Reporting asks
For nonprofit leaders
Bring the question your current reporting cannot answer.
Whether the question comes from your board, a funder, a program redesign, or a possible expansion, Prova helps your team work out what you can responsibly say now and what—if anything—you genuinely need to learn next. Prova combines frontier AI with proprietary evidence systems and accountable professional judgment. The systems connect important statements to sources and make gaps visible for professional review.
You do not need a perfect data system or a new evaluation plan before starting.
Recognizable situations
The next decision is often already visible.
- 01
A board or funder is asking a question no single report was designed to answer.
- 02
You need to decide whether to continue, redesign, expand, or stop something.
- 03
Reporting takes substantial staff time without making the next decision clearer.
- 04
Your program has changed, but its theory of change or measurement has not caught up.
- 05
You think a result is real but are unsure how strongly the current evidence supports the claim.
Reporting and decisions
Reporting describes what happened. A decision asks what to do next.
A decision asks
What should we continue, change, expand, or stop—or what may we say?
What you can honestly say to the board is also a decision. Your current reporting may already contain useful evidence. Prova helps determine which questions it can answer, where the answer stops, and whether more work would actually improve the decision.
Common questions
The work follows the question.
You do not need to know which category your problem belongs to. Prova recommends the smallest useful piece of work around the decision.
What can we actually say?
- Read across reports, data, feedback, research, and staff knowledge
- Check where a claim holds and where it becomes uncertain
Evidence synthesis and claim checking
How is our program meant to create change?
- Clarify the theory of change
- Separate outside research from assumptions the program still needs to test
Theory of change and program assumptions
What should we measure—and what can we stop collecting?
- Work backward from the decisions the information must serve
- Keep measures that help your team learn or decide, and identify which others can be reduced or retired within your reporting requirements.
Measurement design
Are we ready for the next move?
- Test readiness to continue, redesign, expand, or stop
- Identify the evidence that would materially change the decision
Program evaluation and readiness
What the work can look like
A useful answer stays with the files behind it, and with the limits that matter.
These worked examples show the form of possible work. They are not client work and do not describe a real program.
Claims & SourcesWhat can we responsibly say?Which statements are supported, which are only partly supported, and where the current files are silent.Possible output: A map connecting material claims to the sources and limits behind themView the illustrative Claims & Sources exampleClose example
Illustrative example — not client work
Claims & Sources Map
Claims and Sources MapIllustrative
What the current material supports
Every claim the organization already makes, set against what is actually behind it.
| The claim | Support | What it rests on |
|---|---|---|
| 85% of enrolled participants complete the program | strong | Attendance logs and program files, 2022–2024 |
| Employer partners value the placements | partial | Three partner interviews; no wider follow-up |
| Participants leave with more confidence | thin | Described in staff notes; never measured directly |
| The program leads to lasting employment | absent | No contact with participants after they exit |
absentthinpartialstrong
Program LogicHow is the program expected to create change?Make the theory of change and its assumptions explicit, including what outside research supports and what must still be learned locally.Possible output: A research-linked theory of change with open assumptions clearly markedView the illustrative Program Logic exampleClose example
Illustrative example — not client work
Research-Linked Theory of Change
Research-Linked Theory of ChangeIllustrative
Which links outside research supports—and which the program still needs to test
A fictional research review showing how support can differ between links in a program’s theory of change.
Young people complete a 12-week training placement
strong
In this fictional example, the reviewed research supports skill gains.
They gain skills employers are hiring for
partial
Whether those gains lead to job offers remains uncertain and needs to be examined in the local labor market.
They find steady work within six months
thinMust be learned locally
In this fictional example, few reviewed studies follow households this far, and none are from a comparable setting.
Their households become more financially stable
absentthinpartialstrong
MeasurementWhich measures are worth the staff time?Keep the measures that inform a real decision, and make collection burden visible alongside decision value.Possible output: A practical plan for what to measure, how to use it, and what the team can sustainView the illustrative Measurement exampleClose example
Illustrative example — not client work
Measurement Architecture
Measurement ArchitectureIllustrative
Which measures earn the time they take
Current and possible measures, placed by staff time and how much they could inform a decision. Illustrative relative ratings from 0–100, based on the fictional team's judgment.
Collected todayCandidate to stopMissing, worth adding
The same measures as a table
| Measure | Staff time rating | Decision value rating | Status |
|---|---|---|---|
| Monthly attendance | 20 | 70 | Collected today |
| Employer feedback survey | 33 | 80 | Collected today |
| Exit interview | 54 | 68 | Collected today |
| Staff case notes | 66 | 58 | Collected today |
| Satisfaction star rating | 18 | 22 | Collected today |
| Intake demographics form | 42 | 34 | Collected today |
| Weekly activity log | 72 | 11 | Candidate to stop |
| Quarterly funder narrative | 90 | 20 | Candidate to stop |
| Six-month follow-up | 62 | 94 | Not collected today |
See a full Claims & Sources Map and what you receive
Explore a Measurement & Learning Plan
Follow a worked example from a draft claim through the files already kept
What the work should leave behind
Leave with a decision—and a clear view of what, if anything, is worth learning next.
Make the strongest case the evidence supports, with the important limits clear. A limited claim is not a verdict on the program. The next step may be more work. It may also be to stop collecting, narrow the claim, or decide that nothing further would help.
Direct and foundation-funded work
Keep control of the question, the materials, and what gets shared.
Your team sets the question and decides how the work will be used.
Prova works directly with your materials, program, and decision.
A foundation may fund access without owning your question.
Your organization chooses whether to participate, sets the question and materials, and controls what may be shared.
AI and trust
AI helps Prova examine more of the material your team already has.
Important conclusions remain connected to sources and are reviewed by accountable people.
A practical first step
Bring one decision your team needs to make.
A first conversation can clarify the answer you need, the information that already exists, and whether Prova is the right fit.
Discuss a nonprofit decision