Use cases · guide

Per-connection vs per-API-call pricing for unified HRIS APIs: how to model the cost

For a benefits platform with 30 to 50 customers and daily syncs, per-call pricing is usually cheaper at list prices for read volume alone (Unified.to's Grow plan matches $65 per connection at about 65,000 calls per connection per month), while per-connection pricing wins on predictability, and coverage of your customers' payroll systems and the write you need narrows the field more than price does.

By Eli Wilson · Published 2026-10-01

Short answer: for a benefits platform with 30 to 50 customers and daily syncs, per-call pricing is usually cheaper at list prices for read volume alone, and per-connection pricing wins on predictability. Using published numbers, Unified.to’s per-call Grow plan ($750/month including 750,000 calls, then $1.00 per 1,000) costs the same as $65 per connection (Merge Launch and Finch Starter list price) at about 65,000 calls per connection per month. A daily full read of a 10,000-employee employer is far below that. But the cheaper bill only matters if the vendor covers your customers’ payroll systems and the write you need, and that narrows the field more than price does.

Seven vendors, three pricing units: Bindbee, Merge, Finch, Kombo and Knit price per connection or connected account; Apideck prices per active consumer (one customer, unlimited connections); Unified.to prices per API call with unlimited connections. Knit says it also offers API-call pricing, and Unified.to offers connections-based plans on Enterprise.

How the seven vendors price (as of 1 Oct 2026)

Plan gates that matter for a benefits product

The cost formulas

If you pay per call, cost follows how often you read and how many pages each read takes:

monthly cost (per-call) = plan base + max(0, total calls − included calls) × price per call
total calls = Σ over customers (calls per read × reads per month) + billable webhook events

Per-connection pricing makes each customer a fixed cost:

monthly cost (per-connection) = active connections × price per connection (+ any platform fee)

Per-consumer pricing (Apideck) works the same way, with a customer as the unit and published tiers.

Where the breakeven sits

Once a per-call plan is past its included calls:

breakeven calls per connection per month = price per connection ÷ price per call

Below that volume, per-call is cheaper. Above it, per-connection is cheaper.

With Unified.to’s Grow overage rate ($1.00 per 1,000 calls, or $0.001 a call) against $65 per connection: $65 ÷ $0.001 = 65,000 calls per connection per month. On Unified.to’s Pro rate ($0.75 per 1,000) the same calculation gives about 86,700; on Scale ($0.50 per 1,000), 130,000. Those higher plans also carry higher base fees, so run the full formula.

Worked math: 30 to 50 customers at list prices

Check for 30 connections: 30 × 65,000 = 1,950,000 calls; $750 + (1,950,000 − 750,000) × $0.001 = $1,950. From 10 accounts up, Merge Launch works out to $65 per account, so the breakeven is the same 65,000 at every count shown.

For comparison, Apideck’s published Launch tiers are $599 for 25 active consumers and $999 for 50, and its Scale tier is $1,299 for 100. Finch Starter is $65 per connection but stops at 15 connections. Bindbee, Kombo and Knit (beyond Start Up’s $499/month) quote on request; Knit’s pricing page bills connected accounts on contract on every plan, so ask what Start Up’s $499 covers.

To compare a quote from a vendor that doesn’t publish prices (Bindbee, Kombo, Knit beyond Start Up), divide it by your active connections and set it against the figures above, together with the plan gates. Merge Launch also includes 3 free accounts, so at very small counts Merge’s total can be lower.

Estimating your own call volume

Estimate calls per connection from your read design. At Unified.to, a list page of 100 employees is one call, so a full employee read of a 10,000-employee employer is about 100 calls. Once a day for 30 days is 3,000 calls; reading five object types daily (for example employees, deductions, benefits, payslips and groups) is about 15,000. That is far below 65,000, so for read volume alone a per-call plan at these rates comes out cheaper than $65 per connection at list. Employer size does not, on its own, push per-call past breakeven.

What does move you toward breakeven is read frequency. A pass-through API like Unified.to has no stored copy, so if you poll hourly instead of daily, calls multiply by 24. Virtual-webhook runs that find new data and native webhook deliveries are billable too. Run the formula with your actual page sizes and polling plan.

What per-connection buys you instead

The case for per-connection or per-consumer pricing is predictability, not a lower bill at every volume. Open enrollment, mass rehires, a large group’s first full sync and reconciliation passes all push call volume up, and none of them change a per-connection bill. Your cost per customer is known when you sign the customer. Check what each plan includes rather than comparing unit prices alone: unlimited calls (Bindbee), unlimited data (Merge, Knit), fair-use call caps (Knit), API-call allowances that aren’t documented (Apideck).

Which option fits which platform?

It depends on what your product has to do, so here is the decision rule rather than one name:

1. Read-only, at most 15 employers, each under 500 employees. Finch Starter ($65 per connection) is an option; so are Apideck Launch ($599 for 25 consumers, one API category) and Unified.to Grow ($750 flat up to 750,000 calls). Compare against the systems your employers run.
2. Read-only at 30 to 50 customers with daily sync. Price a per-call plan with the formulas above alongside per-connection and per-consumer plans. At list prices, per-call or per-consumer can be cheaper for read volume; per-connection wins on predictability.
3. You need deduction write back to payroll. Price only vendors that document it on your customers’ systems: Bindbee Pro, Finch Pro (including its employer fee) and Knit, plus Unified.to if your book is Paylocity-heavy. Merge, Kombo and Apideck don’t document a unified deduction write. Coverage of your customers’ payroll systems will decide this before price does.
4. You need data residency. Bindbee offers US and EU, customer-selected. Merge lists US, EU and AP; Unified.to lists US, EU and AU (data region sync on Scale); Knit states US, EU and India (confirm the plan); Kombo offers multi-region residency on Enterprise only; Apideck hosts in the EU by default with no US region documented.

For a typical Series A benefits platform that needs benefits read plus deduction write, rule 3 is the one that matters. The full selection checklist is in choosing an HRIS integration vendor for a benefits platform.

Questions to settle before you sign

1. What counts as billable? Bindbee bills an “active” connection (authenticated and synced at least once in the period); Apideck an “active consumer” (at least one API call or webhook event that month); Unified.to each list page, native webhook delivery and virtual-webhook run that finds new data. Ask every vendor how they treat a churned customer whose connection is still authorized, a sandbox, or a customer with two payroll systems.
2. Does the price change with features? Webhooks, write, custom fields, data scopes, API categories and benefits models can sit on higher tiers. At Bindbee, Basic/Lite covers the Employee model only, read-only, and webhooks and write are Pro. At Apideck, Launch includes one API category. At Unified.to, data region sync and the BAA are on Scale.
3. Is there a platform fee or a minimum? Kombo has a fixed annual platform fee. Unified.to’s plans start at a monthly base that includes a call allowance.
4. Are there volume discounts? Finch, Kombo and Knit say per-account pricing decreases with scale; Apideck’s published tiers fall from about $24 to $8.50 per consumer. Get the curve in writing.
5. Can you get it in writing? Ask for a written quote for your expected connection count at months 6, 12 and 24, with the tier named.

Limits by vendor

Coverage gaps cost money too. A low unit price only matters if the vendor supports your customers’ systems; every customer you can’t connect is an integration you build or a deal you lose.

Bindbee

• The per-connection price is not published. You will need a quote, and you should get it in writing.
• The cheapest tier won’t run a benefits product. Basic/Lite is Employee-only and read-only. Benefits models, payroll models, write and webhooks need Pro.
• Bindbee supports 102+ HRIS integrations; the public integrations page at docs.bindbee.dev lists 90 HRIS/payroll connector entries today; ask Bindbee for the full list. Bindbee doesn’t list Gusto, and its deduction write is Supported.
• Sync cadence: Basic is listed at 24 hours, Pro at 12 to 24 hours, Enterprise configurable. A connection’s schedule is changed through Bindbee support; confirm whether that changes your tier.

Merge

• Launch is the only published price and it is limited: daily sync, 100 requests/min, 3-day log retention and email support.
• No unified deduction write is documented, and passthrough needs Professional or Enterprise on contract pricing.

Finch

• Starter is read-only, capped at 15 connections and 24 providers, for employers under 500 employees.
• Deductions and the full 250+ provider list are on Pro, whose price is not published, and Finch marks Deductions “Employer Fee Required”.

Kombo

• Price is not published, and a fixed annual platform fee sits on top of per-connection fees.
• Multi-region data residency is Enterprise only. No deduction write is documented.

Apideck

• The API-call allowance and any overage are not documented on the pricing page.
• Launch covers one Unified API category and lacks custom field mapping and data scopes; new connector builds are a paid add-on on every plan.
• No deduction write and no benefits model, whatever the plan.

Knit

• Only Start Up has a published price ($499/month), and Knit’s own guide says it isn’t for teams integrating enterprise apps like Workday. Scale Up and Enterprise are contract-only.
• Start Up syncs every 24 hours with a 100K API calls/month fair-use cap.
• Deduction write is documented on specific payroll systems only (enrollment on ADP Run, ADP WFN, Paycor, Paychex, Paylocity, PrismHR, Avionte and QuickBooks Payroll).

Unified.to

• Cost tracks your read pattern. Hourly polling, large first syncs and virtual-webhook runs that find data all add billable calls.
• Prices are “from” figures ($750+, $1,500+, $3,000+). The pricing table’s rate limits (5,000 / 7,500 / 10,000) don’t state a unit.
• Deduction write is listed only on Paylocity among the US payroll systems checked, so a cheaper bill may not cover the write you need.

Next step

Ask each shortlisted vendor for a written quote at your expected connection counts for months 6, 12 and 24, with the tier named, and estimate your monthly calls from your actual read design before comparing per-call plans.

Frequently asked

Is per-call pricing cheaper for a benefits platform?

Usually, for read volume at daily cadence. At $65 per connection and Unified.to's Grow overage of $0.001 per call, breakeven is about 65,000 calls per connection per month, and a daily read of a 10,000-employee employer across five object types is about 15,000. Per-connection wins on predictability and on what the plan includes.

Which vendors publish prices?

Merge (Launch), Finch (Starter), Apideck (Launch and Scale tiers), Knit (Start Up) and Unified.to (Grow, Pro and Scale base prices). Bindbee and Kombo quote on request.

Source limits

Sources checked

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