Service Level Agreement
Version 1.1 · Last updated: August 24, 2026 · Versión en español (the Spanish version is authoritative; this English version is provided for convenience)
This Service Level Agreement applies to the Scale and Enterprise plans and forms part of the Terms of Service (§12.2). The Agentic and Growth plans are governed by the operational availability target in §12.1 of the Terms, with no entitlement to credits. With respect to remedies, this SLA is construed consistently with Section 14 of the Terms and clause 8.3 of the DPA: the service credit is the sole remedy.
1. The two planes of the Service
The Service has two planes with distinct risk profiles, and this SLA treats them separately — because losing data is not the same as waiting for a report:
| Plane | What it is | Consequence of an outage |
| Measurement plane (pixel ingestion) | The receipt of events from the Customer's sites, served with replica redundancy | Loss of data for the outage period — unrecoverable. This is the critical plane |
| Query plane (dashboard, API, exports) | Access to reports and already-processed data | Waiting — no data is lost: received events are queued in a persistent messaging system and processed once the service recovers |
This asymmetry is a design property: unavailability of the processing or query components delays reports but does not lose events already ingested.
2. Availability commitments
| Plane | Monthly commitment | Max. equivalent unavailability/month |
| Measurement plane (ingestion) | 99.5% | ~3 h 39 min |
| Query plane (dashboard + API) | 99.0% | ~7 h 18 min |
Monthly availability (%) = (minutes in the month − minutes of Unavailability) / minutes in the month × 100, as measured by Sealmetrics’ internal monitoring (continuous health checks of the public endpoints, with automated alerting). The Customer may request the measurement details for a specific incident.
Unavailability: a continuous period of 15 minutes or longer (rounded to the minute) during which the relevant plane fails to respond or returns errors for customers generally, as confirmed by Sealmetrics’ monitoring. Partial degradation that does not prevent the essential function of the plane (e.g. elevated latency with correct ingestion) does not constitute Unavailability. Interruptions shorter than 15 consecutive minutes do not count.
3. Exclusions
The following do not count as Unavailability:
- Scheduled maintenance: notified at least 24 hours in advance, in a low-traffic window, up to a cumulative 8 h/month.
- Emergency maintenance reasonably necessary for the security or integrity of the Service (notified as soon as practicable).
- Failures of the Customer’s or third parties’ networks or infrastructure beyond Sealmetrics’ reasonable control (the Customer’s DNS, its CDN, its hosting provider, regional Internet outages).
- Customer instrumentation or configuration errors (misinstalled tracker, blocking on the Customer’s site, exceeded plan limits).
- Force majeure.
- Features marked as beta or experimental, and the AI component to the extent it depends on the inference provider’s capacity (degradation of the AI assistant does not affect measurement or reporting).
- Suspensions pursuant to the Terms of Service (non-payment, abuse, legal requirement).
4. Service credits
| Monthly availability of the affected plane | Credit against the monthly Service fee* |
| < 99.5% (measurement) / < 99.0% (query) and ≥ 98.0% | 5% |
| < 98.0% and ≥ 95.0% | 10% |
| < 95.0% | 20% |
* For annual plans, one twelfth of the annual fee. If both planes fall short in the same month, the higher credit applies (credits do not stack).
The Customer requests the credit by email to support within 15 days following the end of the affected month, stating the dates and times of the incident. Credits are applied against future fees, are not redeemable for cash, and their annual total will not exceed the equivalent of one month’s fees. The credit is the sole and exclusive remedy for availability shortfalls.
5. Technical support — response times
Support hours: business days (Barcelona calendar), 9:00–18:00 CET/CEST. The Enterprise plan may agree extended coverage (including 24×7 for S1) in its individual agreement.
| Severity | Definition | First response — Scale | First response — Enterprise |
| S1 — Critical | Measurement plane down or data loss in progress | 8 business hours | 4 business hours |
| S2 — High | Query plane down; essential functionality inoperative with no workaround | 8 business hours | 4 business hours |
| S3 — Medium | Degradation with a reasonable workaround; non-blocking errors | 1 business day | 8 business hours |
| S4 — Low | Inquiries, enhancement requests, usage questions | 2 business days | 1 business day |
These times refer to the first qualified response, not to resolution. Sealmetrics keeps the incident updated until closure.
6. Incident communication
Confirmed S1/S2 incidents are communicated to affected Customers by email without undue delay, with periodic updates until resolution and a post-incident summary for S1. If a measurement-plane incident involves data loss, the communication will state this expressly, together with the affected interval.
7. Continuity and data protection (informational)
Internal recovery objectives, stated for information only (they do not by themselves give rise to credits): encrypted daily backups with 30-day retention — with the session-pseudonymisation cryptographic material excluded from backups by design, because continuity is never bought at the expense of privacy; an RPO of 24 h for configuration data (already-ingested measurement events are additionally protected by the persistent queue); an RTO of 24 business hours after a major disaster; and active redundancy on the measurement plane.
8. Review
This SLA is reviewed at least annually. Amendments are communicated with reasonable notice; amendments that reduce the commitments take effect only upon the Customer’s next renewal.
This SLA forms part of the Terms of Service. Questions: support@sealmetrics.com.