What should GEO service levels cover? A buyer's scorecard

Set service levels for controllable delivery, not promised visibility

A useful GEO service level says what the supplier will deliver, how quality will be checked, when the buyer must act, what happens after an error, and which evidence will be available. A vague promise of responsiveness is difficult to manage. A guaranteed ranking, crawl, indexation, AI mention, citation, traffic or enquiry is worse: those outcomes depend on external systems, users, competitors, the buyer and time as well as the supplier.

Start with the service workflow. Separate commitments the supplier can control—such as delivery dates, evidence fields, review turnaround, correction handling and handover—from monitored outcomes the parties will observe but cannot honestly guarantee. GOV.UK sourcing guidance similarly says performance measures should be relevant and proportionate, while payment and accountability should relate to outputs a supplier can influence.

Complete one six-part GEO service-level scorecard

Write one row for each material deliverable or workflow. Define the clock, the required evidence, buyer dependencies, exclusions, measurement source and remedy before accepting the target. The same headline target can otherwise mean different things to each party.

Define the clock and dependencies precisely

For every time-based commitment, state the time zone, service hours, working-day calendar, start event, due event, priority rule and pause conditions. ‘Respond within one day’ is ambiguous: it could mean an automated acknowledgement, a qualified assessment or a verified correction. Use separate fields where those events matter.

Record buyer dependencies beside the supplier target. A delivery clock might start only after the buyer supplies approved facts, access and a named approver; a correction may pause while a legal or data-protection decision is required. Relief should apply only to the documented dependency that actually caused delay, with notice and evidence. Do not let a generic dependency clause erase accountability for work the supplier controls.

Make evidence quality measurable

Timeliness alone can reward fast but unusable output. Pair each cadence or response target with a quality gate. For an AI visibility observation, that may include the question, interface, market, language, time, source links, answer capture and separate fields for mention, citation and accuracy. For a content or technical change, it may include the approved source, affected URL, review record, test result, release evidence and rollback route.

Keep the number of top-level measures small enough to manage. The Sourcing Playbook warns that too many KPIs create complexity and ambiguity; the risk-allocation guidance calls for measures that are objective, understandable and sensibly measurable. Supporting fields can be detailed without turning every field into a headline KPI.

• Use the system of record named in the contract rather than an untraceable dashboard screenshot.
• State the sample and denominator whenever a percentage is reported.
• Distinguish supplier error, buyer delay, platform change and an unresolved cause.
• Preserve exceptions and failed checks instead of reporting only the successful output.

Use remedies to restore service and evidence

Design the operational response before debating service credits. A missed or failed target should trigger a proportionate sequence: record the event, protect the current state, notify the right owner, correct or re-perform the work, verify the result, identify the cause and update the control. Repeated or serious failure may require a rectification plan, added review, scope change or exit decision.

Commercial consequences must match the contract, value and risk. GOV.UK guidance cautions that incentives, deductions and risk transfer can produce unintended behaviour when they are disproportionate or attached to the wrong measure. A GEO buyer should not reward an attractive metric that encourages weak sampling, unsupported claims or rushed publication. Obtain legal and commercial advice for contractual remedies rather than copying a generic service-credit formula.

Run a service-level acceptance exercise

Choose one normal delivery and one material exception. For the normal case, submit a complete brief and follow it through production, evidence, review, approval and handover. For the exception, introduce an incorrect bilingual fact, unavailable approver or changed tool and observe acknowledgement, triage, containment, correction, verification, notice and closure.

The scorecard passes when both parties calculate the same result from the same records, buyer dependencies are visible, quality gates prevent unsafe release, exceptions produce usable evidence and the handover can be opened by someone outside the supplier team. Record any disagreement as a definition or control defect before relying on the metric in live reporting.

How SALIENS supports measurable GEO delivery

SALIENS provides GEO audits, AI visibility monitoring, content and technical improvements, and ongoing GEO delivery. An agreed scope can include a proportionate service-level scorecard, delivery calendar, source-linked evidence, bilingual review gates, priority and correction workflow, change records, regular performance review and exit handover. These controls make service performance easier to inspect; they do not guarantee crawling, indexation, AI mentions, citations, rankings, traffic or enquiries.

Email info@oxfordintelligence.co.uk with your website, markets and languages, intended deliverables, critical dates, approval owners, evidence requirements and existing supplier measures. SALIENS is operated by Oxford Intelligence Limited. This guide is a procurement framework, not legal or commercial advice; buyers should tailor targets, liabilities and remedies to their own risk and contract.