The SLA your payroll deserves.

A failed sync, from your chair

The Product page walks a failure through the platform (→ When something breaks). Here is the same event from your side of the screen — say the 06:00 sync fails on the 24th, the day before cutoff.

1. The 24th

   The failure finds us first. The failed run alerts OmniSync engineers before anyone else. You didn’t have to notice anything.

2. Then it finds you — from us, not from a payslip. An email or Slack alert names the failed run. Not a discrepancy someone spots at cutoff. Not a dashboard you had to remember to check.

3. Same day

   You don’t open a ticket. The re-run is in the SLA, not in a queue: the failed sync is re-run the same business day whether you ever contact us or not. Recovery is the default, not a favor you request.

4. You can watch, if you want. The failure and its re-run both land in your dashboard and sync history — shown, not smoothed over.

5. Your to-do list from all of this: usually nothing. If the alert also flags quarantined records, fix them at the source and hit Re-run — or let the next sync pick them up. If the gap is in the mapping, that fix is ours, not yours.

   How quarantine works, on the Product page →

6. Before cutoff, the reconciliation report accounts for every record — synced or held — before you sign off.

The version of this story you’ve lived before — the one where you find out at cutoff — is the one this SLA is written to end.

How the number is measured ¶

Most SLAs go vague exactly here. The definition, in plain terms:

Per pair.

Each integration pair is measured on its own runs — a good month on one pair never papers over a bad month on another. Run two pairs, and each carries its own number and its own credits.

Monthly.

Your pair’s successful scheduled syncs as a share of the scheduled syncs due that calendar month.

Quarantine is not failure.

A run that completes with records quarantined by validation is a successful run — per-record quarantine is the platform doing its job, and it never blocks or fails a sync. How quarantine works →

The re-run has a deadline — and it’s measured.

A failed sync that’s re-run successfully the same business day counts as met. Miss the recovery clock, and the miss stands in that month’s rate: a late fix repairs your payroll, not our statistic.

Exclusions leave the math.

Runs excluded below — a vendor outage, revoked client-side access, a maintenance window — are removed from the month’s calculation, not counted as successes. And they show as excluded in your sync history, so you can audit what was left out, not just what was counted. (Quarantine is the exception: those runs stay in the count, as successes.)

Two more terms, pinned down — because “same business day” is a promise only if you know whose day:

Business day

a Philippine business day: Monday to Friday, excluding Philippine public holidays. All times on this page are PHT — the same clock your support hours are quoted in. The same-day re-run and next-business-day first response are both measured against this calendar.

Payroll window

the days around your payroll cutoffs, agreed at onboarding from your payroll calendar and named in your contract — and updatable by written notice when your calendar changes. Premium’s emergency on-call line is active throughout it.

This is the same definition your dashboard uses: one number, one definition, computed where you can read it. Your dashboard is the status page for the only integration you care about — we don’t point you at a platform-average status page.

Service credits — what they are, and what they aren’t ¶

The table is the easy part. Here is the part a burned buyer actually checks:

• The number can’t be argued about. The rate the credits key off is the one computed in your dashboard (above) — there’s no dispute over a vendor’s private measurement, because the measurement isn’t private.
• And there’s nothing to file. Most SLA credits come with homework: notice the miss yourself, claim it within 30 days, attach your logs. Ours is measured in a dashboard we both read — if a month closes below 99.5%, the credit is applied against your next invoice. No claim form, no filing window, no deadline to catch us by.
• How credits reach you. Against your next monthly invoice — the remedy is written into the contract, not negotiated after a bad month. Paid the year up front? Credits accrue and net against your renewal invoice, and if you don’t renew, they’re refunded in cash. Prepaying never weakens the remedy; it only changes where the credit lands.
• What a credit is — and isn’t. A price on our failure, not compensation for yours. A 25% credit doesn’t un-pay an overpayment — no service credit anywhere does, whatever the brochure implies. That’s why this page leads with the same-day re-run. The failed syncs are still re-run the same business day, credit or no credit.

What the SLA doesn’t cover — and what we still do when it applies ¶

The exclusions list is where most SLAs quietly take back what the headline promised. Ours is six items long, and five of them carry a commitment anyway — because when the credit doesn’t apply, the question that matters is whether the vendor still shows up.

An exclusion never means silence: when a run fails — whichever row explains it — it shows in your dashboard and you hear about it from us, never from a payslip. And if something on this list feels like a loophole, bring it to the scoping call — walking the exclusions against your actual stack is a better test than reading them.

Support: one SLA, two tiers ¶

First, the distinction the table below depends on: the tiers set how fast a human answers you. They don’t set how fast a failure gets fixed. A failed sync alerts OmniSync engineers first and is re-run the same business day — on Standard exactly as on Premium. That’s an SLA obligation, not a ticket you file, and no queue sits in front of it. There’s no severity matrix to argue, either: everything we run is the sync feeding your payroll, and its urgency is set by your calendar. You will never have to argue a failure into the right priority tier to get it re-run today.

The 99.5%, the re-run, and the credits are identical on both tiers — paying less never buys a weaker promise. What the tiers buy is human depth:

The rows, in plain English:

• Hours — when support is on duty. Standard covers Philippine business hours; Premium covers 7:00–22:00 PHT, every day — early enough for a cutoff-morning problem, late enough for an end-of-day one. And the timezone is the point, not a constraint: these hours sit where your Philippine payroll’s cutoffs sit, so a human is on duty at the moments that decide a payslip — wherever your HQ is.
• First response — a person who has read your ticket, never an auto-acknowledgement. The clock starts when your ticket reaches us and runs within your tier’s support hours. Next business day on Standard; 4 hours on Premium.
• Payroll-window on-call — Premium’s emergency line, live during your payroll windows, for the problems that can’t wait in any queue. In plain terms below.
• Support queue — on Premium, priority means exactly that: Premium tickets are worked first, ahead of the standard queue. It changes how quickly a person is on your problem — never the promise.
• Success manager — Standard shares one; Premium has a dedicated one who knows your integration.

No metered hours, no per-ticket fees — on either tier. Support is in the monthly fee, full stop.

Who reads your ticket — and why questions are free ¶

The usual enterprise support path: a first responder working from a script, an escalation ladder, and your problem retold three times before it reaches someone who can see your integration. Not here. Your ticket lands with the engineering team that builds and operates the integrations — the same people your failure alerts reach first. There is no tier 1 to escalate past, because there is no tier 1. A support ticket and a failure alert end at the same desk — the desk on the hook for the SLA above.

Your first cutoff is the one we watch hardest. Every go-live is followed by 30 days of hypercare — heightened monitoring, on either tier. How implementation works →

And you’ll never ration a question. No metered hours, no per-ticket fees, on either tier. Vendors who bill by the question teach payroll teams to stay quiet until small problems are emergencies. We need the opposite: tell us early, tell us often — it never touches your invoice.