Merchant Cash Advance. Underwrite the receipts. Understand the merchant.

Canada: working capital and existing debt.

ISED’s 2025 Credit Conditions Survey covers Canadian businesses with 1–99 employees. The survey of 1,812 respondents reports that 20% requested debt financing; 97% of debt requests received full or partial approval. The published categories do not isolate MCA. ^[1]

Among Canadian small businesses requesting debt financing, 45% intended working or operating capital, 24% debt consolidation and 22% fixed assets. These are financing purposes, not shares of funded dollars. ^[2]

For an MCA assessment, this context makes the use of funds consequential: supporting a trading cycle, buying an asset and replacing existing obligations require different evidence. Establish what cash will reach the merchant, which payments continue, and which obligations are actually settled. A broad approval statistic cannot establish that an individual advance is affordable.

United States: a separately measured MCA channel.

The Federal Reserve Banks’ 2026 report on the 2025 Small Business Credit Survey reports MCA applications from 8% of employer firms, alongside 26% for loans and 24% for lines of credit. Responses overlap. These are application shares, not market size, approval rates or defaults. ^[3]

The US results establish a distinct channel for merchant finance, but a product’s popularity is not evidence of affordability. Assess the merchant’s operating receipts, existing obligations and effective agreement rather than assigning capacity from a national application rate.

Canada and the US: compare the context, preserve the definitions.

These datasets answer different questions. Read each measure within its own population and product coverage. Comparing Canada’s broad debt request rate with the US MCA application rate would mix different products and populations. Neither subtraction nor averaging would produce a meaningful MCA market comparison.

For a two-country portfolio, retain country, currency, agreement type and source coverage as separate fields. Build country-specific evaluation cohorts before testing whether a pooled model improves results. State any exchange-rate convention when aggregating monetary exposure; do not combine CAD and USD amounts as though they were one currency.

Start with the agreement, not a generic loan template.

An MCA can use a percentage of receipts, estimated periodic withdrawals subject to reconciliation, or other documented collection terms. Capture the purchase price, cash actually delivered, purchased amount, eligible-receipts definition, collection frequency, exclusions, fees and applicable adjustments. Identify whether the assessed arrangement is a receivables purchase or another financing structure under the applicable agreement and jurisdiction; a product label does not resolve its legal treatment. ^[5]

For a concrete example, SG Building Blocks’ January 2024 SEC-filed agreement lists a USD 1,180,000 purchase price, USD 1,062,000 net funds provided and USD 215,575 received directly by the merchant. It specifies 5% of receipts and an initial USD 49,150 weekly estimate subject to reconciliation. These distinct fields show why the headline amount cannot stand in for spendable cash. ^[4]

The difference between the two net-funding fields is USD 846,425. Review the associated instructions and supporting transfers before assigning that difference to a specific obligation. This is one disclosed agreement, not standard market terms. Its mechanics illustrate an evidence requirement, not a recommended financing structure. ^[4]

What Amlok can provide across the MCA lifecycle.

Insights brings an application and its permitted financial evidence into an agreed full-underwriting workflow. Verification, calculations and analytical investigation can be included stages. For MCA use, the implementation can organize merchant identity, source coverage, reconciled receipts, existing obligations and the proposed agreement into an evidence-backed assessment. The lender retains the policy, approval decision and funding controls. ^{[6] [7]}

The MCA research specification focuses evaluation on eligible receipts, remittance burden, volatility, collection coverage and contract-defined outcomes. Scope these calculations against your agreements and supported data sources. Preserve evidence references and unresolved differences so an analyst can inspect why an amount was included or excluded. Do not turn an unavailable calculation into a guessed value. ^[5]

Signal uses an approved baseline and newly available authorized evidence to support monitoring and reassessment. It can help your team review persistent receipts changes, new supported obligations and unusual remittances within the configured workflow. Freshness depends on source capability and consent; requesting daily monitoring does not make every provider refresh daily. ^[7]

Build an evidence package around the merchant’s operating cycle.

Use enough history to represent the business’s normal calendar and the proposed assessment window. A seasonal retailer may need comparable prior-season evidence; a recently opened location needs a separate treatment for limited history. Record the accounts, storefronts, currencies and processors included. Missing channels are a coverage limitation, not zero revenue.

The following map is a proposed implementation specification. Confirm supported connections, fields and document workflows before promising a particular assessment. Obtain appropriate authorization and retain the source, economic date, posting date and retrieval time for each relevant record. ^{[7] [5]}

Follow the sale all the way into usable cash.

First link orders or captured sales to processor transactions; then link settlement batches to bank postings. Carry refunds, chargebacks, fees and reserve changes through the bridge. A payout can cover several sales days, and a bank posting can arrive after a weekend. Compare economic periods before interpreting a deposit gap as a decline in trading. ^[5]

Use a reconciliation ledger with matched, partially matched and unresolved items. Keep the unmatched amount and the reason visible. Avoid counting the same receipt from the commerce system, payment processor and bank three times. Measure coverage by value as well as count: a few unmatched large settlements can matter more than hundreds of small matched transactions.

Gross receipts, eligible receipts and available cash answer different questions. Eligibility follows the contract. Available operating cash additionally reflects settlement deductions and business expenses. Neither should silently substitute for the other when calculating burden or an estimated collection.

Measure the pressure on receipts, with explicit denominators.

Use a short set of auditable measures rather than an unexplained composite score. Every measure needs a defined window, source coverage and treatment of adjustments. Compare the merchant with its own comparable periods before introducing a peer benchmark. Any policy cutoff is selected and tested by the financing institution. ^[5]

• Eligible receipts: supported receipts after the specific agreement’s exclusions. Reconcile to sales and settlements without assuming that processor fees reduce the contractual base.
• Remittance burden: actual remittances divided by eligible receipts for the same window. Show the proposed arrangement separately from combined existing financing cash outflows.
• Operating residual: reconciled available cash less operating cash costs and financing outflows. Track upcoming taxes, payroll and supplier commitments that are absent from past cash payments.
• Downside variation: compare rolling windows and corresponding seasonal periods. State days covered and distinguish closed trading days from missing source records.
• Collection coverage: actual collections against the contractual eligible share or due amount after approved adjustments. A valid reduction in variable remittance is not automatically a missed loan installment.
• Outcome measures: record shortfall, contract breach, restructuring, settlement, recovery and realized loss separately, with stated exposure and observation horizons.

Worked example: a 20% sales decline changes the whole cash bridge.

Consider an invented merchant with CAD 100,000 monthly captured sales, CAD 4,000 refunds, CAD 3,000 processor fees and a CAD 3,000 increase in withheld reserves. Assume all resulting payouts settle during the month, with no additional timing differences. Available settlement cash is CAD 90,000. After CAD 65,000 operating cash costs and CAD 4,000 other financing outflows, CAD 21,000 remains before the proposed arrangement.

Assume the agreement defines eligible receipts as sales less refunds and collects 12% of that base. Eligible receipts are CAD 96,000; the proposed remittance is CAD 11,520, leaving CAD 9,480. In the downside month, sales and proportional refunds and fees fall 20%, reserve withholding remains CAD 3,000, and operating cash costs fall only to CAD 58,000. The business is left with CAD 184 after the variable remittance.

If an estimated collection instead stays at CAD 11,520 during that downside window, the residual becomes negative CAD 2,120. That creates a CAD 2,304 difference from the assumed 12% share. Whether and how it is reconciled depends on the actual agreement and approved process. This scenario does not assume an automatic refund or authorize a change in withdrawals.

Separate the factor, the cash delivered and the time at risk.

In a second invented example, a CAD 50,000 purchase price at a 1.30 factor produces a CAD 65,000 purchased amount. If CAD 2,000 is withheld in fees, the merchant receives CAD 48,000 before any other deductions. The CAD 17,000 difference between that net receipt and the purchased amount equals 35.4% of cash delivered. It is a simple amount comparison, not an annualized rate.

A factor does not encode timing. Under an intentionally simplified constant-receipts scenario, CAD 500 per collection day reaches CAD 65,000 in 130 collection days; CAD 350 takes 186 days with a smaller final collection. Neither is a guaranteed contractual term. An annualized measure needs the full dated cash-flow schedule, fees, actual or clearly specified assumed timing, and the applicable disclosure method. ^[5]

Treasury’s January 2025 small-business financing brief highlights difficulties comparing products with different fees and variable repayment arrangements. Operationally, make those differences visible in the assessment rather than presenting unlike structures as interchangeable prices. ^[8]

For renewals, distinguish the new gross amount, deductions, any supported payoff and fresh cash delivered. Keep the original and replacement exposures linked without double counting collections. Confirm extinguishment using payoff and settlement evidence before removing an old obligation. A new advance can temporarily lift balances while increasing the merchant’s future burden.

Monitor changes without confusing missing data with distress.

After the approved baseline exists, configure Signal around the evidence your team can act on: persistent receipt changes, unusual reserve movements, higher refunds, new supported financing obligations and deviations between expected and observed remittances. Pair each alert with its comparison window, freshness and source coverage. An expired connection should be handled as unavailable evidence, not a sales collapse. ^{[7] [5]}

For multiple advances, reconcile the merchant’s combined outflows while preserving each agreement’s own eligible base and adjustments. A counterparty string in a bank feed is a lead for review, not proof of undisclosed financing. Obtain supporting terms or records before assigning an obligation. Likewise, a new payment processor may explain a channel shift rather than declining sales.

Route supported changes to the institution’s existing review process. Keep a record of the analyst’s conclusion and any supplied contract amendments so future comparisons use the correct baseline. Amlok provides risk outputs and evidence; the institution owns merchant communication, collection changes, funding and recovery actions. ^[6]

Connect the assessment to your operating systems.

Begin with one defined MCA product and its documented policy. Confirm supported processors, bank sources, accounting records and agreement formats. Create the business and application, obtain authorization, synchronize records and inspect the ready input manifest for the intended period. Freeze the evidence window before starting the approved Insights underwriting run. ^[7]

Include the agreement version and available obligations in the assessment specification. Retain calculation inputs, source identifiers, exceptions and the immutable risk snapshot alongside the institution’s decision. Verification and analytical investigation operate as included stages within the permitted product workflow, subject to agreed scope. ^[6]

For an originated or existing portfolio, use a ready approved monitoring baseline from the completed assessment or agreed onboarding process. Configure the permitted cadence and rules, consume signed result notifications and retrieve the supporting evidence. Have distinct handling for incomplete data, a supported risk change and a disputed interpretation. Test source expiration and late corrections before expanding coverage. ^[7]

Evaluate MCA performance on MCA evidence.

Build a merchant-level, time-separated evaluation with only evidence available at the decision date. Keep later collections, reconciliations and payoff information out of the historical input snapshot. Segment results by trading model, seasonal exposure, channel coverage and agreement mechanics. Multiple records or renewals for the same merchant must not leak into both training and holdout cohorts. ^[5]

First audit the receipt bridge: matching precision, missed matches, duplicate revenue, unexplained settlement value and contract-field accuracy. Then evaluate cash-flow calculations and alert relevance against reviewed cases. Compare analyst effort on the same evidence, report incomplete-file coverage and retain corrections; a quicker output is not an improvement if it omits material obligations.

For performance outcomes, define the denominator and maturity window before examining results. Report contractual adjustments separately from missed obligations, and actual recoveries separately from expected collections. Do not treat unsettled recent advances as successful completions. Evaluate uncertainty and cohort differences before claiming a portfolio benefit.

Amlok’s published experiments cover specific structured-loan and financial-evidence retrieval tasks. They do not establish MCA default prediction, reduced losses or a safe advance size. An MCA deployment claim needs representative validation of the workflow and population in question. ^[9]

• Define: one agreement family, an evidence window, outcome definitions and accountable reviewers.
• Replay: reconstruct historical decisions and audit receipt matching and calculations.
• Shadow: compare Amlok outputs with the current assessment and monitoring process.
• Expand: add sources or merchant cohorts only after reviewing coverage, errors and operational usefulness.