Retrieval · comparison

API vs SFTP connectors for benefits data

Use an API connector when you need to write deductions or new hires, and an SFTP connector when the API doesn't expose the data you need or the source only offers file exports — but file connectors are read-only in Bindbee's catalog, bound to the customer's export template, and not offered at all by Apideck, Knit or Unified.to.

By Eli Wilson · Published 2026-10-01

Short answer: use an API connector when you need to write (deductions, new hires) or want the platform to decide which fields you get. Use an SFTP (file-based) connector when the API doesn’t expose the data you need, or when the source only offers file exports. Not every unified API offers file connectors: of the seven compared below, Bindbee lists SFTP connectors (not tier-gated in the sources reviewed), Merge on Professional and Enterprise, Kombo from its Scale plan up, and Apideck, Knit and Unified.to document none. In Bindbee’s catalog, SFTP connectors are read-only: none of them support write. The fields you receive are whatever the customer’s export template includes.

Bindbee states 102+ HRIS integrations; its public integrations page at docs.bindbee.dev lists 90 HRIS/payroll connector entries today. Most are API connectors, and a minority are SFTP.

The clearest example is ADP Workforce Now in Bindbee’s catalog: the ADP API connector does not provide benefits models, while the ADP SFTP connector provides all four. For a benefits platform, the file route is sometimes the only way to get enrollment data from a given employer.

Comparison: file-based options by vendor (as of 1 Oct 2026)

Finch’s assisted connections are a different mechanism from SFTP. They are included here because, like file connectors, they trade freshness for coverage. “Not documented” means the named pages were checked and no file connector was found; it does not rule out a custom arrangement.

Where each approach wins

This table uses Bindbee’s connectors, the one catalog here with both kinds for the same systems.

The ADP example

ADP Workforce Now appears twice in Bindbee’s integration list. They behave differently for a benefits platform:

Other vendors split ADP differently. Finch’s Benefits API (Early Access) starts with ADP WFN; Knit documents deduction enrollment on ADP WFN; Unified.to lists Deduction read on ADP WFN; Apideck lists ADP WFN as a live HRIS connector. None of those is a file route. In Bindbee, if you need both enrollment read and deduction write for an ADP employer, that means two connections: SFTP for reading elections, API for writing deductions. Each is an active connection under Bindbee’s per-connection pricing, so include both in your cost model. The two connections also give you two records per employee. Pick a join key the employer’s export and payroll both carry, such as employee number, and match on it.

Other systems also appear more than once on the integrations page, including Paychex, Paycom, Paylocity, Oracle and Rippling. Check the model availability matrix for each variant rather than assuming that the two variants match.

Running SFTP connectors well

1. Agree on the export template first. List the fields you need (dependents with DOB, coverage tier, plan identifiers, termination date) and get the employer’s admin to include them. Missing fields cannot be added later on the Bindbee side.
2. Match schedules. If the customer exports weekly, daily syncs don’t make the data daily. Ask for an export schedule that matches your deadlines (see sync frequency).
3. Watch for files that stop arriving. A missing file doesn’t fail the sync: Bindbee re-reads the last file it has, and nothing errors. Track when each connection’s data last changed and alert when that gap passes the agreed export schedule. Use connector_sync_error (Pro and Enterprise) for genuine failures, such as a file that doesn’t match the template, and because exhausted webhook deliveries aren’t redelivered, run a scheduled modified_after check as well.
4. Keep the raw row. include_raw_data returns the original record next to the unified one, which helps when a ben-admin export uses its own labels for plans or tiers.
5. Plan the write path separately. If the employer also needs deductions written, that has to come from an API connector to their payroll system, or from your own file delivery outside Bindbee.

Limits by vendor

• Bindbee: no SFTP connector supports write, including the four ben-admin systems and all SFTP payroll variants. Data is bound to the export; Bindbee normalizes what arrives and cannot request fields the file doesn’t carry. A missing file looks like frozen data, with no error, so monitor file freshness yourself. Not real-time: 24-hour default sync (Pro listed as 12–24 hours, Enterprise configurable), webhooks fire after a sync, and a file source can be no fresher than its export. Bindbee does not generate EDI 834s for carriers, though it can ingest legacy 834 files delivered over SFTP and serve them as normalized API data (EDI 834 explainer). Gusto is not listed, by API or SFTP.
• Merge: SFTP is not available on the Launch plan, so a buyer needing it is on a Professional or Enterprise contract.
• Finch: assisted connections refresh every 7 days and write back within 2 business days.
• Kombo: SFTP requires the Scale plan or above.
• Apideck, Knit, Unified.to: no file connectors documented. If a customer’s benefits data is only available as an export, you would need your own file pipeline alongside the unified API.

Frequently asked

Can I write deductions to an employer whose only connector is SFTP?

Not through the file connector. In Bindbee, you need an API connector for their payroll system with deduction write Supported.

Why does ADP SFTP have benefits data when ADP API doesn't?

They are separate connectors with separate model coverage, and the model availability page lists them separately. Always check the specific variant.

Source limits

Sources checked

Wrong or out of date? payrolldataspecs@gmail.com — corrections are free. How claims are scored: methodology.