Support When It Matters

Support tiers are described in generic language. In this category the only question that matters is what happens when payroll cannot run.

Severity defined by the calendar

Severity one: payroll cannot be produced, or produces wrong figures, with a deadline approaching.

Severity two: a period cannot be closed, or the export is failing, with time before the deadline.

Severity three: a group cannot record time.

Severity four: everything else.

Write these definitions into the contract rather than accepting the vendor's, which are usually about system availability rather than about your calendar.

What to specify

Response time by severity, with a remedy if missed.

Whether response means acknowledgement or engagement. Most tiers mean the former, and an acknowledgement does not run payroll.

Availability window relative to your time zone, not theirs.

An escalation path with named roles, and a route that does not depend on a portal being up.

Coverage during your peak, which for payroll is the same days every period and is entirely predictable.

Telling a real commitment from a tier name

Ask what happened the last time a customer raised a severity one. A vendor with a process answers; one without describes the tier.

Ask for the measured response times, not the target.

Ask a reference customer, who will tell you plainly.

Ask what is excluded: configuration questions, integration failures at your end, anything they classify as training. The exclusions are where support tiers become narrow.

What to measure

Tickets by category, internal and vendor.

Time to resolution for anything blocking payroll.

Vendor response against the contracted time, which is the evidence for a renewal conversation.

Questions per person in the first month after a change, which tells you whether a change was well made.

A rising internal load after an upgrade is the signal that something in the interface changed, and it arrives before anyone reports the change itself.

Your own first line

Most questions are configuration and use, and they should not reach the vendor.

A named internal contact, which is the administrator role.

A short answers page for the recurring dozen questions, which covers most of the volume.

Log what people ask. Repeated questions are a design finding, and fixing the cause removes the load permanently.

A rising internal load after an upgrade is the earliest signal that something in the interface changed, arriving before anyone reports the change itself.

Test ownership across suppliers

A connection adds a second support boundary. This integration page provides a concrete scenario for asking who diagnoses failed transfers.