Clause Forensics Sample

In Year 1, when the platform was first signed, the service level lived in the body of the master agreement and read as a straightforward promise: miss 99.9% and the customer gets money back. At that point the vendor's own ops team published a clean monthly uptime report and emailed it to the customer's IT lead, so the credit was effectively self-documenting and enforcement was almost automatic.

At the first renewal in Year 2, the credit language moved out of the main agreement and into an order-form exhibit, and two phrases were added that shifted all of the work onto the customer: the request now had to be 'in writing within fifteen business days' and had to include 'log data evidencing the claimed periods of Unavailability.' The proactive monthly report quietly stopped being a contractual deliverable; uptime became something the customer had to go find rather than something the vendor delivered.

By the Year 3 renewal, measurement was redefined to be 'determined solely by reference to Provider's monitoring systems as published on the Provider status page,' and the status page itself moved to a new URL with only a rolling 90-day incident history. A 'sole and exclusive remedy' line and a 20% cap were also folded in. Each change looked administrative in isolation, but together they converted a clear refund into a time-boxed claim the customer has to assemble from a source the vendor controls and periodically overwrites.

Today the live agreement exists only as the original MSA plus two renewal order forms and an amendment cover page, and the credit terms are spread across all of them. No single document states the current rule end to end, and the 90-day status page means evidence for older outages has already aged out. The entitlement is still technically in force, but reconstructing it now requires stitching three documents together against a data source that no longer reaches back far enough.

The clause breaks at the handoff between IT operations, who feel the outage, and accounts payable, who pay the invoice. IT operations knows the platform went down because they fielded the tickets, but nothing in their workflow translates an incident into a contractual credit request. Finance approves the monthly invoice on its normal cycle without ever being told an outage occurred, so the two halves of the claim, awareness of the downtime and control of the money, never meet inside the 15-business-day window.

The evidence requirement compounds the timing problem. Filing requires log data that lines up with the vendor's status page, and pulling and reconciling that data competes with live incident response and every other vendor the same small team manages. By the time anyone has bandwidth to assemble it, the deadline has usually passed, and because the vendor controls the measurement and only retains a rolling 90-day history, the supporting evidence may already be gone. The sole-remedy language then closes the door: there is no fallback path once the window lapses.

There is also a relationship dynamic that suppresses claims. The same account representative who would receive the credit request is the person the customer relies on for renewals, escalations, and roadmap asks, so teams quietly decide a few thousand dollars is not worth the friction. Multiply that decision across every outage month and the credit is never exercised at all.