Recurring reconciliation agent case study

Recurring Reconciliation Agent Case Study

A $200M company's month-end close included 47 recurring reconciliations — prepaid amortization, deferred revenue, fixed asset depreciation, intercompany, accruals, prepaids — each with documented schedule and reviewer sign-off. AI agent automated the production work; close cycle dropped from 9 to 4 days.

Client profile: Composite case study based on a $200M B2B services company on NetSuite. Monthly close included 47 named recurring reconciliations across prepaid, deferred, accrued, fixed-asset, intercompany categories. 6-person finance team; 5–7 person-days per cycle on reconciliations alone.

See measured results Talk through your situation

Company context Before — what was actually broken What Ledger Summit implemented Recurring reconciliation agent mechanics Implementation timeline Measured results

Company context

The client is a $200M B2B services company on NetSuite. Monthly close included 47 named recurring reconciliations spanning prepaid amortization (~12 reconciliations), deferred revenue (~8), fixed-asset depreciation (~9), intercompany (~7), accruals (~11). Each had a documented schedule and reviewer sign-off. The AP / accounting team spent 5–7 person-days per cycle reproducing them in spreadsheets, posting JEs, and reconciling to GL.

These reconciliations are perfect candidates for AI agents because the math is deterministic, the data is structured, and the volume is too low to justify a dedicated tool but too high to leave manual.

$200M B2B services
NetSuite GL
47 named recurring reconciliations
12 prepaid amortizations
8 deferred revenue schedules
9 fixed-asset depreciation calculations
7 intercompany reconciliations
11 accrual calculations
5–7 days per cycle on these
6-person finance team

Before — what was actually broken

Each reconciliation in its own spreadsheet
Schedule re-entered each cycle
JE manually posted
Sub-ledger to GL reconciled at end
Reviewer sign-off via paper or email
Evidence pack reconstructed at audit time

What Ledger Summit implemented

Recurring reconciliation agent: per-reconciliation schedule preserved; auto-calculate next period; auto-post JE; reconcile to GL
Prepaid amortization: schedule pulled; current period amortization calculated; JE posted; remaining-prepaid balance updated
Deferred revenue: contract data pulled; recognition pattern applied; JE posted; deferred balance updated; rollforward auto-generated
Fixed-asset depreciation: register pulled; current period depreciation calculated per asset; JE posted; accumulated depreciation updated
Intercompany: matched per entity-pair; eliminations posted; differences flagged for review
Accruals: source data pulled (e.g., AP for unbilled, payroll for accrued comp); accrual calculated; JE posted; reverses next period
Reviewer routing: each reconciliation marked complete; reviewer signs off in queue
Evidence pack per reconciliation: source data, calculation, JE, reviewer approval
Exception queue for reconciliations that don't pass auto-reconciliation
Period-end report: which reconciliations passed, which had exceptions, summary

Recurring reconciliation agent mechanics

Reconciliation type	Source data	Calculation	Output
Prepaid amortization	Prepaid asset balance + amortization schedule	Period amortization	JE: amortization expense, reduce prepaid
Deferred revenue	Contract data + recognition pattern	Period revenue recognition	JE: deferred revenue → revenue
Fixed-asset depreciation	FA register	Per-asset depreciation	JE: depreciation expense, reduce FA
Intercompany matching	Entity-pair postings	Match + eliminations	Elimination JE; exception queue
Accrual — payroll	Payroll provider	Period accrual	JE: accrue, reverse next period
Accrual — AP unbilled	AP system + open POs	Period accrual	JE: accrue, reverse on bill
Accrual — bonus	HRIS + plan rules	Period accrual	JE: accrue, reverse on payment
Sub-ledger to GL	Sub-ledger + GL	Auto-tie-out	Reconciliation report; exception flag
Bank reconciliation	Bank feed + GL	Auto-match	Match report; exception queue

Implementation timeline

Weeks 1–2: Discovery: 47 reconciliation inventory; documentation; reviewer routing
Weeks 3–5: Build: per-reconciliation logic; integration with NetSuite; reviewer queue
Weeks 6–7: Pilot: 12 highest-volume reconciliations automated; shadow run
Weeks 8–9: Full deployment: 47 reconciliations live; reviewer training
Weeks 10–11: Hypercare; rule refinement; auditor walkthrough

Measured results

Metric	Before	After	Delta
Recurring reconciliations	47 manual	47 automated	—
Person-days / cycle on reconciliations	5–7	0.5	−92%
Close cycle	9 days	4 days	−56%
Reconciliation variance at sign-off	Reconciling differences	$0	−100%
Audit fieldwork days (close)	3	1	−2 days
Evidence pack assembly	Audit-time reconstruction	Per-cycle, automated	—

Alternatives considered

Option	Time	Cost	Strengths	Weaknesses
BlackLine recurring reconciliations	5–7 months	$420K–$680K + license	Best-in-class	Cost; over-scoped at $200M
Trintech / Cadency	5–7 months	$320K–$520K	Enterprise-grade	Over-scoped
FloQast (close mgmt)	3 months	$140K–$220K + license	Modern	Less depth on auto-calc
Numeric	3 months	$120K–$180K + license	Modern	Newer; less depth
Ledger Summit AI agent (selected)	11 weeks	$140K–$220K	Right-sized; SOX-clean	Maintenance ongoing

When this approach fits

$50–500M companies with material recurring reconciliations
20+ recurring reconciliations per cycle
NetSuite, Sage Intacct, or comparable GL with API access
Documented schedules already exist
Annual external audit with control posture pressure
Existing accounting team (not bookkeeper-only)

Lessons learned

Inventory the reconciliations first. Knowing 47 vs. estimating ~25 changes scope materially.
Determinism is the value. Calculation is templatable; data integration is the work.
Reviewer routing per reconciliation. Some always reviewed (intercompany); some sample-reviewed (prepaid amortization).
Exception queue for fails. Auto-reconciliation that fails goes to a human; never auto-pass with variance.
Evidence pack per cycle, automated. Auditor walkthrough goes from days to hours.

Frequently asked questions

What's a "recurring reconciliation"?

Reconciliation that repeats with predictable rules each cycle: prepaid amortization, deferred revenue, fixed-asset depreciation, accruals.

Why not BlackLine?

Excellent for $500M+ companies with 200+ reconciliations and SOX 404(b). For $200M with 47 reconciliations, the implementation cost rarely pays back.

How does this differ from BlackLine?

Same outcome (automated reconciliation + evidence) at lower scale and cost. Trade-off: less polish, less mature.

What about ad-hoc reconciliations?

Out of scope for the agent; routine human work. The agent handles the recurring 80%.

How does this support audit?

Per-reconciliation evidence pack: source data, calc, JE, reviewer approval. Walkthrough goes from days to hours.

What if a reconciliation fails?

Exception queue for human review. Agent never auto-pass with variance.

Can the agent handle period-cutoff reconciliations?

Yes — same pattern applies; period-end specific logic for accruals, etc.

What about year-end reconciliations?

Year-end specific reconciliations supported: bonus accruals, audit fees, contingencies, etc.

How do you handle changes to reconciliation schedules?

Change management: documented reason, reviewer approval, agent updated. Same as any control change.

What's the typical cost?

$140–280K project + ongoing maintenance. Pay-back through close-cycle reduction + audit fee reduction.

Recurring reconciliations eating your close cycle?

A 30-minute call walks your reconciliation list and tells you what AI agent automation looks like.

Book a free call