Colorado Tech ◆ Illustrative sample · synthetic data · no real client

Data Truth Report

Where three systems report three different numbers — and what the reconciled figure actually is.

Prepared for
Ridgeline Provisions Co.
(fictional sample brand)
Period reviewed
March close
Scope
Commerce · Fulfillment · Finance
Prepared by
Colorado Tech
01 — Executive summary

Three systems, three numbers, one truth.

Ridgeline runs a Shopify + Amazon storefront, a third-party warehouse, and QuickBooks. Each reports March net revenue differently, and none of them is right. Below is the reconciled figure and the three defects that move the most money — the full register follows in §5.

Commerce (storefront)order date · gross
$4.51M
Fulfillment (3PL)ship-confirmed basis
$4.06M
Finance (GL)refunds netted, mis-dated
$4.33M
− Phantom volume removedduplicates · transfers · unshipped at cutoff
−$0.34M
± Refund re-periodizationmatched to original order month
+$0.02M
Reconciled net revenueone basis · ship-confirmed, in-period
$4.19M
Net revenue (March)
$4.19M
Storefront overstates by ~$0.32M (≈8%)
Duplicates, unshipped-at-cutoff orders, and mis-dated refunds.
Gross margin
38.6%
Reported 42.0% — overstated 3.4 pts
Landed freight & duty never capitalized into COGS.
On-hand · SKU-2481
1,090 u
Storefront shows 1,240 — 150 phantom
In-transit transfer counted as available-to-sell.
02 — Source systems & system of record

What each system is authoritative for.

The reconciliation starts by naming, for every number, which system is the system of record — and where a second system is quietly reporting the same thing on a different basis.

SystemRoleSystem of record forReports revenue asWhere it breaks
ShopifyDTC storefrontOrders, customers, discountsGross, at order dateGross of refunds, discounts & tax; UTC timestamps
AmazonMarketplaceMarketplace orders & feesNet settlement, ~14-day cycleStraddles month-end; net of referral/FBA fees
3PL / WMSFulfillmentPhysical on-hand, ship-confirmShip-confirmed units × priceExcludes unshipped-at-cutoff; no landed cost
QuickBooksFinance / GLRecognized revenue, COGS, taxNet, at deposit dateRefunds mis-dated; landed cost in opex; tax in revenue
Payment processorsSettlementFees, chargebacks, reservesNet payout, batchedCollapsed into one deposit line
03 — Where the truth breaks

The map of the reconciliation.

Every source flows into one reconciled layer. The marked breaks (◆) are where a number is lost, double-counted, or mis-timed on the way in — each is addressed in §4 and §5.

Shopify · Amazon 3PL / WMS QuickBooks Processors ◆ dupes · tz · discounts ◆ unshipped cutoff ◆ refund period · landed cost ◆ net fees · chargebacks RECONCILED LAYER $4.19M P&L Margin Inventory
04 — The reconciliations

Three numbers, reconciled to the cent.

Each headline number is walked from the conflicting sources to the true figure — the mechanism, the steps, the durable fix, and what trusting the wrong number would have cost.

Net revenue — March

Commerce $4.51M · 3PL $4.06M · Finance $4.33M → reconciled $4.19M

Root cause

The systems disagree for one reason and then all miss the truth for a second. Commerce still shows original sales at full value. Finance has already netted $0.18M of refunds against March — which is why it sits below Commerce — but it booked them on the processor’s settlement date instead of matching each to its original order’s period, so the contra-revenue landed in the wrong month. Fulfillment’s $4.06M is on a ship-confirmed basis, so it excludes orders unshipped at the 3/31 cutoff. And none of the three has stripped $0.34M of phantom volume: duplicate marketplace-connector re-imports, internal stock-transfer records posted as orders, and unshipped-at-cutoff orders the storefront counted as sales. Correcting the period match and removing the phantom volume brings every source to $4.19M.

Reconciliation steps

  1. Anchor on Commerce total sales of $4.51M; confirm it is gross of refund reversals and includes shipping/tax lines.
  2. Remove $0.34M of non-revenue volume: dedup marketplace orders the connector re-imported against native order IDs, back out internal stock-transfer pseudo-orders, and exclude orders unshipped at cutoff (recognize on ship-confirm). Subtotal $4.17M.
  3. Re-period the $0.18M of refunds: match each refund to its original order’s month rather than the settlement date. Net in-vs-out-of-month effect moves March only ~$0.02M — landing net revenue at $4.19M.
  4. Prove Finance ties: $4.33M equals Commerce less the refunds it netted but with phantom volume still in and refunds mis-dated; removing the duplicates/unshipped and re-periodizing reconciles $4.33M down to $4.19M as well.
Durable fix. Recognize revenue on the 3PL ship-confirm event in one standardized timezone; enforce idempotent order dedup on native channel order IDs across every ingestion path; and accrue a same-period refund reserve at close, matching every refund to its original order. The month-end plug becomes a quantified bridge, not a guess.
If left alone. Trusting the storefront’s $4.51M overstates March by ~$0.32M (≈8%) — inflating commissions, board-reported growth, and any earn-out or covenant tied to top line. Even Finance’s $4.33M overstates by ~$0.14M and hides that refunds chronically land in the wrong month, so every close whipsaws and margin looks volatile with no operational cause.

Gross margin

Reported 42.0% → actual 38.6% · −3.4 pts

Root cause

Inventory is received at supplier PO price only. The freight-forwarder, customs-duty, and brokerage invoices for the same inbound shipments arrive later, from different vendors, and are coded straight to an opex “Freight-in / Import expense” line instead of riding on the units. Because those landed costs never capitalize into inventory, COGS is understated and reported margin looks artificially clean — until the CPA capitalizes them at year-end and forces the number back down. GAAP requires inbound freight, duty, and brokerage to be capitalized into inventory and released to COGS as the units sell.

Reconciliation steps

  1. Pull every inbound-shipment cost for the period — supplier invoice, freight-forwarder, customs duty, brokerage/handling — keyed to the PO/shipment.
  2. Allocate across the units on each shipment: by invoice value for supplier-linked costs, by weight/volume for freight, by HTS rate for duty — deriving a true per-unit landed cost.
  3. Capitalize into inventory and release to COGS only on units actually sold in the period, leaving unsold landed cost on the balance sheet.
  4. Reconcile allocated landed cost back to the actual vendor invoices so nothing is dropped or double-counted; recompute margin — 42.0% falls to 38.6%.
Durable fix. A landed-cost layer keyed to each inbound shipment that ingests supplier, freight, duty, and brokerage invoices, allocates them to units, and posts a per-unit landed cost flowing to COGS on sale — reconciled to vendor invoices every month so margin never drifts between closes.
If left alone. A 3.4-point overstatement on a ~$50M-GMV brand is roughly $1.7M of phantom gross profit a year — underpricing SKUs, greenlighting promotions that lose money, distorting channel and product mix, and setting up a painful year-end restatement.

On-hand units — SKU-2481

Storefront 1,240 · 3PL 1,090 · −150

Root cause

The storefront’s available number was computed as on-hand plus a 150-unit inbound transfer that has not yet been physically received and scanned at the 3PL. The channel treated in-transit stock as sellable for the whole window between origin decrement and destination receipt, so it advertises 1,240 while the shelf holds 1,090. Available-to-sell should equal destination on-hand minus open allocations, per fulfillable location — never on-hand plus in-transit.

Reconciliation steps

  1. Confirm the 3PL/WMS physical on-hand of 1,090 as the system of record for the unit.
  2. Trace the 150-unit gap to a specific open transfer decremented at origin but not yet receipt-scanned at the destination 3PL.
  3. Reclassify those 150 units to an explicit in-transit / unreceived state, excluded from channel available-to-sell until the 3PL scans them into sellable on-hand.
  4. Recompute available-to-sell as 3PL on-hand (1,090) − open allocations − safety stock, pushed to every channel from one shared allocation ledger.
Durable fix. Model a distinct in-transit state that only rolls into sellable on-hand on the 3PL receipt scan, designate the 3PL/WMS as the single system of record for physical on-hand, and drive every channel’s available from one per-location allocation ledger — with a signed variance alert when channel and 3PL diverge beyond a threshold.
If left alone. Advertising 150 units the warehouse doesn’t have drives overselling a fast mover: cancellations, delayed shipments, Amazon late-shipment and order-defect penalties that can suppress the listing, expedite/reship costs, and support load — plus an inventory-asset misstatement at close.
05 — Findings register

Everything the audit surfaced.

The three headline reconciliations above sit inside a broader register. Each finding names the systems in conflict, the mechanism, and the directional exposure; severity reflects dollar impact and how often it distorts a decision.

#FindingSystemsCategoryExposureSev.
1Landed cost not in COGS3PL/customs vs GLMargin3–8 pts marginHigh
2Net deposits booked as revenueProcessors vs GLRevenue~3–5% of GMVHigh
3In-transit counted as availableChannels vs 3PLInventory1–4% oversellsHigh
4Month-end cutoff / timezoneChannels vs GL vs 3PLOps1–4% mis-timedHigh
5Refund period + no restockChannels vs GL vs 3PLMargin1–4% rev · 1–3% invHigh
6Sales tax booked as revenueChannels vs GL vs filerRevenue3–6% of revenueMed
7Margin omits fees + fulfillmentFee/carrier vs ERPOps8–20 pts SKU marginHigh
8No canonical product masterAll systemsOps2–6% unit varianceMed
9Gift cards booked on issuanceShopify vs GLRevenue3–8% of Q4Med
10Bundles not decomposed to BOMChannels vs 3PL vs GLMargin2–5% COGSMed
11Discounts/promos gross vs netChannels vs GLRevenue8–15% of grossMed
123PL shrink not booked to GL3PL vs GLInventory0.5–2% shrinkLow
06 — The 90-day fix roadmap

From findings to a source of truth.

The audit is the diagnosis; this is the prioritized path to a warehouse everyone trusts — highest-dollar reconciliations first, then the durable pipelines that keep them true.

Phase 1
Weeks 1–3
Outcome: one reconciled March number, verified.
  • Stand up the landing zone and warehouse; ingest Shopify, Amazon, the 3PL, QuickBooks, and processor settlements.
  • Build the master SKU crosswalk and the settlement-level payout decomposition.
  • Deliver the reconciled revenue, margin, and inventory numbers from this report — signed off against the bank.
Phase 2
Weeks 4–8
Outcome: the high-dollar defects fixed at the source.
  • Landed-cost layer capitalizing freight/duty/brokerage into COGS; contribution margin rebuilt net of fees + fulfillment.
  • Ship-confirm revenue recognition in one timezone; same-period refund reserve; order dedup on native IDs.
  • One per-location allocation ledger driving available-to-sell across every channel.
Phase 3
Weeks 9–12
Outcome: a close that ties, monitored.
  • Rebuild the P&L, margin-by-channel, and inventory reports the team actually uses on the reconciled layer.
  • Tax-liability split, gift-card deferral, bundle BOM, and 3PL shrink journals folded in.
  • Variance alerts on every source-to-truth divergence; month-end close cut from days to hours.
07 — What a Data Truth Audit delivers

What you receive.

A written Data Truth Report

This document, for your own systems — ~15–20 pages, delivered in about three weeks, yours to keep whether or not you build with us.

A source-by-source data map

Every system, what it’s authoritative for, and exactly where it diverges from the others (§2).

Your top KPIs, reconciled

Revenue, margin, and inventory walked to the true figure with the root cause of every gap (§4).

A prioritized 90-day roadmap

The fix sequence and a fixed-price build quote — highest-dollar reconciliations first (§6).