Read a Grantee Report Across Qual, Quant, Financial & Social | Sopact Academy

A grantee report is four reports in one — a narrative, a set of numbers, a budget, and an outcomes story — and they rarely agree. This chapter reads all four on arrival, grades every claim by the evidence behind it, flags what's missing or contradictory, and produces one variance-ready summary — so you catch risk mid-grant instead of at the final evaluation.

Who this is for

Grant and program officers who receive dozens of grantee reports a season and can't read every page of every one — and who get burned when a problem that was visible in month four only surfaces in the final report. If you monitor grantees against targets and owe your board an honest account, this is the read that makes it possible.

What "reading across four signals" means

Most review reads a grantee report once, for the story. But a report makes four kinds of claim, and they fail differently:

• Qualitative — the narrative: what the grantee says happened.
• Quantitative — the numbers: participants served, targets, rates.
• Financial — the money: budget vs actual, spend against milestones.
• Social — the outcomes: what changed for people, and the evidence for it.

The risk lives in the gaps between them — a confident narrative with no numbers, a target hit on paper but underspent, an outcome claimed with no measure. Reading the four together, on arrival, is what turns a report from a document you file into a signal you act on.

The process — four prompts, about 30 minutes

You paste the report once; four prompts read it from four angles and reconcile. Each returns something you can see. Run in order; re-run on the next report unchanged.

Worked example throughout: the annual report from Riverside Works, a grant-funded workforce program.

Step 1 · Extract — the graded four-signal table

Paste the report. This prompt reads only what the text states and returns one row per claim, tagged by signal type and graded by a fixed rule — GREEN when a quotable figure backs it, AMBER when stated without a figure, RED when it's expected but absent. Because the rule is mechanical, the table comes out the same every run.

Read the grantee report in the SOURCE using ONLY what it states: [PASTE REPORT]. Extract every claim into one row, tagged by SIGNAL — Qualitative, Quantitative, Financial, or Social. For each: the shortest exact phrase from the source, any figure in a Figure column, and a Grade — GREEN if stated with a quotable figure (number, %, rate, $, timeframe, defined population); AMBER if stated with no figure; RED if a signal expected for this grant type is not stated. Quote the source line for each row. Never invent claims; if a signal has none, write "Not stated". Output: Signal | Claim | Figure | Grade | Source line. The same source must give the same table every run.

Step 2 · Reconcile — contradictions & gaps

The table is the evidence; this prompt cross-checks the four signals against each other and names where they don't line up — the places a single-pass read misses.

Using the graded table above, cross-check the four signals. List every CONTRADICTION (one signal implies something another denies or omits) and every GAP (a claim with no supporting signal). For each, name the two signals involved and quote both source lines. Rank by materiality to the grant's stated outcome. Do not infer beyond the source.

Step 3 · Variance — the on-track / at-risk / off-track call

Now compare to what the grant promised. This prompt places the report against targets and budget and returns one classification with its reason — the line your board and your monitoring log both need.

Compare the graded table to the grant's TARGETS and BUDGET: [PASTE TARGETS]. Classify the grant on-track / at-risk / off-track, with the single deciding reason and the evidence line. Separately state budget status (spent vs elapsed) and whether spend and outcomes are proportionate. One classification only; cite the source.

Step 4 · Clarify — one request, not a reload

The report isn't wrong; it's incomplete. This prompt drafts a single, specific clarification — only the fields that are missing or contradictory — so the grantee closes the gap without re-reporting everything.

Draft a short, warm clarification email to the grantee requesting ONLY the missing/contradictory items from the reconcile list — nothing already provided. For each ask, say why it matters in one line. Keep it under 150 words; funder-to-grantee tone, not an audit.

What you have at the end

Four prompts, and you hold a complete read of one grantee report: a graded four-signal table, the contradictions and gaps named, one honest variance call, and a single clarification already drafted — all cited to the report's own words, all repeatable on the next report without redoing anything. The problem that used to surface at the final evaluation now surfaces the week the report lands.

Want this on every report automatically?

Sopact Sense runs these four reads the moment a grantee report arrives, flags the at-risk ones, and drafts the clarification — so you read the exceptions, not the pile.

Try it in Sopact →

Next: Monitor Progress Against Targets + Real-Time Risk Alerts → — turn this per-report read into a live watch across your whole portfolio.