimize each business unit individually, but instead consider all investment projects collectively?”
The projects are modeled in a single portfolio.
StratePlan calculates an optimized combination while taking the defined budget limit into account.
A comparison with the previous plan can show whether a different allocation of capital would enable a higher overall economic contribution.
Scenario B: Mandatorily Include Strategic Projects
The CEO asks:
“What economic contribution might we be giving up if certain projects must be implemented regardless of their individual economic viability?”
The projects in question are treated as binding constraints.
The portfolio is recalculated.
The comparison shows the impact of this requirement on the expected net value contribution.
Scenario C: Budget Reduction
The CFO wants to know:
“Which combination of projects should we choose if we have to reduce our CAPEX budget by 20%?”
The budget constraint is adjusted.
StratePlan calculates a new combination of projects.
Management can see which measures are being eliminated and how the expected net value contribution changes.
Scenario D: Additional Investment Capital
Another member of the executive board asks:
“How much additional economic value could we generate if we invested another 10 million euros?”
The budget is increased, and a new optimization is performed.
The results allow for a comparison between the additional capital investment and the expected value contribution.
Scenario E: Changed Strategic Priorities
The board asks:
“How does the portfolio change if we place greater emphasis on energy efficiency, digitalization, or specific growth areas?”
The corresponding parameters are adjusted.
The portfolio is recalculated and compared with the previous scenarios.
Change strategic parameters. Recalculate the portfolio. Compare economic impacts. Make more informed decisions.
This allows even decisions that deviate from purely economic optimization to be made in a conscious and transparent manner.
11. Opportunity Costs: How far is our portfolio from the optimal state?
Every investment decision ties up capital and resources.
This capital is then no longer available for other projects.
This results in opportunity costs.
They describe the lost benefits of the best alternative that was not chosen.
Mathematical portfolio optimization makes it possible to compare the originally planned investment portfolio with an optimized combination of projects.
In doing so, the following key metrics, among others, can be compared:
- Total investment requirement
- Expected net value contribution
- Portfolio ROI
- Selected and non-selected projects
- Remaining budget flexibility
- Economic differences between alternative portfolios
The difference between the modeled results shows the potential for economic improvement under the underlying assumptions.
If a global optimum has been mathematically proven, the distance between a selected project combination and this optimum can be precisely determined within the model.
For approximately optimal solutions, the difference from the best solution found is first reported.
The key added value lies in identifying potential economic drawbacks of an existing investment plan before capital is committed.
12. Sensitivity Analysis: How Robust Is the Global Optimum?
Investment decisions are based on assumptions about future economic developments.
Project costs may rise. Revenue expectations may fall. Resources may become scarcer. Strategic priorities may change.
Therefore, it makes sense not to limit oneself to a single portfolio optimization.
Sensitivity analyses allow for the examination of various economic scenarios.
Typical questions include:
- Which projects will remain part of the portfolio even with a reduced budget?
- How does the optimal project mix change as investment costs rise?
- Which measures are particularly sensitive to lower revenue expectations?
- How does the portfolio change under additional resource constraints?
- Which projects remain economically attractive under various scenarios?
These analyses can show how stable a calculated project mix remains under changing assumptions.
A strategically sound decision therefore takes into account not only the best outcome under a single assumption but also the effects of possible changes.
13. Case Study: Klybeck Site in Basel – Approximately 63% Higher Return on Investment
The StratePlan case study of the Klybeck site in Basel illustrates the economic potential that mathematical portfolio optimization can reveal.
As part of the portfolio optimization for Central Real Estate AG and Swiss Life AG, existing investment and project alternatives were analyzed mathematically.
The modeled result:
- Original return on investment: 7.0%
- Optimized return on investment: 11.4%
- Relative return improvement: approximately 63%
- Additional modeled economic value contribution: approximately 30 million CHF
The improvement was modeled by changing the combination of existing investment alternatives while maintaining the same budget.
These project-specific results do not constitute a general guarantee of return.
The key finding: A better combination of existing investment alternatives can enable a significantly higher overall economic contribution—without an additional investment budget.
14. Where local optima particularly often limit economic potential
The risk of isolated optimization of individual measures exists in numerous economic and public sectors.
Industry and CAPEX Planning
Production facilities, automation projects, capacity expansions, and site investments compete for shared investment budgets.
The isolated optimization of individual plants or business units can prevent a more economically efficient allocation of capital across all areas.
Digitalization and AI Transformation
Companies often identify numerous potential AI and digitalization projects.
Simply prioritizing individual use cases does not automatically reveal which combination yields the highest overall economic contribution.
Real Estate
Project developments, renovations, modernizations, and location investments can be considered within a shared capital budget.
The selection of individual projects influences the financing options for the entire real estate portfolio.
Energy and Infrastructure
Energy projects, grid investments, efficiency measures, and infrastructure modernizations often require long-term capital commitments.
Mathematical portfolio optimization can help compare different investment combinations under economic and technical constraints.
Municipalities and Public Budgets
Cities and municipalities must finance numerous investment projects within limited budgetary constraints.
Schools, roads, administrative buildings, infrastructure, and social services compete for the same public resources.
Prioritizing individual departments in isolation does not automatically lead to the combination of measures with the highest defined overall social benefit.
Mathematical optimization can examine different combinations of projects while taking into account budget constraints, mandatory responsibilities, and policy objectives.
The common thread among all use cases is the question of whether optimizing individual measures actually leads to the best possible overall solution.
15. Budget Simulation and Simultaneous Portfolio Optimization: Overcoming Local Optima in the Public Sector
The problem of local optimization becomes particularly evident in the planning of public investment budgets.
Municipal departments often have their own project lists and priorities.
The school district prioritizes investments in education. The building department evaluates infrastructure projects. Other departments plan digitization, energy, or real estate projects.
Each of these prioritizations may make sense within the respective area of responsibility.
The crucial question, however, is:
Which combination of all municipal projects achieves the highest defined overall economic or social benefit within the available budget?
A simplified example:
A city has an investment budget of 81 million euros.
At the same time, there are 50 projects to choose from with a total investment requirement of 220 million euros.
With 50 binary project decisions, there are theoretically 2^50 = 1,125,899,906,842,624 possible combinations.
Mathematical portfolio optimization can examine various permissible project combinations within the defined budget and impact constraints.
A simultaneous budget simulation allows for different financial conditions to be taken into account.
For example:
- Which project combination yields the highest modeled total benefit with a budget of 81 million euros?
- How does the portfolio change if the budget is reduced?
- What additional measures could be implemented if revenues were higher?
- How do mandatory policy projects affect the outcome?
- What are the effects of different weightings for education, infrastructure, or climate protection?
The quality of the results depends on the data used and the definition of the economic and social targets.
Combining budget simulation with mathematical portfolio optimization can help translate department-specific priorities into a comprehensive, transparent basis for decision-making.
16. From Business Intelligence to Decision Intelligence: Optimizing the Quality of the Decision Itself
Many companies already have powerful systems for analyzing and managing their business processes.
ERP systems track resources and business transactions.
Business intelligence solutions visualize key metrics and trends.
Controlling systems analyze budget variances and financial results.
These systems answer important questions:
What happened? How are our KPIs trending? Which projects are within budget?
Decision Intelligence expands this perspective by providing mathematical support for future decisions.
Which combination of projects should we choose? What constraints limit our potential value contribution? Which alternatives yield a better overall result?
StratePlan supplements existing planning and controlling processes with a layer of mathematical optimization.
The necessary project data can initially be provided via simple Excel spreadsheets.
The strategic advancement lies not merely in measuring the results of decisions already made, but in mathematically improving the selection and combination of the projects themselves.
17. Conclusion: The global optimum is a matter of the entire decision space
Organizations make economically sound decisions every day.
They evaluate projects, optimize processes, prioritize investments, and allocate resources.
However, a locally optimal solution is not necessarily globally optimal.
When numerous projects compete simultaneously for limited budgets and resources, a combinatorial decision space emerges.
Even with just 50 projects, there are more than one quadrillion theoretically possible combinations.
Therefore, evaluating individual projects in isolation does not automatically answer the question of what the economically best overall solution is.
Mathematical portfolio optimization makes it possible to systematically examine different project combinations and compare their economic impacts.
To this end, StratePlan combines methods from operations research with AI-based approaches.
This enables companies, executive boards, and public decision-makers to calculate alternative portfolios, examine strategic scenarios, and identify untapped economic or societal potential.
The crucial question is not: Have we successfully optimized our investment projects?
But rather: Have we selected the economically optimal combination of our investment projects?
This is precisely where the transition from local optimization to mathematical portfolio optimization begins.
StratePlan – Don’t just optimize individual projects. Calculate the best combination.
FAQ – Understanding Local Optima and the Global Optimum
What does “optimum” mean in mathematics?
An optimum refers to a solution that maximizes or minimizes a specific objective value within a defined decision space.
In a maximization problem, the highest achievable value is sought.
In a minimization problem, the lowest achievable value is sought.
What is the difference between “optimum” and “optima”?
"Optimum" is the singular form. "Optima" is the plural form.
In mathematical optimization, multiple local optima may exist.
Several different solutions can also achieve the same globally optimal objective value.
What is a local optimum?
A local optimum is a solution that cannot be improved by any permissible change within a defined neighborhood.
However, better solutions may exist outside this neighborhood.
What is a global optimum?
A global optimum is a solution that achieves the best value of the defined objective function within the entire feasible decision space.
In a maximization problem, no other feasible solution has a higher objective function value.
Why do companies often settle for local solutions?
Because investment decisions are often made within individual business units, departments, or predefined scenarios.
As a result, not all economically relevant combinations of projects are automatically considered.
Why isn’t the best individual investment automatically part of the best portfolio?
Because projects compete for shared budgets and resources.
An economically attractive individual project can tie up so much capital that several other projects with a higher combined net value contribution can no longer be financed.
What does 2^N mean in portfolio optimization?
2^N describes the number of theoretically possible combinations for N independent binary project decisions.
Each project can either be selected or not selected.
With 20 projects, there are 2^20 = 1,048,576 possible combinations.
With 50 projects, there are 2^50 = 1,125,899,906,842,624 combinations.
Do all 2^N combinations have to be calculated individually?
No.
Modern mathematical optimization methods can efficiently explore large decision spaces without having to fully evaluate every theoretically possible combination individually.
Which methods are suitable depends on the mathematical structure of the problem and the defined constraints.
Can AI compute the global optimum?
With suitable mathematical models and optimization methods, a global optimum can be calculated and verified.
For particularly complex problems, approximately optimal solutions can also be used.
The statement regarding the quality of the solution depends on the method used and, if applicable, a proof of optimality.
Can Excel solve global optimization problems?
Yes. Excel can handle mathematical optimization tasks using appropriate Solver functions.
However, for large combinatorial decision spaces and complex constraints, specialized optimization methods may be necessary or more efficient.
What data does StratePlan need for an initial optimization?
For a basic economic optimization model, three pieces of information are initially required for each project:
- Project ID
- Expenses – planned expenditures
- Income – expected revenue or economic returns
In addition, the available investment budget must be specified.
More complex models can take additional constraints and evaluation parameters into account.
What does “Maximize Net Contribution” mean?
“Maximize Net Contribution” refers to maximizing the total expected net value contribution of a project portfolio within defined constraints.
The portfolio ROI can also be reported as an economic metric.
Maximizing the absolute net value contribution and maximizing the portfolio ROI are different mathematical objectives.
Can StratePlan highlight the economic differences between local and global solutions?
By comparing an existing project mix with an optimized alternative, the difference in expected net value contributions can be calculated.
This difference illustrates the modeled economic improvement potential based on the underlying assumptions.
An exact measure of the gap to the global optimum requires that the global optimum has been mathematically proven.
Is the global optimum always the right management decision?
Not necessarily.
A mathematically optimal solution always refers to the defined objective function and the constraints taken into account.
Strategic, regulatory, or societal requirements may necessitate additional constraints.
Mathematical optimization helps make the effects of such requirements transparent.
For which organizations is global portfolio optimization particularly relevant?
It is particularly relevant for organizations with numerous competing investment projects and limited resources.
Typical areas of application include manufacturing, CAPEX planning, digitalization, AI use cases, real estate, energy, infrastructure, and municipal budget planning.
What is the most important takeaway for CEOs, CFOs, and public policymakers?
The economically optimal individual decision does not automatically lead to the best overall solution.
Mathematical portfolio optimization makes it possible to systematically examine alternative project combinations and improve capital allocation at the organization-wide level.
The crucial question is: Are we optimizing individual projects—or are we optimizing the selection of our projects?
Do you know the global optimum of your investment portfolio?
Find out which combination of your planned investment projects can achieve the highest expected net value contribution within your available budget.
Compare your current project planning with mathematically optimized investment scenarios and examine whether a different combination of existing projects could yield a higher overall economic contribution.
You can optimize up to 15 projects for free.