Service Level Agreement (SLA) Clauses: Uptime, Credits and Real Remedies

A Service Level Agreement (SLA) clause names a measurable service standard, usually uptime as a percentage, response and resolution times, and support hours, and states what happens if the vendor misses it. The remedy is almost always a service credit: a percentage rebate on your fees, not cash, and usually not compensation for the real cost of the outage. The one thing most people get wrong: they read "99.9% uptime, service credits apply" as a promise, when the actual legal effect is closer to a discount coupon with a low ceiling. This guide (published by Adira, which makes contract review and CLM software, a commercial stake in you understanding contracts well, but this explainer stands on its own) walks through the maths, the drafting traps, and the Indian law behind what an SLA credit clause can take away from you.

Plain meaning

An SLA clause does two things in one place. First, it sets a number: "the Service will be available 99.9% of the time each month," or "Vendor will respond to Severity 1 tickets within 1 hour." Second, it sets a consequence for missing that number, almost always a service credit, a percentage of the fees for the affected period credited back or knocked off the next invoice.

That second part is where most of the risk sits. A service credit is not damages in the ordinary sense. It does not compensate you for lost revenue, lost customers, or a failed product launch caused by the outage. It is a pre-agreed, capped rebate, and in most commercially drafted SLAs, it is also stated to be your only remedy, whatever the outage actually cost you.

Who it protects and what triggers it

The clause exists for the customer, since the customer suffers when a paid service goes down. But it is drafted by the vendor, and it protects the vendor from open-ended liability far more than it protects the customer from the cost of downtime.

The trigger is the service falling below the stated metric during the defined measurement window, most commonly a calendar month. It only "bites" if three things line up: the outage counts as Downtime under the clause's own definition, it is not caught by an exclusion, and you claim the credit within whatever window the clause allows, which can be as short as seven days.

What to look for

Four mechanics decide whether an SLA clause gives a real remedy or a symbolic one.

1. How is uptime measured, and by whom? Monthly measurement is standard and fair; annual measurement can hide a bad week inside a good year. Watch for the vendor's own internal monitoring being the sole source of truth, with no audit right for you.
2. What counts as excluded Downtime? Scheduled maintenance, force majeure, and customer-caused issues are standard exclusions. The risk is scope: an unlimited, loosely-notified maintenance window, or a catch-all like "any factor outside Vendor's reasonable control," can swallow most real outages.
3. Is the credit the sole remedy, and is the cap meaningful? This is the single most important line. "Sole and exclusive remedy" language forecloses claiming the real cost of downtime, however large. Combine that with a small cap, commonly 5 to 25% of the monthly fee, and recovery for a catastrophic outage is a small fraction of one month's bill.
4. Is there an exit ramp for chronic failure? A vendor that pays the same small credit every month has no incentive to fix anything, unless repeated misses trigger a right to terminate for cause and get money back.

Quick test: is uptime measured monthly, is the exclusion list specific rather than open-ended, and does repeated failure let you leave without penalty. Three "no" answers means treat the SLA as decorative.

The uptime maths, worked out

Vendors quote uptime in "nines," and the gap between them is bigger than it looks. Over a 365-day year (8,760 hours):

• 99% uptime allows 87.6 hours of downtime a year, roughly 3 days 15 hours.
• 99.5% uptime allows 43.8 hours a year, about 1 day 20 hours.
• 99.9% uptime ("three nines," the most common SaaS commitment) allows 8.76 hours a year, about 43.8 minutes a month.
• 99.99% uptime ("four nines," typical for infrastructure-tier vendors) allows 52.6 minutes a year, about 4.3 minutes a month.

Run this test: multiply the allowed-downtime figure by your cost of an hour of outage (lost sales, idle staff, refunds), and compare that to the maximum credit the clause pays. For a payment gateway or a CLM system holding live deal documents, an hour of downtime in a critical window can cost far more than a month's fees, while the credit cap rarely exceeds 25% of one month's fees. That gap is the entire point of the clause, from the vendor's side.

The Indian position: Section 74 and what a credit ceiling actually means

An SLA credit is, in substance, a sum a contract names as payable when a specific breach happens, missing the stated service level. That puts it squarely inside Section 74 of the Indian Contract Act, 1872, the provision that governs liquidated damages and penalty clauses generally. It reads:

"When a contract has been broken, if a sum is named in the contract as the amount to be paid in case of such breach, or if the contract contains any other stipulation by way of penalty, the party complaining of the breach is entitled, whether or not actual damage or loss is proved to have been caused thereby, to receive from the party who has broken the contract reasonable compensation not exceeding the amount so named or, as the case may be, the penalty stipulated for." Source: Section 74, Indian Contract Act, 1872

Two consequences follow, and they cut in opposite directions.

First, the named credit is a ceiling, not a guarantee. Naming a 25% credit does not automatically entitle you to it; a court can award less if no real loss is shown, since the Supreme Court read "whether or not actual damage or loss is proved" narrowly in Kailash Nath Associates v DDA, treating loss as generally a precondition for compensation under Section 74. In practice this rarely bites customers, since vendors apply the credit as a billing adjustment rather than fighting for a lower figure.

Second, and this is the part that actually matters to you: the ceiling problem runs the other way when the clause says the credit is your "sole and exclusive remedy." That language is not really about Section 74 at all, it is a limitation-of-liability clause wearing an SLA-credit costume. Indian courts generally give effect to negotiated limitation and exclusion-of-remedy clauses between commercial parties under ordinary freedom of contract, so a clearly worded "sole remedy" clause is usually enforced on its own terms, separately from what Section 74 says about the credit figure. The result: Section 74 stops a court awarding more than the named figure, but nothing stops your vendor's exclusive-remedy clause from capping your entire claim at the credit, however large your actual loss. There is no reported Indian appellate decision squarely on SLA service credits; this is the general Section 74 liquidated-damages framework applied by direct analogy, and the framework itself is well settled.

A named Indian case: what "reasonable compensation" means here

Kailash Nath Associates v DDA, (2015) 4 SCC 136, is the case an SLA credit dispute would actually be argued around. The Delhi Development Authority tried to forfeit earnest money after a land allotment fell through, arguing Section 74 let it keep the named sum regardless of actual loss. The Supreme Court disagreed: a stipulated sum is recoverable as reasonable compensation only where it is a genuine pre-estimate of loss, and damage or loss is generally a precondition for compensation under Section 74. On the facts, DDA had suffered no loss, so it recovered nothing, despite the clause saying the sum was payable automatically.

Practically: a vendor cannot usually be forced to pay more than the stated credit, but a customer arguing for the full credit needs to show the outage happened and caused some loss, not merely that the uptime number was missed on paper. Read alongside ONGC v Saw Pipes Ltd, (2003) 5 SCC 705, where the Supreme Court held precise proof of quantum is not required where loss is real in kind but hard to measure exactly, a customer whose SaaS tool went down in a business-critical window does not need an exact rupee figure to justify the stated credit; that difficulty of quantifying real loss is itself why SLA credits exist as a pre-agreed number.

Red flags

Normal	Red flag	Why it matters
Uptime measured monthly	Measured annually or quarterly	A single bad day hides inside a good yearly average, so the SLA never triggers
Maintenance capped (e.g. 4 hours/month), advance notice required	Unlimited or loosely-notified maintenance excluded from downtime	Vendor can schedule its way out of an SLA miss
Exclusions specific: force majeure, customer-caused faults	Catch-all like "any factor outside Vendor's reasonable control"	A vague exclusion can be stretched to cover almost any outage
Credit cap meaningful, commonly 25 to 50% for severe misses	Flat, small cap (e.g. 5%) regardless of severity	Recovery stays trivial next to real business impact
Credit is primary remedy, with carve-out for termination and gross negligence	Stated as "sole and exclusive remedy," no carve-out	Forecloses recovering real loss for even egregious failures
Chronic-failure clause: termination-for-cause right and pro-rata refund	Same small credit every month, no escalation or exit	Vendor has no incentive to fix a recurring problem
Reasonable claim window (60 to 90 days)	Short window (7 to 15 days), easy to miss	Designed so genuine breaches go unclaimed and lapse
Uptime verifiable by a method both sides can audit	"As measured solely by Vendor's internal systems"	Vendor becomes scorekeeper and judge of its own breach

Bad clause → better clause

Bad: "Vendor shall use commercially reasonable efforts to maintain 99.9% uptime, measured on an annual basis. Scheduled maintenance and any circumstances outside Vendor's reasonable control shall be excluded from downtime calculations. If Vendor fails to meet this target, Customer's sole and exclusive remedy shall be a service credit of up to 5% of the monthly fees for the affected month, which Customer must claim in writing within 7 days of the incident, failing which the right to claim is waived."

Better: "Vendor shall maintain 99.9% Uptime, measured monthly as (Total Minutes in Month minus Downtime Minutes) divided by Total Minutes in Month. 'Downtime' excludes only (a) Scheduled Maintenance not exceeding 4 hours per month with at least 48 hours' notice, and (b) Force Majeure events as defined in Clause X. If Uptime falls below 99.9% in a month, Customer shall receive the Service Credit set out in Schedule A, ranging from 5% of that month's fees for 99.0 to 99.89% Uptime up to 25% for Uptime below 95%. These credits are Customer's primary remedy for Downtime but do not limit Customer's right to terminate for cause under Clause Y, or to claim damages arising from Vendor's gross negligence or wilful misconduct. Customer must submit a claim within 60 days of the end of the affected month. If Uptime falls below 99.9% in 3 or more months within any rolling 6-month period, Customer may terminate for cause and receive a pro-rata refund of prepaid fees, in addition to any Service Credits owed."

What changed: monthly measurement, a capped and notified maintenance window, a tiered credit schedule, "primary" instead of "sole and exclusive" remedy with named carve-outs, a workable claim window, and a chronic-failure termination right with a refund.

How it interacts with related clauses

• Limitation of liability. "Sole and exclusive remedy" language in an SLA is really a limitation clause. Check whether it sits inside your general liability cap or is a separate, harder ceiling for downtime.
• Termination for cause. A chronic-failure termination right is worthless if the general termination clause requires a long cure period a vendor can always technically meet.
• Force majeure. SLA exclusions often cross-reference the force majeure clause; a broadly drafted one widens what counts as excluded downtime too.

You can mark up an SLA clause for free, alongside the clauses it depends on, using Weave, before you push back on a vendor's draft.

US and global contrast

US and global SaaS contracts use broadly the same mechanics, uptime tiers, service credits, sole-remedy language, but the legal backdrop differs. US courts generally enforce a clearly worded "sole and exclusive remedy" clause as written, with no statutory filter like Section 74 asking whether the stated sum is reasonable compensation. Because a service credit is usually framed as a fee adjustment rather than a penalty, it mostly avoids the American penalty-doctrine fight, and courts rarely second-guess the cap. The practical effect is similar either way, a sole-remedy clause is hard to get around in both countries, but in India Section 74 at least keeps "reasonable compensation" in play if a customer can show the credit was never a genuine attempt to reflect real loss.

FAQ

Can I claim more than my SLA credit if downtime causes real business loss in India? Usually not, if the clause clearly states the credit is your sole and exclusive remedy. That language operates as a limitation of liability, which Indian courts generally enforce between commercial parties, separate from how Section 74 treats the credit figure itself. Negotiate a carve-out for gross negligence or wilful misconduct if this risk matters to you.

Is a service credit the same thing as liquidated damages under Indian law? Functionally yes. It is a sum named as payable on a specific breach, missing the stated service level, so Section 74 applies the way it applies to any liquidated damages clause: the named figure is a ceiling, not an automatic entitlement.

Why does it matter whether uptime is measured monthly or annually? Annual measurement lets a single catastrophic outage, say three full days down, average out against eleven good months and still show a "compliant" yearly figure, so the SLA never triggers. Monthly measurement catches a bad month on its own.

What is a fair credit cap for a business-critical SaaS tool? There is no fixed legal number, but caps in the 25 to 50% range for severe misses give the clause real teeth. A flat 5% cap regardless of severity is common in vendor-favourable drafts and rarely reflects real business impact.

What should I do if a vendor keeps missing its SLA every month? Check for a chronic-failure or repeated-breach termination right. Without one, an unreliable vendor can keep paying the same small credit indefinitely with no real consequence. Also check whether the uptime promise is worded as "commercially reasonable efforts to maintain 99.9%" rather than "shall maintain 99.9%": the first is hard to prove was breached, the second is a measurable outcome with a clear trigger.

This guide gets you to understanding what an SLA credit clause does and how Section 74 and limitation-of-liability principles apply to it in India. It does not tell you whether a specific SLA is fair, or whether a sole-remedy clause would hold up in your dispute, that depends on the facts and the scale of loss involved, and is not legal advice. Talk to a lawyer before you rely on, negotiate, or enforce an SLA clause in a live deal.