Vercel's uptime commitment lives in a contract, not the status page

The workload

Vercel's own Service Level Agreement, last updated 18 July 2024, requires a customer seeking a credit to 'notify Vercel in writing within thirty (30) days from the time you become eligible' and to 'provide log files' documenting the downtime with specific dates and times, verified — a narrower window than AWS's two-billing-cycle deadline elsewhere in this batch, and one that puts the evidentiary burden on the customer. Separately, Vercel's public status page tracks 'Uptime over the past 90 days' for dozens of components (API, CDN, regional edge nodes), a monitoring record Vercel publishes itself, verified, distinct from the claim process above.

What the documents show

Verified, from the SLA text: Vercel commits to '99.99% of the time, excluding any Excused Downtime events,' measured over one calendar month, with credits of 10% of the monthly fee for 99.1%–99.98% actual uptime, 25% for 95%–99%, and 50% below 95%. The document is titled the 'Enterprise Service Level Agreement,' signaling which tier it is written for, though the retrieved text does not itemize whether Hobby or Pro accounts are excluded outright — a founder on a lower tier should not assume these figures apply without checking directly. Verified exclusions ('Excused Downtime') include third-party vendor failures, internet outages, force majeure, and the customer's own software or hardware. The status page states no contractual remedy at all; it is a transparency tool, not a credit mechanism.

The operating cost

Credits are capped, verified, at '50% of the amount due...for the applicable month,' applied 'within sixty (60) days after the Credit was requested' — a bill credit, not a cash refund and not compensation for downstream losses. The SLA also carves out, verified, 'components of the Vercel platform that do not serve customer content...such as APIs or CLI,' meaning an outage confined to build tooling may not qualify for a credit even if it stops a deploy cold.

The stop condition

The 30-day notice window is the hard deadline stated, verified; past it, eligibility lapses regardless of how well-documented the outage was. Editorially, because the SLA is titled for Enterprise and the status page carries no credit terms, a founder on free or Pro tiers should treat the absent plan-specific commitment as the operative fact — there may be nothing contractual to invoke below Enterprise, and the status page's uptime figures are not a promise.

• Is the account actually on the Enterprise plan the SLA is titled for, and if not, has Vercel confirmed in writing whether any uptime commitment applies?
• Would the specific outage affect 'customer content' as the SLA defines eligible downtime, or would it fall into the excluded API/CLI category?
• Are the log files the SLA requires for a claim something the deployment's own monitoring already captures, or would they need to be assembled after the fact?

A status page and a service level agreement answer different questions: one reports what happened, the other states what was promised and to whom. Vercel publishes both, but only the second is a contract, and only for the tier it names.