Sopact is a technology based social enterprise committed to helping organizations measure impact by directly involving their stakeholders.
Copyright 2015-2026 © sopact. All rights reserved.
a blended report from the same records that proved your outcomes — social impact beside unit economics, classified on the IMP Five Dimensions with the thin dimension flagged, and a path to self-funding whose assumptions an investor can pressure-test.
For: social enterprises and impact-driven teams reporting to impact investors, lenders, and boards — anyone whose capital expects outcomes and a business model.
Why: an impact investor asks two questions at once — did it work, and does the model sustain itself? A funder report answers only the first, and traditional case management answers neither, because it has no view of the financial side at all.
Outcome: a blended report from the same records that proved your outcomes — social impact beside unit economics, classified on the IMP Five Dimensions with the thin dimension flagged, and a path to self-funding whose assumptions an investor can pressure-test.
This is Chapter 10b of the Case Intelligence series — the capstone’s second half. In Chapter 10a the RiseWorks cohort became a funder impact report: outcomes against targets, every claim cited and graded, cohort close as a query. This chapter re-queries the identical records with an investor lens. Not a new dataset, not a parallel tracking system — the same eight stores, the same persistent IDs, the same on-arrival scores, read for a different audience. That is the capstone’s punchline, and this chapter exists to make it concrete.
The running example is unchanged: RiseWorks Foundation / Pathways 2027, now read through its earned-revenue side. RiseWorks charges employers placement fees, so the cohort that produced 29 placements also produced fee revenue of roughly $282k earned / $185k collected, with about $27k clawed back on placements that did not hold. Those numbers sit on the same placement rows as the wage gains and confidence scores — which is the whole story of this chapter.
As in every chapter, each step is tagged [DIY] or [SENSE]. Structuring the investor report and choosing your unit-economics metrics is thinking work for any chat window. Blending social and financial value across a cohort, and modeling self-funding from observed fee data, is what the product does over the stores — a standalone prompt cannot blend records it never received, and it cannot reconcile two systems it does not hold.
A grantmaker asks whether the program changed lives. An impact investor asks that, then asks the question a funder never raises: does the model pay for itself, or will it always need subsidy? To an investor, a beautiful outcome with no path to sustainability is a grant in disguise — fundable, perhaps, but not investable.
Most reporting stacks cannot answer the two questions together, because the two answers live in different systems. The outcomes sit in a case-management tool; the fee revenue sits in accounting; the clawbacks sit in a spreadsheet someone keeps by hand. At report time the impact deck and the finance deck get built separately, and when an investor notices that the placements in one do not reconcile with the fees in the other, the diligence conversation stops being about the program.
The Case Intelligence answer is structural: the money lives on the same records as the outcomes. A placement row holds the outcome — a durable, retained placement, joined by requisition_id and email to everything Chapters 1 through 9 collected — and the economics: the fee earned, whether it was collected, whether retention held or the fee clawed back. The fee-at-risk flag that Chapter 9’s Intelligent Cell set on every match was this chapter’s data being born. There is no separate finance export because there is no separate finance dataset. Switching from the funder report to the investor report is a different query, not a different system.
The investor lens, in six sections: blended value (social and financial, side by side) · unit economics (fees earned vs collected, cost-recovery, net cost per placement, clawback) · SROI / return (carried from Chapter 7) · the IMP Five Dimensions (What / Who / How Much / Contribution / Risk) · sustainability and the path to self-funding · scale (what growth does to the economics).
What you do. Draft the report skeleton against the six sections above, in any capable AI, before touching data. Keep the same discipline as the funder outline: sections answer investor questions, and the honesty rule — measured value labeled separately from benchmarked — applies to the financial figures exactly as it applied to the SROI.
What you get. A fixed outline where each section names the question it answers and the fields it needs — so you discover before close whether a section’s data was ever collected.
Why it matters. The investor report fails in a specific way the funder report does not: it makes financial claims, and financial claims get audited harder. An outline that ties every section to named fields is the difference between “fee revenue of $282k” as a queryable fact and as a number someone remembers differently in the board meeting.
Real example — RiseWorks. The outline that came back maps each section to the placement-row fields: blended value → outcome fields plus fee fields; unit economics → fee earned, fee collected, retention checkpoints, program cost; Five Dimensions → the evidence already graded in 10a plus the risk fields; self-funding → collection rate and clawback observed, not assumed. One flag raised in drafting: “scale” had no field of its own — it is a model over the others, and the outline says so rather than promising data that does not exist.
Prompt 1: Investor Report Outline
PROMPT 1 — INVESTOR REPORT OUTLINE [DIY]
Chapter 10b · Case Intelligence series · Sopact Academy
Use in: any capable AI (Claude, ChatGPT). You structure this once, by hand.
Purpose: shape a social-enterprise investor / board report around the two
questions impact capital asks — did it work, and does the model sustain
itself?
------------------------------------------------------------
You are an impact-and-finance reporting editor for a social enterprise.
Your job is to structure an investor report where every section ties to
named data fields — financial claims get audited harder than social ones.
TASK
From my program, revenue model, and capital-stack description below, build
a report outline using ONLY these six candidate sections (drop any that do
not apply; add none):
1. Blended value — social outcomes AND financial value, side by side, from
the same records.
2. Unit economics — fee revenue earned vs collected, cost-recovery ratio,
net cost per outcome, clawback.
3. SROI / return — social value per dollar (carried in from an existing,
sourced calculation — never re-derived for the report).
4. IMP Five Dimensions — What / Who / How Much / Contribution / Risk, each
rated with evidence, the thinnest flagged rather than padded.
5. Sustainability & path to self-funding — where fees cover cost vs where
subsidy remains, modeled on observed rates.
6. Scale — what growth does to the economics (a model over the other
sections, not a data section — say so).
For each section you keep, write one line naming: the investor question it
answers, and the FIELDS it requires (e.g. fee earned, fee collected,
retention checkpoints, program cost, outcome scores).
HONESTY RULE (apply to every section): measured values stay labeled
separately from benchmarked ones — the same rule the SROI uses.
INPUT — paste between the markers:
<<<
[describe your program, your revenue model (who pays fees, when, on what
terms, clawback conditions), and your capital stack (grant, debt, equity,
board reporting)]
>>>
OUTPUT FORMAT
A. The kept sections, in order: Section | Investor question | Fields
required.
B. The dropped sections, each with the one-line reason.
C. FIELD GAPS — any required field my records do not currently collect,
listed plainly (these are fixable mid-cohort; at close they are not).
RULES
- No section may be invented beyond the six candidates.
- Every kept section must name its fields; no section may rest on data
that would need a separate finance system to produce.
- The same input must give the same output every run.
------------------------------------------------------------
What you do. Different capital wants different numbers: a revenue-based lender watches cost-recovery, an equity investor watches net cost per outcome and the slope toward self-funding, a board watches clawback because it measures placement durability. Choose the metrics your investors will read first — and for each, name the field it needs and whether it leads or lags sustainability.
What you get. A short metric → fields → leading/lagging table, which doubles as a data-collection audit: any metric pointing at an uncollected field is a gap you can still fix mid-cohort.
Why it matters. Choosing metrics up front is what lets the rollup compute them from data already flowing, instead of triggering the scramble this series exists to end. And the leading/lagging split keeps the report honest about time: collection rate predicts next cohort’s sustainability; clawback confirms what happened to this one.
Real example — RiseWorks. The chosen set:
Every field in the middle column already exists on the placement rows — nothing here demanded a new system, only that the fee fields live where the outcome fields live.
Prompt 2: Unit-Economics Selector
PROMPT 2 — UNIT-ECONOMICS SELECTOR [DIY]
Chapter 10b · Case Intelligence series · Sopact Academy
Use in: any capable AI (Claude, ChatGPT). Different capital watches
different numbers — choose yours before close, while field gaps are still
fixable.
Purpose: pick the unit-economics metrics your investors read first, with
the field each needs and its leading/lagging role.
------------------------------------------------------------
You are helping me choose the unit-economics metrics an impact investor in
my model actually reads. The output doubles as a data-collection audit.
TASK
From my revenue model and capital source below:
1. Select the metrics that matter for MY capital stack, drawing only from
this fixed menu (no invented metrics):
- fee revenue, earned vs collected
- cost-recovery ratio (fees collected / program cost)
- net cost per outcome (program cost minus fees collected, per durable
outcome)
- clawback (fees lost to broken retention)
- fee yield per placement
- collection rate (collected / earned)
2. For each selected metric, name the FIELD(S) it needs — e.g. fee earned,
fee collected, retention checkpoints, program cost, placements — so it
computes from records already collected, not from a scramble later.
3. Classify each metric LEADING (predicts sustainability — e.g. collection
rate) or LAGGING (confirms it — e.g. clawback), and say why in one
line.
4. Flag any metric whose required field my records do not collect —
FIELD GAP, fix mid-cohort or drop the metric; never estimate it at
close.
INPUT — paste between the markers:
<<<
[describe your revenue model — who pays, how much, when, clawback terms —
and where your capital comes from: lender, equity, board, blended]
>>>
OUTPUT FORMAT
One table, exactly: Metric | Field(s) needed | Leading / Lagging | Why this
investor cares — followed by a FIELD GAPS list.
RULES
- Metrics come only from the fixed menu; fields must be named per metric.
- Leading/lagging is assigned by the mechanical test: does the metric
predict next cohort's sustainability, or confirm this one's?
- The same input must give the same output every run.
------------------------------------------------------------
Here is the record that makes the whole two-reports claim concrete. Marcus D. — the strong match Chapter 9 scored, placed as a MIG Welder Apprentice — has one placement row, and both reports read it:
Join keys: email (persistent participant ID) · requisition_id · employer_name
Outcome fields (the funder report read these): credential AWS D1.1, earned at exit · placement accepted, start by Nov 15 · retention checkpoints · wage at six-month follow-up · confidence trajectory on the joined waves
Finance fields (this report reads these): fee earned per the employer agreement · fee collected (or aging) · clawback trigger if retention breaks · fee-at-risk flag — set LOW by Chapter 9’s Intelligent Cell, from his match band and retention signals
Rolled up across all 29 placements: fees earned ≈ $282k · collected ≈ $185k · clawed back ≈ $27k
Nothing was collected twice. The wage gain that proves impact and the fee that proves the model belong to the same person, on the same row, joined the day the placement was made. When 10a rolled this row up, it read the left-hand fields; this chapter reads both.
From here on, this is product output, not a prompt you run. A chat prompt could compute a cost-recovery ratio from two numbers you paste — but it cannot keep 29 placements’ fee statuses current, and it cannot notice that a retention checkpoint failing today turns an earned fee into a clawback. Sense can, because the finance fields are store fields, scored like everything else.
Two mechanisms. The Intelligent Cell on each placement row keeps the economics honest on arrival: as retention checkpoints land, it updates the fee status (earned → collected, or earned → clawed back), and it maintains the fee-at-risk flag so projected revenue never silently includes shaky placements. The Intelligent Row assembles the Placement Economics Profile — one report per placement holding the outcome story and the money story together: this person, this credential, this wage trajectory, this fee, this risk.
Marcus’s Row reads the way an investor wishes all evidence read: a strong-band match with a low fee-at-risk flag, a credential on record, retention holding — an outcome and a receivable, one record. And the aggregate honesty falls out of the per-record honesty: the $282k earned versus $185k collected gap is not an accounting embarrassment discovered at close; it is the live sum of rows whose fee status says aging. The $27k clawback is the sum of rows where retention broke. Nobody reconciles anything, because nothing was ever separate.
Prompt 3 — The Blended Record
PROMPT 3 — THE BLENDED RECORD [SENSE]
Chapter 10b · Case Intelligence series · Sopact Academy
This is product configuration — it runs on arrival in Sopact Sense, not in
a chat window. The blend is not computed at close; it exists because the
fee fields live on the same records as the outcome fields, and both are
scored as they land.
------------------------------------------------------------
WHERE IT SITS
An Intelligent Cell on each placement row in the Placements store. The row
joins the rest of the workspace by reference keys: email (persistent
participant ID), requisition_id, employer_name — so one row reaches the
credential (Exit/Completion), the wage trajectory (Six-Month Follow-up),
and the fee terms (Employer Accounts).
THE CELL'S INSTRUCTION
On each placement row, as records land:
1. Track the FEE STATUS through its lifecycle: earned (placement made) →
collected (payment received) or aging (earned, not yet collected).
2. When a retention checkpoint fails per the fee terms, move the fee to
CLAWED BACK and record the trigger.
3. Maintain the FEE-AT-RISK flag (initialized by Chapter 9's match Cell
from band and retention signals) so projected revenue never silently
includes shaky placements.
4. Keep MEASURED financial facts (fees on record) labeled separately from
any BENCHMARKED or modeled values — the SROI honesty rule, applied to
money.
THE ROW'S ASSEMBLY (Intelligent Row — Placement Economics Profile)
One report per placement, both lenses on one record:
outcome side: credential · placement and start date · retention
checkpoints · wage at follow-up · confidence trajectory
finance side: fee earned · fee status (collected / aging / clawed back)
· fee-at-risk flag · program cost allocation
SAMPLE RETURN (real record — Marcus D. · MIG Welder Apprentice placement)
outcome: AWS D1.1 earned at exit · strong-band match (Chapter 9) ·
start by Nov 15 · retention holding
finance: fee earned per the employer agreement · fee-at-risk: LOW ·
no clawback trigger
→ an outcome and a receivable, one record.
WHAT THE PER-RECORD HONESTY PRODUCES IN AGGREGATE (RiseWorks, 29 placements)
fees earned ≈ $282k · collected ≈ $185k (the gap = rows whose status
says aging) · clawed back ≈ $27k (rows where retention broke).
Nobody reconciles anything at close, because nothing was ever separate.
WHY THIS CAN'T BE A CHAT PROMPT
A chat prompt can divide two numbers you paste; it cannot keep 29 fee
statuses current, notice a retention checkpoint failing today, or hold the
join between a wage gain and the fee the same placement earned. The blend
is a property of the store, not of a prompt.
------------------------------------------------------------
The close, again — same mechanism as 10a, different questions. The Sopact Assistant (with the Claude MCP connection) reads the same stores and returns the investor report as answers.
The blended rollup. Social and financial value from one query, measured kept separate from benchmarked:
The wage gain that anchors the social row and the fees that anchor the financial row come from the same 29 people. That is what “blended” means when it is real: not two analyses stapled together, one dataset read twice.
The Five Dimensions, rated with evidence. Investors read impact through the IMP’s five questions, so the Assistant classifies the cohort that way — and flags the thin dimension instead of padding it:
A thin dimension named is a diligence item; a thin dimension hidden is a red flag when they find it — and they find it.
The path to self-funding. This is the answer to the investor’s second question, and the arithmetic runs on numbers already in this article. Cost-per-outcome of ≈ $20,076 across 29 placements puts the cohort’s cost in the neighborhood of $580k; $185k collected against it means the fee model already funds roughly a third of a cohort — the rest is the subsidy still required, stated as a number instead of an aspiration. The levers are explicit, and each is an assumption an investor can test rather than a promise:
A model built from observed collection and clawback rates, with its assumptions flagged, reads as credible. The same model built from hoped-for rates reads as a pitch — and investors have seen the pitch.
Prompt 4 — Blended Rollup and Sustainability Model
PROMPT 4 — BLENDED ROLLUP AND SUSTAINABILITY MODEL [SENSE]
Chapter 10b · Case Intelligence series · Sopact Academy
Run this in the Sopact Sense Assistant over your store. Same mechanism as
the Chapter 10a close, different questions — the investor report comes
back as answers from the same records.
------------------------------------------------------------
ASSISTANT PROMPT A — the blended rollup
"Roll this cohort up as a blended social + financial report from the
placement rows and their joined waves:
1. SOCIAL value — the measured outcomes (confidence, wage, employment)
with the funnel.
2. FINANCIAL value — fee revenue earned vs collected, clawback, and
cost-recovery from fees collected against program cost.
3. Carry the SROI in from its sourced calculation; keep MEASURED value
labeled separately from BENCHMARKED throughout.
Reconcile nothing — both lenses read the same records."
WHAT IT RETURNS (RiseWorks):
Social (measured): confidence 4.3 → 7.4 · wage $9.96 → $25.11 median ·
employment 82% completers vs 45% non-completers · funnel
80 → 62 → 58 → 29.
Financial: fees ≈ $282k earned / ≈ $185k collected · clawback ≈ $27k ·
cost-recovery from fees collected against program cost.
Return: SROI ≈ 2.44:1 (Chapter 7), measured separate from benchmarked.
The social row and the financial row come from the same 29 people.
ASSISTANT PROMPT B — the Five Dimensions
"Classify the cohort's impact on the IMP Five Dimensions — What / Who /
How Much / Contribution / Risk. Rate each dimension with the evidence
behind it, and FLAG the thinnest dimension rather than padding it."
WHAT IT RETURNS (RiseWorks):
What durable living-wage placements in skilled trades EVIDENCED
Who youth 18–29 · ~22% justice-involved · ~55% first-
generation · coverage tracked EVIDENCED
How Much 29 placed · wage roughly doubled · confidence
+3.1 — modest scale, real depth EVIDENCED
Contribution counterfactual attribution captured at exit EVIDENCED
Risk clawback ≈ $27k · demand gap 25 unfilled vs 29
surplus FLAGGED (thin)
ASSISTANT PROMPT C — the sustainability model
"From the observed fee and cost data: compute cost-recovery today, state
the subsidy gap, and model the path to self-funding through the explicit
levers — collection rate (earned vs collected), clawback via retention,
cohort size. FLAG every assumption so an investor can pressure-test it.
Use observed rates only; never aspirational ones."
WHAT IT RETURNS (RiseWorks):
Cost-recovery today: with cost-per-outcome ≈ $20,076 across 29
placements (cohort cost in the neighborhood of $580k), the ≈ $185k
collected funds roughly a third of a cohort; the remainder is the
subsidy still required, stated as a number.
Levers, each with its assumption flagged:
- close the earned–collected gap (≈ $282k vs ≈ $185k — a collections
lever visible per-employer in the store)
- cut the ≈ $27k clawback via 90-day retention, targeted by the
fee-at-risk flags before placements break
- grow cohort size — only if collection and retention hold at scale.
------------------------------------------------------------
Running impact and finance as two systems. Two decks that do not reconcile is the fastest way to turn a diligence call into an audit. Put the fee and cost fields on the same records as the outcomes — that one design decision is most of this chapter.
Reporting earned as collected. $282k earned and $185k collected are different facts, and an investor will find the gap in your receivables even if your report hides it. Report both, and let the gap be the collections lever it actually is.
Padding the thin dimension. Every program has one — for RiseWorks it is Risk. Padding it with prose signals either that you do not know your weakness or that you hope the reader will not notice. Flag it, attach the numbers ($27k clawback, 25-vs-29 gap), and make it a managed item.
Burying the clawback. Fees lost to early exits feel like failure, so they vanish from decks — which converts a retention metric into a trust problem. The clawback is placement durability priced in dollars; report it and show the lever against it.
Modeling self-funding on aspirational rates. A sustainability model built on the collection rate you intend to have is fiction with a spreadsheet. Build it on the observed rates in the store, flag every assumption, and let the investor stress-test the levers.
A social-enterprise investor report queried from the same records that produced the funder report: blended social and financial value with measured kept separate from benchmarked, unit economics computed from fields on the placement rows, the cohort classified on the Five Dimensions with the thin one flagged, and a self-funding model whose levers — collection, retention, scale — carry observed numbers and named assumptions. And you have the capstone’s proof in hand: one Case Intelligence workspace, two audiences, two reports, zero re-collection.
Because the fee and cost fields were scored on arrival on the same records as the outcomes, the blended view is retrieval, not reconciliation — the impact deck and the finance deck cannot disagree, because they are the same data. Because measured stays labeled apart from benchmarked, the financial claims survive the same diligence the social claims do. And because the sustainability model runs on observed collection and clawback rates with assumptions flagged, the answer to “does this sustain itself?” is a testable model instead of a hope. The skeptic’s one-liner: one dataset, two reports — the funder saw the change; the investor sees the change and the model, from the same records.
Add fee and cost fields to the records you already score for outcomes — fee earned, fee collected, program cost allocated, a retention-linked clawback flag. That single schema change is the whole difference between a program that can produce one report and one that can produce both. The queries come later; the fields have to start filling now.
A screen-by-screen walkthrough — the placement row with both lenses visible, the blended rollup coming back in one query, and the self-funding model with its levers live — is in production. Check back on the Academy.
Social enterprises whose board deck and impact deck are built by different people from different exports. Workforce programs with earned revenue who suspect their fee model funds more of the work than anyone can currently prove. Impact-first teams preparing for investment diligence and dreading the question “do these two reports reconcile?” If your outcomes and your economics have never lived on the same record, this is the fix.
Write your investor report in Sopact Sense — sopact.com/academy.
Next in the series: the loop closes. Take the chain you have watched RiseWorks run — How to Build a Theory of Change through both capstone reports — back to Chapter 0 and Chapter 1, and run it on your own program: stand up one store, prove it in an afternoon, extend.
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact