play icon for videos
Agents

Impact Measurement & Management (IMM) for Portfolios

Run IMM from due diligence to LP reports on one platform

US
By Unmesh Sheth
·
11
min read
<!DOCTYPE html><html><head><meta charset='utf-8'><style>*{margin:0;padding:0;box-sizing:border-box}html,body{background:transparent}body{font-family:Inter,Arial,sans-serif;color:#1A1915}#wrap{padding:2px}.kick{font-size:11px;letter-spacing:.12em;text-transform:uppercase;color:#76716A;font-weight:600;margin-bottom:12px}.tabs{display:flex;gap:5px;margin-bottom:14px;white-space:nowrap}.tab{border:1px solid #D8D2C4;background:#FAF9F5;border-radius:999px;padding:6px 10px;font-size:11.5px;font-weight:600;cursor:pointer;color:#3D3A33;flex:0 0 auto}.tab.on{background:#1A1915;color:#FAF9F5;border-color:#1A1915}.card{background:#FAF9F5;border:1px solid #E3DFD3;border-radius:14px;padding:22px}.st{font-size:11px;color:#C96442;font-weight:700;letter-spacing:.08em;text-transform:uppercase;margin-bottom:8px}.t{font-family:Georgia,'Times New Roman',serif;font-size:24px;line-height:1.2;margin-bottom:10px;font-weight:500}.d{font-size:14px;line-height:1.55;color:#3D3A33;min-height:44px}.bar{display:flex;gap:5px;margin-top:16px}.seg{height:5px;flex:1;border-radius:3px;background:#E3DFD3;cursor:pointer}.seg.on{background:#C96442}</style></head><body><div id='wrap'><div class='kick'>IMM lifecycle · one investee record</div><div class='tabs' id='tabs'></div><div class='card'><div class='st' id='st'></div><div class='t' id='t'></div><div class='d' id='d'></div><div class='bar' id='bar'></div></div></div><script>var S=[{n:'Diligence',t:'Score the thesis',d:'Investee ID minted; Five Dimensions rubric scored from the deal documents with citations.'},{n:'Onboard',t:'Bind the data dictionary',d:'Theory of change confirmed with the investee; data dictionary bound and baseline deployed.'},{n:'Collect',t:'Evidence lands on one ID',d:'Quarterly narratives, financials, and surveys all join the same investee record on arrival.'},{n:'Learn',t:'Re-score and surface drift',d:'Five Dimensions re-scored on new evidence; drift and missing data surface mid-cycle.'},{n:'Report',t:'Reports draft as views',d:'LP letter, board pack, and IRIS+ aligned filing generate as views on the record, not projects.'}];var i=0;function ph(){var w=document.getElementById('wrap');parent.postMessage({ucxWidgetHeight:w.offsetHeight+4},'*');}function r(){var tb=document.getElementById('tabs');tb.innerHTML='';S.forEach(function(s,j){var b=document.createElement('div');b.className='tab'+(j==i?' on':'');b.textContent=(j+1)+' · '+s.n;b.onclick=function(){i=j;r()};tb.appendChild(b)});document.getElementById('st').textContent='Stage '+(i+1)+' of 5';document.getElementById('t').textContent=S[i].t;document.getElementById('d').textContent=S[i].d;var bar=document.getElementById('bar');bar.innerHTML='';S.forEach(function(s,j){var g=document.createElement('div');g.className='seg'+(j<=i?' on':'');g.onclick=function(){i=j;r()};bar.appendChild(g)});ph();}r();window.addEventListener('load',ph);setInterval(function(){i=(i+1)%S.length;r()},5000);</script></body></html>

What is impact measurement and management (IMM)?

Impact measurement and management (IMM) is the discipline of designing an investment or program around intended outcomes, collecting evidence of change as it arrives, and using that evidence to manage — not just report. In an impact fund it spans the full lifecycle: scoring the thesis at diligence against the Five Dimensions of Impact, binding a theory of change at onboarding, re-scoring each quarter, and producing LP, board, and compliance reports from one continuously learning investee record. Measurement is the input; management is the point.

IMM stands for Impact Measurement and Management. The same letters carry unrelated meanings elsewhere — the International Monetary Market in derivatives, a medical abbreviation — but in impact investing and philanthropy, IMM always refers to measuring outcomes and managing against them.

The learning happens the moment the document arrives — or it doesn't happen at all

Every serious fund references the frameworks. The difference between funds that build value from IMM and funds that merely file it is not the framework they chose — it is when the document gets read. A quarterly narrative read in week one of the quarter is management. The same narrative read during year-end assembly is archaeology. What separates the two is whether the document is tested against the framework on arrival.

On the arrival posture, a portco quarterly lands and its fields map to the data dictionary the same day; the open narrative is themed, the financials parse against last quarter, and an audit report reconciles against the investee's own prior submissions so discrepancies flag rather than file away. Missing data goes back to the investee through a unique link while the quarter is still open. An outlier metric — or a narrative signal like fundraising fatigue — flags for a check-in call before it becomes a write-down. By quarter end there is nothing to assemble: the LP letter, board pack, and IRIS+ filing draft themselves from the record, and the reviewer edits rather than assembles.

On the year-end posture, documents accumulate in folders, the analyst re-reads everything in week eleven, re-keys scores into a master spreadsheet, and reconciles against last quarter from memory. The cleanup is the work and the LP letter is its residue. The whole argument of this page is that the second posture is a choice, not a constraint — and that the choice is made by the plumbing, not the framework.

Impact management became impact intelligence — and the legacy stack didn't follow

Three tools dominate impact measurement today, and each one turned from asset into liability at the same seam: none of them connect to the evidence. Framework templates — GIIN, the IMP's Five Dimensions, published SROI spreadsheets — are mature and widely adopted, but they are PDFs and spreadsheets that never touch investee data; the analyst hand-keys the rubric, the numbers drift quarter to quarter, and the auditor can trace nothing. ESG and portfolio platforms — Watershed, Novata, Workiva and their peers — were built for ratings and CSRD/SFDR disclosure, not contribution analysis; they track the standardized metric but never read the memos, narratives, and interviews where a contribution claim actually lives, so they can tell you the number and not defend it. And a raw foundation model will draft a fluent one-investee summary, but run the same prompt twice and you get two different answers — on 2026 enterprise-document benchmarks, every major reasoning model still fabricated information in more than 10% of summaries. "The model said so" is not an audit trail.

Impact intelligence is the realignment: the frameworks stay, but they run on a connected record instead of a folder. Reliable answers from your impact data, in minutes rather than months — and reachable by the many people who need them (the GP, the IC member, the LP-facing analyst, the compliance reviewer) through a plain-language Assistant, not a dashboard only one analyst can drive. Choosing a traditional player now is a bet against where every LP, auditor, and board is already heading.

The data dictionary is where frameworks stop being PDFs

IMP's Five Dimensions, IRIS+, SROI, and a theory of change all reduce to the same operable artifact: a field-by-field definition of what counts as evidence, where each field carries a definition, a unit, a framework binding, and its evidence source. Sopact drafts that dictionary by reading the documents the fund already holds, then enforces it at every intake so twenty investees stay comparable. You edit definitions; you don't author them from a blank page.

Draft the data dictionary for [FUND] from these documents: define each field once — with a definition, a unit, a framework binding (5D / IRIS+ / ToC), and its evidence source — so all [N] investees report against the same schema. Flag any field that duplicates what fund admin already holds, and mark anything you inferred as [INFERRED].
Any framework
IMP · Five DimensionsIRIS+SROITheory of ChangeYour own DD rubric
Field
Framework binding
Evidence source
outcome_confirmed
5D · What
onboarding call
beneficiary_tier
5D · Who
intake · disaggregated
income_change_pct
5D · How Much
KPI pack · IRIS+
attribution_claim
5D · Contribution
interview · cited
toc_drift_signal
ToC · assumption
quarterly narrative

The fund reviews the definitions and locks the schema once. From then on every quarterly submission, financial report, and audit lands against the same fields, and the dictionary scores the Five Dimensions, maps to IRIS+, and tests the theory of change from the same records — across the whole portfolio, not one investee at a time.

The IMM cycle, stage by stage

Whatever the mandate, the cycle underneath has the same six stages, and one investee record inherits all of them. Each stage below: what it does, the prompt to run it, and what to expect back.

Stage 1 — Diligence: score the thesis, mint the ID

What it does. The investee ID is minted at the first DD document and never breaks again. Scoring the Five Dimensions at diligence with evidence citations is its own discipline — covered in depth on due diligence software. The rubric is scored from the deal materials with a citation under each dimension, and the theory of change is extracted from the pitch as testable assumptions.

Score [INVESTEE] against the Five Dimensions from the DD pack: read the investment memo, founder interview, and financial model; score What, Who, How Much, Contribution, and Risk with a citation under each; extract the theory of change as testable assumptions; and flag where the evidence is thin. Mark anything inferred as [INFERRED].

Expected output. A pre-scored, cited IC brief that becomes the binding spine for every quarterly report that follows.

Tips for reliable output. Refine Contribution and Risk by hand — they carry the most judgment. Let the model draft; the IC partner confirms.

Stage 2 — IC & onboarding: bind the dictionary

What it does. The IC brief is generated from the DD record, the theory of change is aligned with the investee on a structured call, and the baseline survey is deployed. The data dictionary is bound here, so every later submission has fields to land on.

From the DD record for [INVESTEE], draft the IC brief and confirm the theory of change with the investee: list the assumptions to test each quarter, bind them to the data dictionary fields, and design the baseline survey that establishes the "compared to what" for every outcome claim.

Expected output. A confirmed record and a bound schema — the yardstick the investment is measured against.

Stage 3 — Quarterly: evidence lands on one ID

What it does. Quarterly narratives, financials, audits, and stakeholder surveys append to the same record on arrival — the cadence side is covered on portfolio monitoring software. The Five Dimensions are re-scored, and missing fields go back to the investee while the quarter is open.

Structure this quarter's submission for [INVESTEE] onto the record: map the KPI pack to the dictionary, theme the written and audio narrative, reconcile the financials against last quarter, and list every field still missing so it can go back to the investee this week with a unique link.

Expected output. A scored quarterly report plus a specific missing-data list — management, not archaeology.

Tips for reliable output. Let the investee submit what it already produces. Bespoke impact forms layered on top of the work are how submission rates die.

Stage 4 — Score & learn: re-score and surface drift

What it does. The dimensions are re-scored against last quarter with citations, and the current evidence is tested against the theory of change bound at onboarding. The assumption that stopped holding flags itself, mid-cycle, with the paragraph or metric that contradicts it.

Re-score [INVESTEE] on the Five Dimensions against last quarter with citations, and test the current evidence against the theory of change bound at onboarding. Flag any assumption that stopped holding, the narrative paragraph or metric that contradicts it, and any outlier worth a check-in call before the LP sees it.

Expected output. A drift-and-risk view with citations — the outlier surfaced while there is still time to act on it.

Stage 5 — Report: every output is a view

What it does. The LP letter, board and IC pack, IRIS+ filing, and bespoke per-LP report each generate as a view on the same records — every figure carrying its citation, every view regenerating when the data changes. The reporting side in depth: impact reporting software.

Draft the [LP / board / IRIS+] report for [PERIOD] from the portfolio records: roll up the Five Dimensions by investee and fund, pull the IRIS+-aligned indicators, join fund-admin capital data where the report needs spend beside outcomes, and keep every figure one click from its source. Format for [AUDIENCE].

Expected output. A report that reads as learning and survives an audit — produced overnight, on evidence a reviewer can verify.

Tips for reliable output. Different LPs want different shapes — that's fine. Each destination is a mapping on the same records; map once, regenerate every quarter.

Stage 6 — Exit: defensible against the underwriting

What it does. The exit summary pulls the full history — final score with the citation trail, the theory of change validated or amended for the next vintage.

Assemble the exit summary for [INVESTEE] from the full record: final Five Dimensions score with the citation trail, the movement per dimension across the hold, whether the theory of change held, and what should carry into the next vintage's underwriting.

Expected output. An exit view that compares outcome to underwriting — evidence, not narrative.

One investee, six stages — one record that inherits every prior stage

The point of the cycle is what accumulates. The ID minted at the first DD document is the same record at exit; nothing is re-keyed between stages, so by year three the same row holds the full investment lifecycle, queryable in seconds. Click through the stages below and watch the record compound — Identity captured once, Five Dimensions scored then re-scored, theory of change extracted then tested, stakeholder voice added at baseline, reports drafting as views.

VISTA HEALTH PARTNERS · A4217
One connected record
Illustrative · 2024 vintage · health, climate, livelihood
$4.2Mdeployed
6submissions
24 mointo hold

What the record carries
Identity
Five Dimensions
Theory of Change
Stakeholder voice
Reports
The LP report drafts itself. The cleanup that used to take weeks now runs overnight, every cycle — the reviewer edits a generated draft; nobody assembles it from scratch. The structure holds for every investee in the portfolio.

The legacy stack tunes for the annual report — the real job is the Tuesday question

The annual report is the artifact everyone optimizes for and almost nobody actually reads under pressure. The questions that decide things arrive on a Tuesday: the GP asking about a portco, the LP asking how a contribution claim is defended, the IC member asking what the brief actually says. Answered on one record, each is a query; answered on the legacy stack, each is a project.

QWhat's the portfolio-level Five Dimensions roll-up this quarter, with citations?
On one record

One query. Scores read from each investee record, joined to current-quarter submissions, returned with a citation under every figure — verifiable by clicking to the source memo or survey response.

On the legacy stack

Three weeks, two analysts. Pull each submission, re-key the scores into a master spreadsheet, reconcile against last quarter's rubric, rebuild the chart — and hope the citations reassemble before the LP letter goes out.

QWhich portcos are drifting from their theory of change, and what's the evidence?
On one record

Drift surfaces during the cycle. The theory of change extracted at DD is the same record tested each quarter — the assumption that stopped holding flags itself, with the narrative paragraph or survey response that contradicts it.

On the legacy stack

You find out at exit. The theory of change is a slide in the IC packet. Nobody re-reads it during the hold. Drift becomes visible only when the outcome diverges from the underwriting case.

QWhat does this Series B candidate's IC brief say — and can every claim be verified?
On one record

Generated from the DD record. The brief pulls the rubric, interview themes, model summary, and contribution analysis from the structured record — every claim cited to source. The IC reads a defensible document, not an opinion in a template.

On the legacy stack

One associate writes it from scratch. Re-reads the memo and interview, drafts scores from memory, formats to the template. Takes a week — and nobody can verify the claims without re-reading every document.

Sopact's one bet: one investee record that inherits every stage

Not another dashboard on top of disconnected folders — one record. The ID issued at the first DD document is the same record at exit, and it doesn't break across IC rounds, follow-ons, restructurings, vintages, or platform migrations. Five Dimensions scores, IRIS+ indicators, and financial metrics sit alongside DD memos, IC notes, and quarterly narratives on that one record, and every contribution claim, drift signal, and score cites the specific paragraph, response, or submission behind it — so auditors and LPs verify in one click. This has been Sopact's day job since 2014, before the category had a name. The decision it enables is concrete: when an LP audit asks how a contribution claim is defended, the answer is a citation trail, not a re-reading project.

Sopact connects — it does not replace your fund admin, CRM, or accounting

The boundary question comes up on nearly every call, so here it is plainly: your systems of record stay. Fund administration, the CRM, the data room, and the accounting platform keep doing their jobs; Sopact Sense sits in the middle and holds one connected investee record — capital and valuations flow in from fund admin, evidence flows out to the LP letter, the board view, the IRIS+ filing, and your BI tool. It is "an AND," not a rip-and-replace. Below roughly ten investees, a shared drive and a survey tool is often still the right answer; above about $5B in AUM, the fit is partial and we will say so. The record earns its keep where the qualitative-plus-quantitative join across a portfolio is unmanageable by hand — the mid-market impact fund with multiple LPs, an IC and quarterly cadence, and no full-time data engineering.

Learn the how-to: frameworks and evidence in the Academy

The cycle above is the argument; the Academy walkthroughs are the practice — each runs on your own data.

Impact measurement and management — frequently asked questions

What is impact measurement and management (IMM)?

IMM is the discipline of designing an investment or program around intended outcomes, collecting evidence of change as it arrives, and using that evidence to manage — not just report. In an impact fund it spans the full lifecycle: scoring the thesis at diligence, binding a theory of change at onboarding, re-scoring each quarter, and producing LP, board, and compliance reports from one continuously learning investee record.

What is the full form of IMM, and does it mean other things?

In investing and philanthropy, the full form of IMM is Impact Measurement and Management. The acronym carries unrelated meanings elsewhere — the International Monetary Market in derivatives, internal models in bank risk management, a medical abbreviation. Across impact investing, IMM always means measuring outcomes and managing against them.

What is IMM software?

IMM software is the system of record for impact evidence: it holds the investee identity, the framework scores, the theory-of-change drift, the stakeholder voice, and the citation trail under every claim. It differs from survey tools (collection only), ESG platforms (disclosure only), and BI dashboards (numbers without narrative evidence). The test of real IMM software is whether the score in this quarter's LP letter traces to the same record as the score in the original IC memo.

Which frameworks does IMM use — and do they compete?

The three core frameworks are complementary, not competing. The IMP's Five Dimensions of Impact (What, Who, How Much, Contribution, Risk) is the question structure; IRIS+ is the GIIN's catalog of standardized indicators; a theory of change is the causal logic specific to each investee. One data dictionary binds all three, so a single quarterly submission scores the dimensions, reports IRIS+-aligned metrics, and tests the theory of change at once.

How is IMM different from ESG platforms like Watershed, Novata, or Workiva?

ESG platforms are built for ratings and disclosure; IMM is built for evidence and management. An ESG platform tracks standardized metrics for CSRD/SFDR-style filings but does not read the DD memos, quarterly narratives, and interviews where contribution evidence actually lives — so it can tell you the number, but not defend the claim. IMM systems hold both the numbers and the narrative on one investee record.

What is a data dictionary in IMM?

The field-by-field definition of what counts as evidence — each field with a definition, a unit, a framework binding, and an evidence source. It is the artifact that turns framework PDFs into something operable. Sopact drafts it automatically by reading the fund's own documents, then the team reviews and locks the schema so every investee reports against the same fields.

What happens to missing data and unusual results during a cycle?

They become mid-cycle outputs, not year-end discoveries. When a quarterly submission lands, incomplete fields go back to the investee through a unique link while the quarter is still open. Outliers and narrative risk signals flag for a check-in call before the LP ever sees the report. This is the management half of IMM.

What reports does an IMM system produce?

Each report is a view on the same records: the quarterly LP letter, the board and IC pack, the IRIS+ aligned filing, the annual impact narrative, the exit summary, and bespoke per-LP logic-model reports. Because every view reads from one investee record, a different LP format means a different mapping — not a re-keying project.

Can't a general-purpose AI chatbot do this from our folders?

It can draft a fluent summary; it cannot produce an audit trail. Run the same prompt twice and you get two different answers, and on enterprise-document benchmarks every major model still fabricates information in a meaningful share of summaries. The agentic layer is genuinely useful — but it needs a structured, persistent, citation-bearing record to read from. IMM software is that record; the chat is a view on it.

How is IMM different from monitoring and evaluation (M&E)?

M&E typically evaluates a program after the fact; IMM manages an investment while it can still change course. The two share methods — indicators, baselines, stakeholder feedback — but IMM is wired to the investment lifecycle: diligence scoring, quarterly re-scoring, drift detection, and LP reporting. A nonprofit running M&E on one program and a fund running IMM across 24 investees can use the same record structure; the cadence and audience differ.

Bring a real DD memo. Leave with the citation trail behind every number.

Sixty minutes, no deck. We work a question you already need answered — a portfolio Five Dimensions roll-up, an LP letter, a Tuesday question from your IC — against your real data shape. You leave with a path that doesn't depend on a foundation model returning the same answer twice, or a clear reason it isn't the right fit. Scope a working session → or download Impact Intelligence.