Frequently asked questions
Answers to the questions MCA brokers, ISOs, and funders ask most often about how MetrikData reads bank statements, what it detects, and how your data is handled.
Product
What does MetrikData do?
MetrikData reads merchant bank-statement PDFs and turns them into a reviewable underwriting picture for merchant cash advance (MCA) decisions. It extracts every transaction and daily balance, then surfaces the metrics underwriters care about: deposit volume, true external revenue, negative-balance days, NSF activity, and existing MCA positions.
Every number is backed by the underlying transactions, so you can click through from a metric to the exact deposits or debits that produced it.
Which banks and file types are supported?
MetrikData is built to read statement PDFs from every major US bank — including Chase, Bank of America, Wells Fargo, Citi, U.S. Bank, PNC, Truist, Capital One, TD Bank, and regional and community banks. It handles both text-based PDFs and scanned/image PDFs via OCR.
Uploads must be PDF. If a particular bank’s layout produces an unusual result, the parsed transactions remain fully editable before you run an analysis.
How does MetrikData detect MCA stacking and debt burden?
MetrikData scans debits for the repayment signatures of MCA funders — recurring daily or weekly withdrawals, known funder and ISO naming patterns, and ACH descriptors consistent with advance repayment. Matched debits are grouped into positions and labeled with a confidence level.
From those positions it estimates the merchant’s existing daily/weekly debt service so you can see the stacking burden against deposits, rather than guessing from a raw transaction list.
What is the “external revenue” basis and why does it matter?
Gross deposits overstate real revenue because they include transfers between the merchant’s own accounts, returned items, loan and advance fundings, and other non-revenue credits. MetrikData filters those out to estimate true external revenue — money actually coming in from customers.
Underwriting on external revenue rather than gross deposits gives a more honest view of capacity and reduces the risk of over-advancing.
Is MetrikData a decision engine? Does it approve or decline deals?
No. MetrikData does not output a yes/no, an approval, or a recommended advance amount. It surfaces reviewable formulas and the evidence behind them — the metrics, how each was calculated, and the transactions involved.
The underwriting decision stays with you. MetrikData makes that decision faster and better-informed; it does not make it for you.
Data & security
How is my data secured and how long is it kept?
Uploaded statement PDFs are stored in private, per-workspace object storage. Access is gated by workspace authentication and row-level security, and data is encrypted in transit (TLS) and at rest. One organization cannot read another’s documents.
Statements and their derived data are retained for up to 3 years to support audit and re-review, after which they are eligible for deletion. See the Privacy Policy for full detail.
Do you use my uploaded statements to train AI models?
No. Uploaded statement content is used only to produce your analyses. We do not sell your data and we do not use statement content to train third-party models.
Pricing & data
How much does it cost? Is there a free trial?
You can start free — new workspaces get 4 free statements to run a full analysis end to end before choosing a paid plan. After the trial, paid plans include a monthly statement allowance.
See Pricing for current plans and limits.
What export formats are available?
Reports export as PDF for sharing and filing, and transactions export as CSV for your own spreadsheets or downstream systems. CSV is available both as a raw transaction list and as a tagged version that includes MetrikData’s category and lender labels.
Can I correct the parsed data before running an analysis?
Yes. After a statement is parsed you can review and edit transactions and balances before analysis. This is useful for unusual layouts or for marking transactions that the parser categorized differently than you would. Re-running the analysis recomputes every metric from the corrected data.