← THE PORTAL
PERFORMANCE STEERING · INVESTOR RELATIONS / EXTERNAL NARRATIVE

From hand-tied numbers to a source-linked narrative loop.

Today, every external number is tied out by hand under deadline. In Future Finance, every external metric rests on a certified definition, narrative drafts assemble source-linked, and the numbers reconcile to the ledger at any moment.
LEARNING RETURN 01 SIGNAL number moves · metric shifts vs prior 02 EXPOSURE tie-out risk · disclosure exposure 03 CHOICE confirm {D} escalate 04 OWNER ACTION IR {D} CFO {D} legal 05 OUTCOME did it reconcile? EVIDENCE GATE 06 · VALUE PROOF narrative approved The governed loop, cast for this function – the same circuit as every Finance function, carrying metric tie-out risk instead of a report.
01 · THE TRANSFORMATION

From hand-tied numbers to a source-linked narrative loop.

TODAY
Every external number tied out by hand, under deadline.
Narrative drafted from scratch each quarter.
Consistency checked manually against prior statements.
Peer context assembled ad hoc.
IN FUTURE FINANCE
Every external metric on a certified definition
Narrative drafts assembled source-linked {D} earnings, board, street
Continuous tie-out {D} external numbers reconcile at any moment
Guidance sensitivity modeled before commitments are made
METRIC TIE-OUT → ACTION ROUTE
02 · WHERE TO START – THE WORKFLOWS, RANKED

Five workflows. One operating pattern.

Every workflow in this function becomes the same governed loop – cast differently. Below, each one in full: what it becomes, who does what, what it needs, and where the human boundary sits.
RANKED BY · ownership (who holds the lever) · value (from this domain’s sizing) · autonomy ceiling (Tier 1 = human-only → Tier 4 = highest permitted autonomy) · control sensitivity · scope (core vs conditional)   weighting leans value + ownership
AGENTS prepare DETERMINISTIC SYSTEMS calculate HUMANS approve OWNERS execute FINANCE validates
1Tier 2 of 4
Earnings and Q&A preparation
Manual Q&A prep becomes source-linked draft support {D} the prep book assembled with every KPI tied out, and messaging kept firmly with CFO, IR, and legal.
Who does what
Agentsdraft Q&A and support, source-linked
Systemstie out KPIs to the ledger
HumansCFO / IR / legal approve messaging
ARCHETYPE cadence-to-decisionCADENCE quarterlyDATA KPIs · disclosures · peersSENSITIVITY high
THE BOUNDARY · external messaging is human-approved; agents draft and tie out, they never speak to the street.
OWNERSHIP Finance analystVALUE Effort + risk avoidanceAUTONOMY Tier 2 of 4CONTROL High sensitivitySCOPE Core
2Tier 2 of 4
Guidance support and sensitivity
Manual sensitivity decks become governed scenario support {D} guidance drivers summarized and sensitivities calculated, with the guidance decision kept human.
Who does what
Agentssummarize guidance drivers and prepare sensitivities
Systemscalculate the sensitivity scenarios
Humansdecide guidance
ARCHETYPE scenario-to-choiceCADENCE quarterly / eventDATA forecast sensitivity modelsSENSITIVITY high
THE BOUNDARY · guidance is human-decided; support is preparation and calculation only.
OWNERSHIP Finance analystVALUE Decision qualityAUTONOMY Tier 2 of 4CONTROL High sensitivitySCOPE Core
3Tier 2 of 4
Disclosure support coordination
Email coordination becomes a governed evidence and signoff workflow {D} disclosure evidence tracked, numbers tied out, and signoff kept human.
Who does what
Agentstrack disclosure evidence and status
Systemstie out the disclosure numbers
Humansdisclosure signoff is human-owned
ARCHETYPE close-to-certificationCADENCE quarterlyDATA disclosure controls · evidenceSENSITIVITY high
THE BOUNDARY · disclosure signoff is human-owned; the workflow governs evidence, not the disclosure decision.
OWNERSHIP Finance stewardVALUE Risk / defensibilityAUTONOMY Tier 2 of 4CONTROL High sensitivitySCOPE Core
4Tier 1 of 4
Investor narrative consistency
Manual consistency checks become source-linked narrative support {D} narratives compared against prior statements and metrics tied out, with messaging approved by a person. Deliberately the lowest-autonomy workflow here.
Who does what
Agentscompare narratives against prior statements
Systemstie out metrics across the narrative
Humansapprove messaging
ARCHETYPE policy-to-complianceCADENCE quarterlyDATA narrative source traceabilitySENSITIVITY high
THE BOUNDARY · messaging is human-approved; narrative support stays Tier 1 {D} lowest autonomy, by design.
OWNERSHIP Finance stewardVALUE Risk / defensibilityAUTONOMY Tier 1 of 4CONTROL High sensitivitySCOPE Core
5Tier 1 of 4
Peer and market context
Manual research becomes agent-prepared peer context with caveats {D} public context summarized for the team, with relevance interpreted by a person.
ranked last: interpretation is human; the value is decision support.
Who does what
Agentssummarize public peer and market context
Humansinterpret relevance {D} the judgment is human
ARCHETYPE cadence-to-decisionCADENCE recurringDATA public sources · peer dataSENSITIVITY low
THE BOUNDARY · interpretation is human; agents summarize public context, they do not draw conclusions.
OWNERSHIP Finance analystVALUE Decision supportAUTONOMY Tier 1 of 4CONTROL Low sensitivitySCOPE Core
03 · THE SIZING – FULL EVIDENCE TRAIL

The number, carried the way every claim is carried.

The figure on the front page arrives here as what it is – a governed packet. Range, basis, inputs, benchmarks, derivation, assumptions, and the strongest objection to it, all in one place.
INVESTOR RELATIONS SIZING PACKET
IR-SZ-04 · DISCLOSURE-RISK + EFFORT
OUTSIDE-IN
RANGE
$1–12M / yr · mixed risk + effort
BASIS
Assumption-built · Confidence: Low
WHAT IT IS
Continuously reconciled numbers reduce the probability and severity of a disclosure or tie-out error, plus modest narrative-production efficiency. Sized as expected value (probability × direct cost) plus an effort estimate. The large, speculative market-cap reaction is excluded by choice.
INPUTS
External audit + audit-related fees (event-cost proxy) and total revenue (disclosure surface) of a representative large-cap SaaS company
BENCHMARKS
US restatements/year ~280–300; ‘Big R’ share rising (CAQ / Audit Analytics) · rev rec historically a top restatement cause (Audit Analytics)
DERIVATION
1 · risk (EV): annual probability of a material error 1–3% × direct cost $5–30M = ~$0.05M–0.9M/yr
2 · effort: narrative/tie-out efficiency = ~$1M–11M/yr (assumed)
3 · total → ~$1M–12M/yr
ASSUMPTIONS
Event probability (1–3%), direct cost ($5–30M), and IR effort ($1–11M) are assumed. Market-cap reaction excluded.
SENSITIVITY
The IR-effort assumption dominates the range (the EV component is <$1M) – this is why confidence is Lowsanity bound: the effort component cannot exceed the benchmarked cost of the IR/reporting function; the EV component is small by construction
THE ATTACK
“You’ve either lowballed by ignoring the stock reaction or you’re guessing at effort.” — Both the low probability and the exclusion of the market-cap tail are deliberate conservatism; the effort piece is an estimate pending internal IR-cost and control-incident data.
OUTSIDE-IN · ILLUSTRATIVE · SUBJECT TO VALIDATION
Modeled on a Representative SaaS Company · outside-in, illustrative
A target-state vision · every value claim subject to validation