Bank feeds vs file import: what to know
The trade-offs between direct custodian feeds and file-based import, and how duplicate detection keeps data clean.
- Direct feeds automate ingestion; file import is universal and immediate.
- Duplicate detection prevents double-counting across syncs.
- An aggregator layer bridges many banks with one integration.
The trade-off
Direct custodian feeds pull data automatically but require integrations bank by bank, or an aggregator. File import (CAMT.053, MT940) works with any bank today and needs no integration — at the cost of a manual step.
Keeping it clean
Whichever route, the same statement must not be counted twice. A stable duplicate hash across syncs detects and skips transactions already imported — the difference between a feed you trust and one that creates chaos.
Formats and what they carry
Bank and custodian data arrives in a few common formats. SWIFT MT messages such as MT535 (statement of holdings), MT536/537 (transactions and pending trades) and MT940/MT950 (cash statements) are still widespread; ISO 20022 messages such as camt.053 increasingly replace the cash statements. Many private banks also deliver proprietary CSV or PDF files, and some offer APIs. Each format carries different information: positions with valuations, cash movements, corporate actions or just end-of-day balances.
Before connecting a feed, clarify what it actually contains, how often it is delivered, how instruments are identified (ISIN, internal codes) and how corrections are sent.
Reconciliation and exceptions
Automated feeds remove manual typing, not responsibility. Reconcile positions and cash against the bank's statements regularly, flag unknown instruments and price gaps, and review breaks rather than overwriting them. Keep the original files as evidence, so that every reported figure can be traced back to the source. For assets that no bank reports — private equity, real estate, loans within the family — combine the feed with manual valuations, each carrying its own date.
Practical checklist
For each bank, document the format, frequency, content and contact for technical issues. Map instrument identifiers and account numbers to your data model. Test the feed with historical data before relying on it.
Reconcile positions and cash daily or at least before each report, review breaks and unknown instruments, and keep the original files. Monitor delivery so that a missing file is noticed the same day, and add manual valuations for assets no bank reports.
Common mistakes
The most common mistake is trusting a feed blindly once it is connected. Feeds break silently: a file is not delivered, a new instrument is not mapped, a corporate action is booked differently from expectations. Without daily or at least pre-report reconciliation, these errors surface only when a family member questions a figure. Another mistake is overwriting breaks instead of investigating them, which destroys the audit trail.
Offices also underestimate the effort of instrument mapping and historical data. Moving to automated feeds without importing acquisition dates and costs leaves gaps in tax lots and performance history. Finally, keep the original files: when a bank later corrects a statement, the original is the only way to explain what was reported at the time.
Example from practice
After connecting a new custodian feed, the office reconciles positions daily for the first month. On day nine a spin-off is booked as a new position without cost basis; the reconciliation flags it, the team records the corporate action correctly and the tax lots remain accurate. Without the check, the error would have surfaced only in the year-end tax report.
FAQ
- ISO 20022 CAMT
- PSD2 / open banking
Related guides
Compliance, built into the platform
FLIORE is a family office platform with KYC and compliance built into the core.
Explore FLIORE