Spend trust, on a board a CFO actually reads. — recapture rate, MTTF, MTTR.
Quarterly sampling never produced a spend-trust number — and the post-payment recovery-firm contingency tells you what you already lost, not what is still on the wire. VeraStream rolls the same eight production detectors into a weekly board: weekly detection-vs-recovery funnel, duplicate-payment recapture rate, mean-time-to-flag, and mean-time-to-recover — each tied to the detecting detector IDs underneath.
What "spend trust" means here
The board moves on three loops: signal → action → workpaper
The dashboard is a preview of the v1 spend-trust board. It does not chase a recovery-firm contingency, it is not a quarterly sampling exercise, and it does not re-invent a KPI engine under the data layer. The board reads directly from the eight production detectors — the same pipeline that holds a duplicate pre-payment and that writes the workpaper — so the number on the board is the number the auditor would see. One receipt, one detector fire, one recovery event, one board cell.
The four Tier-1 KPIs
What the v1 CFO spend-trust board looks like
Two lines per tile: the headline number and the trend behind it. Below each tile, the detecting detector IDs — the four to six production detectors that produced the receipts. They are the same detectors that back the weekly finding roll-up and the PCAOB AS 2315 workpaper.
Weekly Detection-vs-Recovery Funnel
$846K ↓ / $612K ✓ / 72%
Receipts flagged → receipts recovered, last 7 days vs. trailing 4-week median.
Weekly · receipts → recovered
What moves this number
Drives off the count of receipts that fired at least one detector during the week and the subset of those receipts whose recovery actually posted (a voided payment, a vendor refund, a stopped ACH). Upstream: detector coverage and ERP cutoff windows. Downstream: AP team pace on void requests.
Detecting detectors
Duplicate-Payment Recapture Rate
94.1%
Share of duplicate-payment receipts that produced a stopped or returned disbursement, last 30 days.
Monthly · recovered ÷ flagged
What moves this number
Numerator: each detectDuplicatePayment or findDuplicateInvoices fire that resulted in a real recovery event (void, return wire, or vendor credit). Denominator: every duplicate receipt dated within the window. Slows when AP voids lapse past the bank cutoff or when the vendor obligation rolls to a credit note rather than cash.
Detecting detectors
Mean-Time-to-Flag
11m 24s
Median time from a posted receipt to a detector fire, last 7 days.
Weekly · ingestion → flag
What moves this number
The cheapest KPI to move. Bounded by the ERP ingest SLA and the four lowest-latency detectors (evaluatePolicy on the approver chain, detectThresholdGaming on the split-purchase cluster, detectRoundDollar on the clone-template cents, detectVendorRisk on the typosquat). Drops with pre-paid connector tiers; rises when the AP stack runs on daily batch instead of real-time.
Detecting detectors
Mean-Time-to-Recover
6h 12m
Median time from flag to a confirmed recovery event, last 7 days.
Weekly · flag → recovery
What moves this number
The recoverable findings only. DetectDuplicatePayment (a stopped ACH or a vendor credit) and detectGhostEmployee (a held pre-payment batch) are the two detectors with a real recovery path. Bound to the AP team’s voiding cadence and the bank’s same-day reversal window. The recovery event — not the flag — is the action this KPI counts.
Detecting detectors
What ships in v1
The v1 board, the v1 pipeline, and the v1 workpaper trail
Five v1 surfaces; four explicit v2 deferrals. The board above is not aspirational — every KPI on it runs against the production pipeline on day one.
- The four Tier-1 KPIs as a single weekly board. Weekly detection-vs-recovery funnel (a stacked bar across detectDuplicatePayment + detectGhostEmployee), duplicate-payment recapture rate (a recovered-÷-flagged line on detectDuplicatePayment + findDuplicateInvoices), mean-time-to-flag (a rolling median anchored on evaluatePolicy, detectThresholdGaming, detectRoundDollar, detectVendorRisk), and mean-time-to-recover (a rolling median anchored on detectDuplicatePayment + detectGhostEmployee). One board, one weekly email, one workpaper trail.
- The full eight-detector pipeline beneath the board. Every KPI reads only from the production receipts that the eight-detector pipeline produced. The board is the consumer; the pipeline is the source. The same eight production detectors — evaluatePolicy, findDuplicateInvoices, detectExpenseAnomalies, detectVendorRisk, detectThresholdGaming, detectRoundDollar, detectDuplicatePayment, detectGhostEmployee — run on every payment.
- Weekly finding roll-ups with the receipt trail. A weekly roll-up — every flagged receipt from the past seven days, grouped by detector and by recovery status, sized for the CFO email and for the AP lead’s workpaper. The same payload backs the weekly newsletter; the same payload backs the external auditor’s PCAOB AS 2315 packet on request.
- PCAOB AS 2315 workpaper evidence on demand. For every receipt on the board, a one-click workpaper: the source invoice, the rule that fired, the corroborators that landed on the same receipt, the override field, and the recovery event. Sized to hand to your external auditor under PCAOB AS 2315, and to the internal-audit team under SOX 404(b), without a separate evidence package.
- Pre-built ERP, corporate-card, and HRIS connectors — no middleware. Live deploys use pre-built connectors to NetSuite, SAP, Oracle, Coupa, Concur, Expensify, Brex, Ramp, and (HRIS) Workday and BambooHR for vendor / approver cross-check. New payments and invoice postings stream into the pipeline in real time; the Tier-1 KPIs are recomputed at the same cadence.
Deferred to v2 — explicit list
- Drill-through by entity / business unit / cost center. A v2 board layer that breaks each KPI out by legal entity, business unit, ERP instance, and cost-center slice — and overlaying it with the per-entity recovery goal. Useful once the v1 weekly board is anchored; not promised in v1.
- Vendor-master-only KPI sub-board. A KPI sub-board that isolates detectVendorRisk and the master-data events from identifyDuplicateInvoice-to-master collisions — useful where the spend-trust question is really “is the master file clean?” rather than “are the disbursements clean?”. v2.
- A custom KPI builder (“bring your own detector”). A workflow that lets a customer author a KPI derived from any of the eight detectors, with their own recovery-status logic and their own numerator/denominator formula. Deferred to v2 — the four Tier-1 KPIs in this preview are the v1 set.
- Forecast / budget overlay and trend-line extrapolation. A weekly forecast of recovered-÷-flagged for the next eight weeks, and a budget overlay that ties the Tier-1 KPIs to the spend-trust target the CFO board has set. Deferred to v2.
See it on real data
What the Tier-1 board looks like against your own AP ledger
A worked sample on a synthetic ledger, the production pipeline against your live AP feed. The CFO board is the consumer — the audit at /audit is where the underlying receipts are surfaced.
Frequently asked
Common questions from CFO, controller, and internal-audit teams
What the v1 CFO spend-trust board contains, what it reads from, and what it deliberately defers to v2 — plain HTML answers, no JavaScript required to read.
Who reads the VeraStream CFO spend-trust dashboard?
The weekly board is built for the CFO, the controller, the head of AP, and the internal-audit lead. Each of the four Tier-1 KPIs is a single number with a trend, a one-line caption, and the detecting detector IDs underneath — so the CFO reads the headline and the audit lead reads the receipt trail in the same view. The same eight production detectors — evaluatePolicy, findDuplicateInvoices, detectExpenseAnomalies, detectVendorRisk, detectThresholdGaming, detectRoundDollar, detectDuplicatePayment, detectGhostEmployee — run on every payment.
What is the Weekly Detection-vs-Recovery Funnel?
A weekly stacked view of receipts flagged by detectDuplicatePayment or detectGhostEmployee — the two detectors that have a real recovery path — against the subset of those receipts that produced an actual recovery event (a voided payment, a vendor refund, a stopped ACH). The ratio is the share of the week's flagged dollars that actually came back, not just the share that surfaced.
How is Duplicate-Payment Recapture Rate different from a recovery rate?
A traditional recovery rate reports dollars recovered against dollars found — often across a recovery-firm engagement and on the firm's contingency. The Duplicate-Payment Recapture Rate here is the share of detectDuplicatePayment and findDuplicateInvoices receipts in the last 30 days that produced a stopped disbursement or a vendor credit, computed continuously from the production pipeline. No contingency, no quarter lag.
How do Mean-Time-to-Flag and Mean-Time-to-Recover relate to quarterly sampling?
Quarterly sampling measures detection roughly once per quarter — near the back of the audit, in the workpaper, not on the disbursement run. Mean-Time-to-Flag (MTTF) here is the median time from the posted receipt to a detector fire, anchored on evaluatePolicy, detectThresholdGaming, detectRoundDollar, and detectVendorRisk — the four lowest-latency detectors. Mean-Time-to-Recover (MTTR) is the median time from flag to a real recovery event, anchored on detectDuplicatePayment and detectGhostEmployee. MTTF catches things the day they post; MTTR measures how fast the AP team actually closes the loop. Sampling measures neither.
What ships in v1, and what is deferred to v2?
v1 ships: the four Tier-1 KPIs in this preview, the full eight-detector pipeline beneath them, weekly finding roll-ups, PCAOB AS 2315 evidence workpapers, and the pre-built ERP / corporate-card / HRIS connectors. v2 defers: drill-through by entity / business unit, a vendor-master-only KPI sub-board, a custom KPI builder, and a forecast / budget overlay. The board in this preview is the v1 set; the v2 items are listed explicitly so nothing is hidden in the gaps.
How does the eight-detector pipeline feed the Tier-1 KPIs?
The same eight production detectors — evaluatePolicy, findDuplicateInvoices, detectExpenseAnomalies, detectVendorRisk, detectThresholdGaming, detectRoundDollar, detectDuplicatePayment, detectGhostEmployee — run on every payment. Every Tier-1 KPI on the board reads only from the production receipts that pipeline produced — there is no separate KPI engine beneath the dashboard. Mean-Time-to-Flag is bounded by the four lowest-latency detectors (evaluatePolicy, detectThresholdGaming, detectRoundDollar, detectVendorRisk); Mean-Time-to-Recover is bounded by the two with a real recovery path (detectDuplicatePayment, detectGhostEmployee); the funnel and the recapture rate are anchored on the same two plus findDuplicateInvoices.
How do I launch the board against my live AP stack — and what does it cost?
Live deploys use pre-built connectors to NetSuite, SAP, Oracle, Coupa, Concur, Expensify, Brex, and Ramp; HRIS connectors read Workday and BambooHR for vendor / approver cross-check. Drop a CSV of the AP ledger at /audit and the same detectors run in the browser against your data — no integration required to test the same board in dry-run. /pricing covers the monitored live deployment against your real spend tier. For the connector target list and the progressive tier-by-tier rollout of each one, see /integrations.
See the v1 board on your own data
Spend trust, on a board a CFO actually reads.
Drop a CSV of your AP ledger at /audit — the same eight detectors run in the browser, the same four Tier-1 KPIs populate against your own data. Use /pricing to launch the board continuously against your live ERP.