Read narrative, delivery, spending and outcome evidence together. Use a worked report to distinguish reported facts, real gaps and conclusions that need more context.
Read a grantee report by checking its narrative, activity figures, financial information and outcome evidence together against the grant’s agreed scope and reporting period. Extract the claims and sources, reconcile genuine inconsistencies, compare results with the relevant targets, then request only the missing information needed for the review. Keep “reported,” “checked” and “not yet known” distinct: a number in a report is not automatically proof, and a missing outcome may not yet be due.
By Sopact Academy · Reviewed September 12, 2026. This lesson combines cited public guidance with practical workflow recommendations. Figures and teaching scenarios are illustrative unless explicitly identified as a published case.
Grant Intelligence · Chapter 14 · Analyze
Bring: one report, its agreed targets and reporting schedule, the relevant budget version and supporting files. Use the collection and clarification practices from chapter 9.
Leave with: a claim-and-source table, a short list of real gaps, a scoped assessment and a focused clarification request. Riverside Works is a fictional workforce program used only for the exercise below.
This lesson is for grant and program officers who need to understand a report well enough to decide what to discuss, verify or act on. It is also useful for portfolio teams preparing summaries from multiple grantee updates.
The task is not to turn every report into a red-or-green judgment. It is to identify what the evidence supports, where interpretation is uncertain and what the responsible reviewer needs to do next.
The four lenses are a practical reading aid, not four mutually exclusive data types. Outcome evidence can be qualitative or quantitative. Financial figures are also quantitative. A participant interview may explain an outcome rather than simply decorate the narrative.
Use the distinctions in the OECD results-based management glossary to keep outputs and outcomes separate. Delivering a service and observing a change are different claims; neither, on its own, establishes that the service caused the change.
Work through four steps, letting the evidence determine the conclusion. Allow time to check source quality and resolve the questions the report raises.
| Step | Reviewer task | Useful output |
|---|---|---|
| Extract | Identify claims, definitions and sources | Claim-and-source table with uncertainty visible |
| Reconcile | Distinguish contradictions, omissions and different scopes | Focused issue list |
| Compare | Use agreed targets, budget and timing | Separate delivery, financial and outcome assessments |
| Clarify | Ask for the evidence needed to resolve the issue | One proportionate request with an owner |
For this exercise, the fictional report supplies the following statements. Treat them as submitted claims, not independently verified facts. No target, budget phasing or follow-up schedule is supplied yet.
Teaching source · fictional report
The source labels make the exercise reproducible. In a real report, use the document name, version and page, table or paragraph reference. Do not manufacture line numbers when the source has none.
Do not grade every claim with a number as strong evidence and every qualitative observation as weak evidence. A precise number can be incorrectly defined; a well-documented interview can provide useful evidence within its scope.
| Claim | Source | Initial status | What to check |
|---|---|---|---|
| 142 enrolled; 118 completed | S1–S2 | Reported; calculation possible | Same cohort and definition? Unique participants? Completion rule? |
| Confidence and job-readiness improved | S3 | Reported; method not supplied | Whose experience, which method and what time comparison? |
| $212,000 of $250,000 spent | S4 | Reported; scope and period need confirmation | Relevant budget version, supporting financial statement and basis? |
| Some moved into employment | S5 | Reported; count and follow-up coverage absent | Number, cohort, follow-up window and known-status denominator? |
| Six-month outcomes not included | S6 | Not reported | Were they required and due at this reporting point? |
Suitable status labels include reported, checked against source, clarification needed and not yet due. If your team uses color, show the text label too and define what it means. A green cell should not silently mean “impact proven.”
Using only the supplied report and agreed requirements, list each material claim with its source reference, period, population or financial scope, and any stated method. Distinguish reported information from checks you actually performed. List missing definitions and absent required evidence separately. Do not assume a numerical claim is verified or a qualitative claim is invalid. Do not invent source passages.
A contradiction requires incompatible claims about the same thing. If one table says 118 completers and another says 112 for the same cohort, period and definition, investigate. If one is cumulative and the other quarterly, the figures may both be correct.
An omission is different. A narrative about confidence without an employment count is not automatically contradictory. Confidence and employment are different outcomes. The issue may be that the report does not provide evidence for a separate employment claim.
For Riverside Works, record the employment count and measurement method as gaps. Check whether the six-month follow-up is due. Do not call the program off-track merely because a later outcome is not yet available.
Review the claim table. Separate incompatible statements from missing information and differences in period, population or scope. For a possible contradiction, identify both source references and explain the mismatch. For an omission, identify the requirement or claim that needs support. Do not treat absence as a contradictory statement.
The reported completion rate is 118÷142 = 83.1%, if the definitions and cohort match. That is the proportion of enrolled participants who completed. It does not mean 83.1% of the grant timeline has elapsed or 83.1% of the funded outcome has been achieved.
Reported spending is 212,000÷250,000 = 84.8% of the grant budget. It does not establish whether spending is on plan without the relevant time-phased budget and accounting basis. The similarity between 83.1% completion and 84.8% spending is not evidence that costs and outcomes are proportionate.
Supported conclusion from this source
Riverside Works reports 142 enrollments, 118 completions and $212,000 in spending. The completion and budget-use calculations can be reproduced from those figures.
We cannot yet judge progress against the agreed plan because the target, phased budget and outcome timing have not been supplied. Employment results require further definition and evidence. This is a limitation of the review material, not proof that the grant has failed.
Keep delivery, financial and outcome assessments separate where they differ. If a board needs one summary status, use the organization’s defined rule and show the underlying reason and uncertainty. Do not force a definitive status from incomplete evidence.
Compare the reported values with the supplied approved targets, reporting schedule and relevant budget version. Check matching definitions and periods first. Show the calculations and assess delivery, financial information and outcomes separately. If a required comparison input is missing, state that the comparison cannot yet be made. Do not infer an overall status from similar percentages.
Watch the explainer · 6 minutes 7 seconds
This Sopact video explains why sources, definitions and audience context matter even when the numbers are clean. Apply that lesson when judging what the submitted report can support and which context is still missing.
Watch on YouTube if playback is unavailable · Browse the video library
Check your own grant record before asking the grantee for material already held. Then request only what resolves the remaining issue. Avoid congratulating or criticizing performance before the evidence supports that judgment.
Draft example · review before sending
Thank you for the Riverside Works report. We have the enrollment, completion and spending figures. To interpret the employment statement, please confirm the number employed, the cohort and follow-up period, and how many participants have a known employment status.
Please also confirm whether the six-month follow-up is due under the agreed schedule. If it is not yet due, we will record that timing rather than request an early result. There is no need to resend the whole report. Please use [approved channel] by [agreed date], or contact [owner] if the information is unavailable.
Do not say that everything else is complete unless the relevant checks have been performed. Keep the request, reply and reviewer’s conclusion with the source record. Use the missing-data follow-up lesson for reminders and unresolved information.
In a configured Sopact workflow, AI can prepare the claim table and flag possible gaps as a report arrives, using the supplied context and definitions. This helps the reviewer start with organized evidence rather than opening an unconnected file.
The reviewer still needs to check extraction, interpretation and the appropriate action. Test complete reports as well as reports with known issues. Do not assume that only flagged reports need attention or that the same prompt guarantees identical outputs every time.
Keep changes in definitions, source versions and review configuration visible over time. That context is what makes the next report comparable; collecting a sequence of files alone does not establish a reliable trend.
Use the Riverside Works source. First supply a reporting schedule showing the six-month outcome is not yet due. Then test a second version where it was due and remains missing. Write the different next action for each.
Next add the actual completion target and phased budget from a fictional agreement you define for the exercise. Recalculate against those inputs. Keep your assumptions labeled so another reader does not mistake them for facts in the original source.
Pass condition: your assessment changes when relevant evidence changes, and it explains why. It never treats a number as automatically verified, an omission as automatically contradictory or a spending percentage as proof of impact.
Check the grant, period, agreed requirements and source versions. Then identify the narrative, delivery, financial and outcome claims with the definitions needed to interpret them.
No. A numerical claim still needs a clear definition, period, scope and suitable source check. Record whether it is reported or actually checked.
Not automatically. Its usefulness depends on the question, method and scope. A well-supported account can explain an experience; it should not be generalized beyond what the evidence supports.
No. An omission is absent information. A contradiction involves incompatible claims about the same scope and period. Check definitions before deciding which problem you have.
No. A participant completion rate and a share of budget spent measure different things. Compare each with its relevant target and timing instead of treating similar percentages as proof of proportional progress.
First check whether the outcome is required and due. Not-yet-due evidence differs from overdue or unavailable evidence, and none alone proves program failure.
AI can help prepare a source-linked review in a configured workflow. Human reviewers still check the evidence and interpretation, resolve uncertainty and decide the next action.
The next chapter examines variance against targets. Keep the claim table, matching definitions and unresolved questions so calculations are based on comparable evidence.
For the final output, use How to Write an Impact Report — the ebook and example reports. For the broader workflow, explore grant reporting or return to the Grant Intelligence course.
Bring one grantee report and its agreed targets. Explore a workflow that connects the evidence, definitions and reviewer’s next action.
Explore Applications & Grants →
