How to design a business support program that actually changes something in the companies

Key Takeaways

• A business support program is an intervention funded with public, development, or corporate money that seeks to change capabilities, decisions, or results in a group of companies. It is judged by what changes in them, not by the activities delivered.
• Between 2022 and 2024, eleven Colombian support organizations accompanied more than 30,000 companies, and 49 investments closed in that period, a 0.1% conversion rate. That figure was the starting point of AuRA, the program led by Fundación Bolívar Davivienda and IDB Lab.
• Design is decided before the first session. What change is expected, who is served, which instrument measures the baseline, and what gets measured after the program ends.
• Measuring sales at three months is a poor way to judge a program. At closing you can verify installed capability and decisions made. Commercial effects are measured at six and twelve months.
• Duration is set by what you are supporting. Each company moves according to its complexity, how fast it makes decisions and its learning pace, and the fiscal calendar shouldn't override that.

Nobody who designs a business support program lands on this page out of curiosity. They come because they are accountable: to a development bank that wants attributable results, to a board that wants satisfied members, or to a government plan that ends before the program matures.

This guide gathers what we have learned designing and running programs for more than 30 institutions in Latin America, with more than 1,200 companies supported. It sets out the order of decisions that, in our experience and also as business owners ourselves, separates a program that changes something in the companies from one that only reports well.

What is a business support program?

A business support program is a set of activities designed so that a defined group of companies gains a capability, makes a decision, or improves a result it would not have achieved on its own in the same time. It is almost always funded by a third party (a government, a chamber of commerce, a multilateral, a foundation, or a large company working with its suppliers) and run by an in-house team or a contracted operator.

The last part of the definition, "would not have achieved on its own," is what forces you to measure. If the companies were going to grow anyway, the program did nothing, even if it delivered everything it promised.

Very different interventions are designed under the same name. It pays to separate them, because each one serves a different kind of change:

Most good programs combine two or three of these formats in sequence. The common mistake is to choose the format first, out of habit or because the previous edition did it that way, and the expected change afterwards.

Why do so many programs fail to change the companies?

The main reason is that they are designed to comply with a format, or to spend some institution's budget, and they are measured by activities. Training hours, workshops held, and companies enrolled are metrics that report well and are easy to measure, but they don't really say whether anything changed in a single company's operations. Even so, what ultimately sustains the next funding round is not having to say that nothing happened.

The figure that sums it up was put on the table at the kickoff of AuRA, the Fundación Bolívar Davivienda and IDB Lab program where we take part alongside ten other entrepreneurship support organizations (ESOs) in Colombia: together we accompanied more than 30,000 companies over three years, and 49 investments closed. There was no shortage of work. The problem was what we were usually asked to measure: our own activities, and very little of what happened in the companies, at least not systematically. We tell the full story in the end of vanity metrics in business support.

Four other causes keep repeating:

• Design is separated from operation. The person who writes the program is often not the one who runs it, or has no field experience, and the document's assumptions don't survive the first week with real companies.
• Duration is set by the budget or by the length of a plan, not by the time the intended change needs, or measuring results months later is never budgeted.
• They want strong results while doing the same as always. If you want different results, you have to change how the program is measured and also its design, duration, formats and support. It isn't easy, because it almost always means changes in organizational culture and internal processes, and sometimes significant investments of time and money.
• Follow-up depends on spreadsheets and message threads. With five companies it works. With sixty, the team ends up spending more hours coordinating than supporting, and reports arrive when they can no longer be used to correct course.

How to design a business support program step by step

Order matters. Each step depends on the previous decision, and skipping one is paid for during operation.

1. Define the change you expect to see in the companies

Before talking about workshops, write down what will be different in a company when the program ends. Not "strengthen digital capabilities," but something observable: "the company has a working online sales channel and knows what margin it leaves." That statement is the heart of the program's theory of change, and everything else follows from it.

A simple test: if the change can't be verified in a half-hour visit to the company, it is still written as an intention, not a result.

2. Decide who you serve and how you select them

The same program produces opposite results depending on who gets in. A cohort of micro businesses that have never sold online needs to enable its first channel; one with working online stores needs to optimize them. Mixing them forces a design that serves neither well.

Define the entry profile by the problem the companies share, on top of size or sector. At Innovalab, the Bogotá Chamber of Commerce program, we supported 67 companies from sectors that had nothing in common, from digital health to the music industry, and most of them stumbled on the same decision: who to sell to. That overlap, more than the sector, is what made group work possible.

3. Build a baseline with the same instrument you will use at closing

Without a baseline there is no way to show progress, and with a baseline different from the exit measurement there isn't either. The entry and exit instruments have to be the same, with the same dimensions and the same scale.

In an e-commerce program for 65 micro and small businesses in Panama we measured nine capability dimensions at the start and at the end. All nine improved, and the average went from 2.81 to 3.13 on a 1 to 5 scale. More than the size, what made the figure useful was where the progress happened: the two largest gains came in digital security and e-commerce strategic planning, exactly the dimensions where the cohort had reported the biggest gap at the start. The program moved what it set out to move, and that can only be claimed because the baseline existed.

In the artificial intelligence workshops we run with companies we apply the same logic on a small scale: before opening any tool, each company writes down how much time and money the problem it will solve costs today. A rough estimate works. A blank does not.

4. Choose the format based on the change, not on habit

With the change, the profile, and the baseline defined, the format almost chooses itself. If the change is a business decision, you need one-on-one support. If it is enabling a tool, training with guided practice may be enough. If the territory is large, training multipliers goes further than a central team.

In El Salvador we trained and certified 105 advisors for the cultural and creative industries, who then supported 210 ventures in 11 of the country's 14 departments, with 591 sessions and 1,397 hours of field work. A central team would not have reached that coverage with the same budget. The details are in how we trained 105 digital transformation agents.

Design asynchronous channels as real channels, not as backup. In Panama, twelve of the 41 companies that finished with their plan reached the same result as the rest with minimal attendance at group sessions, relying on the course and on an artificial intelligence tool available outside session hours. Not everything has to be in person, and not every answer needs a video call. What matters is that each company can move forward in the process and get value from it.

5. Match the duration to the pace of what you support

Duration is not decided by the fiscal calendar. It is decided by the time the change you defined in step 1 needs.

At Caribe Exponencial, the program we run in Barranquilla for several ecosystem partners, we support high-growth companies for seven months, and we are now on our sixth cohort. Each company has its own particularities: some need less time and others more, depending on their complexity, how fast they make decisions and absorb change, their needs, and their learning pace.

When the calendar is compressed for external reasons, decide in advance what will run in parallel and what can't be cut.

6. Separate what you measure at closing from what you measure later

This is the most frequently skipped decision, and the one that does the most damage to the final report. At the end of a three to six month program you can verify installed capability, decisions made, and channels enabled. Sales, new customers, later implementations, decisions that need more time than the program has, new products or services, and in some cases even jobs usually come after the program ends. That doesn't mean they aren't worth measuring.

At Innovalab, twelve companies reported revenue at the start and end of their support, and every one of them went up. We did not present it as a program result: they were self-reported figures, over a short period, with no comparison group. What did change in that time was the clarity of the business model, and in at least eleven of the 67 companies that turned into a pivot or a redefined value proposition. That result doesn't show up on any indicator dashboard, and it is the most valuable one the program produced.

The practical recommendation is to include in the design, and in the budget, a follow-up measurement at three or six months and another at twelve, with a short instrument that asks directly about the impact measures you want to track over time.

A program that only reports the first two rows can't claim it worked. One that promises the last row at closing is promising something it won't be able to back up. How to write each indicator, with its technical sheet and verification method, is covered in indicators for business support programs.

That said, we are not big fans of measuring hours or the number of sessions held, for example, because they don't necessarily mean more impact or more learning. Something we often tell our clients is that more person-hours don't make a program better. What matters is the transformation the program will create in the business or in the people in the organization, and the less time the business owner has to spend to get there, the better, because time is the most valuable resource they have.

Of course, this doesn't mean time shouldn't be measured. It is an important variable, but it should be measured along with everything else and with a clear purpose, beyond quantifying the cost of the program.

7. Design the operation and how you will contract it

The last step is deciding who runs the program, with what system, and under what payment rules. If you are hiring an operator, the terms of reference must ask for what you defined in the previous steps: the expected change, the baseline instrument, the follow-up measurement. If they ask for hours and workshops, that is what you will get. How to write them without leaving ambiguity in payment, figures and responsibilities is covered in how to write terms of reference for an operator.

More and more funders tie part of the operator's payment to verifiable milestones in the companies and leave the rest tied to activities. It is the pay-for-results model we are exploring within AuRA. It demands more from the operator, but it aligns incentives with what the institution needs to report, and it moves the how, or the means, to the background and puts the intended result first.

Real case: 65 businesses in Panama, from zero online stores to a capital fund

Between April and August 2026 we ran an e-commerce program for 65 Panamanian micro and small businesses, eight in ten of them micro businesses and 72% with a woman as legal representative. The baseline showed that none had its own transactional online store.

Of the 65 invited, 53 logged into the learning platform and 41 completed the training with their plan delivered, against a target of 30. Sixteen presented their project to a fund that would allocate resources to carry it out, and nine were awarded a combined 113,100 dollars to carry it out.

Two lessons changed the design of the next edition. First: the digital advisor handles the technical side well, but the business decisions the investment opens up (which segment to sell to first, at what price and margin) need a different profile, which we added as a separate role. Second: claiming commercial impact at closing would have been convenient and incomplete, so we proposed measuring it at three and six months. The full case is in boosting e-commerce for 65 businesses in Panama.

Who runs the program is also a design decision

Many programs are run by entrepreneurship support organizations, known as ESOs: incubators, accelerators, business development centers at chambers of commerce, and specialized operators. Understanding what kind of organization is behind a program, and what its reason for being is, helps you choose who to work with and read their results in the right context. We explain it in what is an ESO.

In this section we will keep adding guides on each piece of the design: theory of change, impact evaluation, and how to select companies in an open call.

Frequently asked questions about designing business support programs

How long should a business support program last?

The sensible answer is that it depends: it should last as long as it takes to verify the change defined as the objective, and it depends on the target segment, that segment's capabilities, the expected change, the available resources, and the time for the follow-up measurement. There is no single recipe; it depends on each case.

What indicators should a business support program have?

It should have indicators at all four levels: activity, output, outcome, and impact, each with its own measurement moment. Activity indicators help manage the operation, not show success. Outcome indicators are the ones that tell you whether the program changed anything in the companies, and impact indicators need time and, if you want to attribute the effect, a comparison group.

Can sales be measured at the end of the program?

They can be recorded, but in general they should not be attributed to the program. There are two exceptions: when there is a comparison group, or when the sale depends directly on something the program enabled that didn't exist before, such as the company's first online sales channel. Even then, in two to six months of support, the sales a company reports depend on too many outside factors, and there is almost never a comparison group. At closing it is better to measure installed capability and decisions made, and to ask about sales again at six and twelve months.

Should the program be designed in-house or with the operator?

Whoever designs it should take part in running it, or the operator should take part in designing it. The programs that fail most often are the ones designed in a document and handed to a different team to execute. If the institution designs it, it should test its assumptions with someone who has run a similar program before publishing the call, or be willing to make changes along the way.

What is pay-for-results in a business support program?

It is a scheme in which part of the operator's fee depends on the companies reaching verifiable milestones, and the rest on activities being delivered. It requires defining the milestones and how they will be verified from the design stage. That is why it works best when step 6 of this guide is settled before contracting.

The program is decided before the first session

A business support program is won or lost in the design. When the expected change is written in verifiable terms, the baseline exists, and the follow-up measurement is budgeted, the operation has a direction and the final report has something to say. When it isn't, the program ends up measuring the only thing it can: its own activities.

If you are designing a program or preparing a call for an operator, we can review the design with you in a working session before it goes out.

Request a design session | How we work with institutions

Read also: Who to sell to: what came up most across 67 companies