Nonprofit Impact Report Template

Cover one period, usually your fiscal year, and give each program its own section. Lead each one with outcomes (what changed for participants, such as reading levels gained) rather than outputs (what you did, such as sessions held). An annual report adds organization-wide finances and thanks, and the annual impact report writer drafts one from your figures.

Every rate divides by a count of people, so build the counts first. For each program, count everyone you reached, enrolled, completed and measured, then those who achieved the outcome. Write the rule beside each count, such as "completed means 24 of 32 sessions."

Each count should be no larger than the one above it. If more people completed than enrolled, the two counts came from different rules or records, so fix that first. Then count distinct people: anyone in two programs appears in both totals, so the sum overstates how many you served.

Put the count beside every percentage. The Urban Institute's guide to analyzing outcome data stresses identifying clearly the population in the denominator¹. Percentages alone, it notes, don't show whether a rate rests on very few people or very many.

Coverage is the share of completers you measured: in the example, 268 of 341, or 78.6%. Report it for every program, and judge it as you would a survey response rate.

As a rule of thumb, the Urban Institute's guide to client surveys calls a response rate of 70% or better very good². Between 50% and 70% can be decent, but may need a check that the people who answered are representative. Results from rates much below 50% deserve some skepticism.

Run that check with your own records: compare the people you measured with those you missed on attendance, starting level and when they stopped coming. In the example, 51 of the 73 untested learners stopped coming before the exit test, in the course's last third, so they are the least likely to have improved.

When the groups differ, don't assume the unmeasured did as well. Publish the range, here 50.1% to 63.8%, and say why it is that wide.

Cost per outcome is a program's full cost divided by the number of people who achieved the outcome. Charity Navigator, which evaluates US charities, says it quantifies the financial investment required to achieve a single successful result³. It rates the figure reasonable, borderline or unreasonable against peers with similar models. Under US federal grant rules, an award's terms can also require unit cost data⁴.

Start with full cost: direct costs plus the program's portion of shared costs, such as finance staff and rent. FASB, which sets US accounting standards, requires a nonprofit's financial statements to describe the methods used to allocate costs among program and support functions⁵. If yours follow them, start from that method, and name any cost you leave out, such as fundraising.

Keep everyone's cost in the figure. The FRIENDS National Center, a federal service for child abuse prevention programs, says cost per outcome counts the cost of serving people who did not reach the outcome⁶. Its cost analysis rules add that costs and counts must cover the same period.

Publish all four, and name the stage on any figure you lead with. The stages can rank programs differently. In the template's three-program example, literacy is the cheapest per person reached, while housing is the cheapest per outcome.

A trend means something only if both years were counted the same way. The Urban Institute's analysis guide warns that major changes to measurement indicators or data collection procedures¹ can make comparisons over time misleading.

So if a definition changed, recount last year from its records on this year's rule. Adjusting the published figure won't work: it has the old rule built in.

In the example, last year's report said 1,842 adults served, counting anyone who came once. This year's says 604, counting enrolled learners only. Side by side, that looks like a 67% drop. Counted the same way in both years, the program grew: 20% more people reached, 8% more enrolled and 14% more completed.

Next, find where any change in outcomes came from. Outcomes are completers times the share measured times the success rate, so swap in this year's figures one at a time and note what each adds. In the example, nearly half of the 33 extra outcomes came from testing more learners, not teaching them better. The success rate among those tested even slipped, from 64.8% to 63.8%.

A case study shows what a result looks like in one life. Say whether the person is typical or exceptional, and use the same outcome measure as the totals. Include what didn't go to plan, and let the person approve the final text.

Then list what the figures can't show. The Urban Institute's analysis guide is blunt: outcome data by themselves can never prove what causes the outcomes¹. Without a comparison group, say participants improved, not that the program caused it. Testing whether it did takes a formal evaluation, which a program evaluation plan sets up.

If your outcomes were never set as measures and thresholds, define them for next year with a logic model. When a funder defines an indicator differently, report it on their terms in your grant report. To start from your own records, open this template in River.