By Sopact · Updated September 12, 2026
Aggregate outcomes across a grant portfolio by mapping each grantee’s evidence to the funder’s intended outcomes, then combining only measures with compatible definitions, populations and periods. Show which grantees are represented, which results remain separate and where evidence is missing. A shared Theory of Change organizes the review; it does not make every result additive or establish that the funding caused the observed change.
This lesson is for foundation, grant and impact teams preparing a portfolio review. Bring the included and excluded record list from the related lesson, the outcome framework and the data dictionary. You will produce an outcome map, a defensible calculation and a coverage note for the report.
- Map evidence to the intended outcomes.
- Separate comparable measures from contextual evidence.
- Check populations, time windows and overlap.
- Calculate totals or rates using the right denominators.
- Disclose missing coverage and remaining limitations.
- Review the result and share it with appropriate access.
Map the portfolio to outcomes before adding numbers
An outcome map shows how the work funded across the portfolio relates to the foundation’s intended change. Keep grantee names and source references visible so a reader can understand where the evidence comes from.
Do not force every contribution into a single metric. Two grantees may contribute to economic mobility through different pathways: paid employment, income stability, affordable transport or access to childcare. Those observations may belong under a shared strategic outcome without being added together as “people with improved mobility.”
| Intended outcome | Evidence to map | What the map tells you |
|---|---|---|
| More secure employment | Comparable employment follow-ups, plus contextual accounts of job quality | Which grantees report relevant observations and what each observation covers |
| Better access to services | Service access measures and accounts of barriers | Whether availability, use and experience are being distinguished |
| An outcome with no current evidence | No qualifying observation in the reviewed package | A coverage gap to investigate, not proof of success or failure |
This is an illustrative map. Use your own Theory of Change and do not relabel unrelated outputs as outcomes simply to fill an empty row.
Decide what can be combined
Before summing a measure, check whether each contributing value means the same thing. A common column name is not sufficient. Retain the definition version, unit, observation window and population for each source.
The Operating Principles’ monitoring guidance discusses the difficulty of portfolio aggregation when definitions and methods differ. The worksheet below turns that concern into a practical review. It does not prescribe a universal set of portfolio indicators.
| Question | If the answer is no or unclear |
|---|---|
| Do the values use the same unit and definition? | Keep them separate or reconcile from the underlying evidence |
| Do they cover compatible observation periods? | Label the different periods; do not create a single trend point |
| Are the populations and eligibility rules compatible? | Explain the difference before comparing rates |
| Could the same person or result appear more than once? | Resolve overlap or label the result as non-unique counts |
| Are values observed rather than projected? | Separate forecasts from recorded results |
| Can a reviewer reopen the source? | Retain the verification gap rather than silently accepting the value |
Agreeing a small shared core can improve future collection. Grantees can keep local questions and instruments: map compatible local fields to the common definition and retain differences that cannot be reconciled. It does not automatically repair earlier observations. If the source cannot support a restatement, show a break in comparability.
Practice: combine two rates without averaging percentages
Fictional exercise. Two grantees report six-month employment outcomes using the same definition and eligible population rules. Grantee A has 100 eligible participants, 80 responses and 48 respondents employed. Grantee B has 50 eligible participants, 40 responses and 28 respondents employed. For this exercise, the cohorts do not overlap and every counted response contains a usable employment status.
A’s respondent employment rate is 48 ÷80 =60%. B’s is 28 ÷40 =70%. Simply averaging 60% and 70% gives 65%, but that treats the two response groups as equally sized. The combined respondent rate is(48 +28) ÷(80 +40) =76 ÷120, or about 63.3%.
| Portfolio measure | Calculation | Interpretation |
|---|---|---|
| Employment among respondents | 76 ÷120 ≈63.3% | Rate among people who responded, not all eligible participants |
| Response coverage | 120 ÷150 =80% | Share of eligible participants with a response |
| Observed employed share of eligible participants | 76 ÷150 ≈50.7% | Known employed participants as a share of the full eligible group; nonrespondent outcomes remain unknown |
Now add Grantee C, which reports 30 employed respondents out of 45 at a three-month follow-up. Keep that observation separate from the six-month rate. It can appear in the outcome map with its own time window, but including it in the same denominator would change the question.
These calculations summarize observations. They do not establish a causal effect of the grants, and nonrespondents may differ from respondents. State those limits alongside the result.
How do you avoid double counting?
Count a person or result once when the intended measure is unique people or unique outcomes. Within one grantee, a stable identifier can connect repeated observations. Across grantees, different identifiers do not establish that the people are different.
If two organizations support the same participants, summing their counts may measure service engagements rather than unique people. Use an authorized, appropriate matching process if unique counting is required. Where overlap cannot be resolved, report grantee-level counts or clearly label the combined figure as potentially overlapping.
Do not collect extra identifying information solely to make a dashboard total convenient. Decide whether unique counting is needed for the reporting purpose and use an appropriate data-sharing and access arrangement.
Show coverage without confusing it with performance
Coverage describes what the portfolio can currently assess. Performance describes what the reviewed observations show against a defined expectation. Keep them separate.
For each intended outcome, list the grantees expected to report, those with usable evidence, the period covered and the outstanding gaps. An outcome with no data should be explicitly marked “not assessed” or “evidence missing.” A red label without explanation can be misread as poor performance.
If you use colors, provide text labels and documented criteria. “Evidence sufficient for this comparison” does not mean “outcome achieved.” Nor does one new indicator turn an outcome green automatically: it may be the wrong measure, the wrong period or too limited to support the claim.
Resolve the gap with a realistic next measurement
Identify what information would improve the next review. Make the proposed collection specific and proportionate rather than asking an assistant to produce a better grade.
- Which outcome or comparison cannot be assessed?
- What observation is missing, and why does it matter?
- Can an existing source supply it without another questionnaire?
- Who can collect or clarify it, and when?
- What will remain uncertain even after it arrives?
For an outcome that takes six months to observe, a new intake question does not provide a six-month result today. Agree the follow-up plan and show the current limitation. Use the reporting collection lesson to design the request.
A practical portfolio roll-up prompt
Supply the actual framework and source records. Replace the placeholders; a demonstration dataset name is not required.
Map the reviewed evidence for [portfolio and period] to [intended outcomes]. Use [dictionary version]. List contributing grantees and exact source references.
For each measure, state whether aggregation is valid based on definition, period, population and overlap. Show included and excluded values, the calculation and coverage. Keep incompatible measures separate. Do not add forecasts to observed results, average rates without their denominators or infer missing outcomes.
Separate evidence coverage from performance. Identify unresolved questions and a realistic next collection or review action. Do not infer that the grants caused the observed change. Produce a table a reviewer can verify before writing the summary.
Write the report after checking the roll-up
Lead with the decision the report informs, then show the result, represented population, exclusions and uncertainty. Name the grantees contributing evidence without calling them the causal drivers unless that claim is supported.
Use the How to Write an Impact Report guide for the narrative and report examples for presentation ideas. Keep the checked source table with the report so another colleague can reproduce the numbers.
Choose access deliberately. A shareable report does not need to expose participant records or restricted partner documents. Verify supported sharing settings and the intended audience before distributing a link; do not assume every report should open publicly without login.
Watch the portfolio reporting demonstration · 2 minutes 16 seconds
A Sopact demonstration of connected portfolio records and reporting. Use the rate and coverage exercise to test aggregation and traceability in your own setup.
What should you test in Sopact?
Use the fictional A/B/C exercise or an authorized portfolio sample. Check that the outcome map preserves the grantee, period and source, that incompatible follow-up windows stay separate, and that the combined rate uses the actual denominators.
Then test a correction and a missing observation. Confirm that the result can be revisited without losing its evidence or the reason for an exclusion. Configure definitions, review rules and sharing access with the team. The value is a repeatable, inspectable reporting process—not an automatic green grade.
Frequently asked questions
How do foundations aggregate outcomes across grantees?
Map results to intended outcomes, check comparability and overlap, then combine only compatible measures. Show source grantees, denominators, exclusions and missing coverage.
What is a portfolio roll-up to a Theory of Change?
It organizes evidence against the funder’s intended outcomes. Some observations can be aggregated; others provide context or remain separate. The map does not itself demonstrate causation.
Can we average grantee outcome percentages?
Use the underlying counts and compatible denominators when calculating a combined rate. A simple average of percentages gives equal weight to groups that may have very different sizes.
What if grantees report the same participants?
Resolve overlap when reporting unique people, using an authorized matching approach. Otherwise keep counts separate or clearly disclose potential double counting.
Should an outcome with no data be red?
Use an explicit missing-evidence label. If a color is used, explain that it represents coverage rather than an observed performance failure.
What should social reporting software show in a portfolio roll-up?
Test outcome mapping, comparable definitions, denominator-based calculations, overlap handling, coverage gaps and source traceability. Use the worked example to assess the tool rather than relying on a dashboard image. The Sopact portfolio solution is the next place to examine the collection and review workflow.
What makes a roll-up weak?
Combining incompatible observations, hiding missing data, double counting, losing source references or overstating causation. Correct the specific weakness rather than changing a presentation grade.
Bring the result back to your module
Continue to SROI across organizational levels. Bring the boundaries and overlap checks from this lesson before considering monetary values.