Service credit calculation: what the decision maker looks at
The decision maker checks three things in order: whether the service level was actually breached, whether the breach is recorded in data both sides accept, and whether the credit formula in the contract produces one number, not a range. If any of the three is missing, the calculation stalls before it starts. Most disputes at this stage are about missing records, not about the formula itself.
What changes once the calculation is triggered
Once a party raises a service credit claim, the clock in the contract starts running. Notice periods for disputing a credit are usually short, often shorter than either side expects. A late notice can waive the right to challenge the figure entirely, regardless of whether the underlying breach was real. This is the point where a technical disagreement turns into a contractual one, and the two are decided differently.
From this point, the conversation moves away from whether the service was good enough and towards whether the paperwork supports the number. Engineers argue about uptime; the decision maker reads tickets, timestamps and the definitions clause.
Documents the decision maker asks for first
| Document | What it is checked against | Typical timing issue |
|---|---|---|
| Service level definitions (the schedule, not the marketing summary) | Whether the metric breached matches a metric that is actually defined | Definitions changed by amendment are often filed separately and missed |
| Monitoring logs or dashboard exports | Whether the outage or degradation is independently recorded, not just reported | Retention periods on logs are shorter than the notice period for a claim |
| Incident tickets and escalation records | Whether the incident was raised through the contractual channel, not informally | Tickets raised outside the named escalation path are sometimes disregarded |
| The credit formula and any cap or basket | Whether the claimed figure matches the arithmetic set out in the contract | Caps are frequently applied per incident, not per period, and missed |
None of these documents proves the case on its own. Together, they let the decision maker reconstruct the sequence: outage, detection, notice, claim. A gap anywhere in that sequence is the first thing raised by the other side.
Steps to take before submitting a claim
- The service owner exports the monitoring data covering the disputed period before the retention window closes.
- The contract manager checks the notice clause and calendars the deadline, not the calendar month.
- The technical lead maps the incident tickets to the specific service level definition said to be breached.
- The finance or contracts function runs the formula against the contract text, not against a prior invoice as a template.
- The claim is sent through the notice channel named in the contract, with the calculation shown as a worked figure, not a conclusion.
Skipping the second step is the most common cause of a claim being rejected on procedure rather than on merit. The other side does not need to dispute the outage if the notice was late.
Who reviews the calculation, and how long it takes
In most technology contracts, the first reviewer is the counterparty's contract or account manager, not a court or arbitrator. This stage is commercial, not judicial, and usually resolves within two to six weeks if the documentation is complete. If the two sides disagree on the figure, the contract's escalation clause typically routes the dispute to named senior contacts on each side, with a further window, often thirty days, before either party can invoke a formal dispute resolution mechanism.
Formal proceedings, where the contract provides for them, run on a separate and longer timeline that this page does not cover, because it depends on the mechanism chosen in the contract itself.
Two scenarios that change the approach
If the service level definitions were amended after signing and the amendment is not attached to the main contract, the decision maker will ask for the amendment first. Absence of that document has stalled more claims than any dispute over the underlying outage.
If the outage affected a service delivered by a subcontractor rather than the direct counterparty, the calculation still runs against the prime contract. The subcontractor relationship is not the decision maker's concern at this stage; the prime contract is. Escrow-related service continuity issues follow a related but separate process — see Escrow set-up technology contract what the decision maker looks for how that decision maker approaches source and continuity questions.
If neither side has retained the monitoring data, the calculation may fall back to the service provider's own reporting, which the receiving party is entitled to challenge but rarely able to replace. This is the weakest position for a claimant and the strongest for a respondent.
Evidence discipline: a comparable pattern
The same discipline — a documented sequence, checked against a defined trigger, submitted within a fixed window — appears in other technology and data disputes that turn on records rather than on argument. Publication practice in tokenised rights explained sets out how a comparable evidentiary sequence is checked when the underlying record is a publication rather than a monitoring log.
Where the underlying data involves personal data processed by a subcontractor, the processing terms sit in a separate document from the service level schedule. See Data processing agreement overview and the evidence for how that document is structured and what it is expected to show.
Sources and limits
This page does not resolve what a specific contract's formula produces, what cap applies, or whether a particular incident meets a particular threshold. Those depend entirely on the wording agreed by the parties, and no general procedure page can substitute for reading that wording. It also does not cover formal dispute resolution once escalation has been exhausted; that stage follows the mechanism named in the contract, not a general timeline.
Where the dispute turns on establishing facts about a counterparty rather than about the contract itself — for example, whether the counterparty still exists in the form it contracted under — a separate check is usually needed. See Counterparty in Lithuania and its limits and its limits for what such a check can and cannot establish. Where the dispute turns on establishing facts about a document itself, see Document in Latvia and the evidence for the equivalent limit on document-based checks.
Related reading
Service credit disputes and escrow release disputes both turn on whether a triggering event is proven by records the other side accepts. Readers dealing with one often need the other: Escrow set-up technology contract what the decision maker looks covers the parallel question for source code release.
Verified
This page was last checked for procedural accuracy on the date of publication. It describes a general commercial procedure common to technology and outsourcing contracts and does not rely on a specific statute, regulation or case citation, because none applies uniformly across the contracts this page addresses. Contract-specific wording always takes precedence over the general sequence described here.
Frequently asked questions
How long does service credit calculation take in practice?
Where documentation is complete, the first review by the counterparty typically takes two to six weeks. A contested figure that reaches formal escalation clauses can add a further thirty days before either side can move to a separate dispute resolution mechanism, if the contract provides one.
Which documents are needed to start?
At minimum: the service level schedule as currently amended, monitoring or dashboard data for the disputed period, the incident tickets raised through the named escalation channel, and the formula clause itself. Missing any one of these usually delays the claim rather than defeating it outright.
What happens if the other side does not respond?
Most contracts set a deemed-acceptance or deemed-rejection outcome after a fixed silence period, and the two produce opposite results. Check the escalation clause for which default applies before assuming silence favours either side.