Impact and outcome indicators for business support programs

Key Takeaways

• A good program indicator tells you what changed in the companies and when it can be verified. Activity indicators help manage the operation, but they don't show that the program worked.
• Colombia's National Planning Department (DNP) classifies indicators into three types along the value chain: management, output and outcome (DNP, Guía para la construcción y análisis de indicadores, 2018). The same guide warns that outcomes can be influenced by external factors.
• Few studies find significant effects of business training on sales or profits, and many measure before a year has passed (McKenzie and Woodruff, World Bank Research Observer, 2014). Promising sales at closing means promising something hard to prove.
• In a trial with 432 small and medium enterprises in Mexico, the largest effect of consulting (about 50% more jobs) showed up in social security records and was still there five years later (Bruhn, Karlan and Schoar, Journal of Political Economy, 2018). A report at closing would have missed it.
• Impact takes time and a point of comparison. Without them, the honest report talks about outcomes and contribution, not impact.

The indicators of a business support program are almost always written at the end, when the document is ready and someone asks for "the matrix." That order produces what most reports show: training hours, workshops held and companies enrolled, because those are the numbers the operator controls and can count without leaving the office.

This guide goes the other way. It starts from the change the program wants to produce and ends with the technical sheet for each indicator, with examples from programs we have run for chambers of commerce, multilaterals and governments in Latin America. It is one of the pieces of how to design a business support program, the main guide in this section.

What is an impact indicator in a business support program?

An impact indicator measures a change in the companies' business (sales, jobs, investment raised, survival) that can be attributed to the program, because it is compared against what would have happened without it. The word that sets it apart is "attributed." If there is no comparison group, or at least a reasonable way to rule out other causes, what you have is an outcome indicator.

The DNP indicator guide organizes indicators by the link of the value chain they measure. Management indicators quantify inputs and activities, output indicators measure the goods and services delivered and who received them, and outcome indicators measure changes in perception, knowledge and well-being. About outcomes, the guide says something that matters a lot in business programs: their effects can be influenced by external factors and are not necessarily produced directly by the public intervention.

Applied to a business support program, the chain looks like this:

The first two rows answer whether the program was carried out. The last two answer whether it was useful.

Why do so many programs end up reporting activities?

Because activities are the only thing the operator fully controls, and funders usually ask for reports before the companies have had time to change. You can see the result in figures like the one that opened AuRA, the program run by Fundación Bolívar Davivienda and IDB Lab: eleven Colombian support organizations served more than 30,000 companies between 2022 and 2024, and 49 investments closed in that period. Every organization had its reports up to date. What almost nobody measured was what happened inside the companies. We wrote about it in the end of vanity metrics in business support.

The opposite extreme exists too. Some programs promise sales growth at the end of a three-month engagement. The evidence doesn't back them: McKenzie and Woodruff's review of business training evaluations in developing countries found that most owners adopt some of the practices taught, that those improvements tend to be modest, and that few studies detect significant effects on profits or sales. It also noted that many evaluations measured before a year had passed.

Between counting workshops and promising sales there is a middle ground, and that is where useful indicators live: verifiable changes in how the company operates, measured with the same instrument before and after, plus a follow-up measurement for what takes longer.

How to build a program's indicators step by step

1. Start from the expected change, not from the list of activities

The first indicator you write is the outcome indicator, and it comes straight from the change the program promises. If the program wants companies to sell online, the indicator is how many have a working online sales channel, not how many attended the e-commerce module.

A useful test is to ask what an evaluator would see if they visited the company the day after closing. If the answer is "a certificate," the indicator sits in the wrong place in the chain.

2. Choose a few indicators per level and drop the rest

A program with thirty indicators can't be tracked and ends up reporting the ones that are easy to fill in. One or two per level are usually enough. To decide which ones stay, the DNP guide proposes the CREMAS criteria: the indicator should be clear, relevant, economical to collect, measurable by a third party, adequate for the dimension it measures, and sensitive to changes over short periods.

Sensitivity is the criterion that rules out the most in business programs. Annual sales don't move in a three-month engagement, even when the program worked. A segment decision or an enabled channel does.

Management indicators aren't dropped; they are used for something else. In El Salvador, the program with 105 digital transformation advisors logged 591 sessions and 1,397 hours of field work with 210 small businesses. Those numbers told us whether coverage was advancing at the expected pace in each department. The outcome was what each business had working after the session.

3. Write the technical sheet for each indicator

The technical sheet, which the DNP calls the indicator's "résumé," is what keeps two people from measuring the same thing differently. It should include name, objective, definition of each variable, formula, unit of measure, source, frequency and owner, plus the baseline and the target.

Here is a sample sheet built on the baseline of an e-commerce program with 65 small businesses in Panama, where none had an online store at the start. The target is left blank because each program sets its own:

The "definition" line is the one most often skipped, and the one that saves the most arguments in the final report.

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

If the entry and exit instruments aren't the same, the difference between them means nothing. Same questions, same dimensions, same scale and, if possible, the same person applying it.

The baseline also helps adjust the design before you start. If most of the cohort has already solved what the program meant to teach, there is still time to shift the focus.

5. Define how each data point is verified

Every indicator needs a verification rule: self-reported by the company, verified by the advisor, or taken from an external record. All three are valid, but they don't carry the same weight in a report, and you should say which one you used.

In Innovalab, the Bogotá Chamber of Commerce program, twelve companies reported revenue at the start and end of their engagement, and the number went up in every case, almost tripling in one. We didn't present it as a program result: the figures were self-reported, over a short period and without a comparison group. What we could verify was that at least eleven of the 67 companies pivoted or redefined their value proposition. The details are in who to sell to: what came up most in 67 companies.

6. Schedule the follow-up measurement and decide how you will talk about impact

Sales, new customers and jobs are measured six and twelve months after closing, with a short instrument the company can answer in a few minutes. That measurement has to be in the budget from the design stage, because after closing the operator no longer has a contract or a team to do it.

To talk about impact you also need a point of comparison. A practical option is to also measure the companies that applied and didn't get in, as long as the selection criteria don't make them too different. If there is no way to compare, the report should talk about outcomes and the program's contribution, and not about impact.

Outcome indicators by type of program

The outcome indicator changes with the program format. This table offers a starting point for each format, which each program adjusts to its expected change:

In every case, the impact indicator measured later (sales, jobs, investment) can be the same. What changes is the intermediate outcome that makes the link between the program and that impact credible.

Real case: 432 small businesses in Mexico measured five years later

The case that best shows why the timing of measurement matters is not ours. Starting in 2008, the state government of Puebla, Mexico, offered a year of subsidized management consulting to small and medium enterprises, and a team of researchers (Miriam Bruhn of the World Bank, Dean Karlan and Antoinette Schoar) used the program to run a randomized trial with 432 firms. Some were assigned consulting and others were not, so the comparison group was defined from the call for applications.

After one year, the firms that received consulting showed higher productivity and return on assets than the comparison group, and the most visible changes were in marketing, financial accounting and long-term planning practices. These were outcome indicators: changes in how the company was run.

The largest effect came later. Using Mexican social security data, the researchers found that the number of employees and the total wage bill of the treated firms grew by about 50% relative to the comparison group, and that the difference persisted five years after the program. The results were published in the Journal of Political Economy in 2018.

For anyone designing indicators, the case leaves three lessons:

The third lesson is the easiest to copy in small programs. An administrative record, such as social security payroll filings or the annual renewal of the business registration, makes it possible to measure jobs and survival years later without surveying anyone again, as long as the company authorized that use of its data when it joined the program.

Frequently asked questions about impact indicators in business programs

What is the difference between an outcome indicator and an impact indicator?

An outcome indicator measures a change in the company; an impact indicator measures the part of that change caused by the program. To measure outcomes, it is enough to compare the company with itself before and after. To measure impact, you need to compare it with similar companies that didn't take part, because a company's sales depend on many things besides the program.

How many indicators should a business support program have?

Few: one or two per level of the chain are usually enough. Each additional indicator has a collection and verification cost, and when there are too many the team ends up filling in the easy ones. Five indicators with a complete technical sheet are better than twenty without definitions.

Is data reported by the companies themselves useful?

Yes, as long as the report says it is self-reported and doesn't present it as a program effect. For outcome indicators, ask for simple evidence (a screenshot, an invoice, a contract) or have the advisor verify it. For sales, self-reported data is almost always the only source available, which is why it should be read with caution.

How do you measure impact without a comparison group?

You can't rigorously attribute the change to the program, but you can report contribution. That means showing the full chain: the company had a gap at baseline, the program worked on that gap, the intermediate outcome was verified, and then the change in the business followed. If the funder needs attribution, the comparison group has to be built into the design from the call for applications.

Which indicators should the terms of reference for an operator require?

Outcome and output indicators, with their technical sheet and verification method, plus the obligation to build a baseline and support the follow-up measurement. If the terms of reference only ask for hours, sessions and attendees, the operator will optimize for those numbers. Asking for verifiable outcomes also opens the door to pay-for-results schemes.

An indicator is worth the decision it lets you make

The test of a good indicator system is that someone changes something after reading it: the operator shifts focus mid-program, the funder decides to repeat or redesign, the next call asks for a different company profile. If the indicators only serve to fill in the final report, they were chosen because they were easy to count.

If you are building the indicator matrix for a program or reviewing the one an operator proposed, we can work on it with you in a design session before the call goes out.

Request a design session | How we work with institutions

Read also: How to design a business support program | What is an ESO