Understanding Telecom SLAs: What Your Carrier Actually Owes You

An SLA is a contract that says "if our service doesn't meet this standard, you get paid." Simple. But the details of what that standard is, how it's measured, what the payment is, and what's excluded — those details determine whether the SLA actually protects you or just creates an illusion of protection.

We've seen businesses with SLAs that look great on paper but don't actually protect against the failures that matter most. We've also seen businesses with modest SLAs that negotiated smartly and have real protection. The difference is understanding what you're actually buying.

What an SLA actually is (and isn't)

An SLA — Service Level Agreement — is the carrier's formal commitment to you that their service will meet certain performance standards. If it doesn't, they owe you money (usually a service credit). It's basically a penalty for missing their own promises.

Here's the critical misunderstanding: the credit is not the same thing as compensation. If your internet goes down for 8 hours and costs you $10,000 in lost productivity, the carrier might owe you a $100 credit. That's not fair. That's just accountability for missing a standard. The real value of an SLA is different: it forces the carrier to prioritize your repair over residential customers' repairs. Your circuit jumps the queue.

Think of it this way. A carrier without SLAs has no obligation to do anything if your service fails. A carrier with SLAs has a contractual obligation to hit certain metrics or pay. That creates pressure to staff appropriately, maintain equipment better, and dispatch repairs to SLA circuits first. The credit is incidental. The dispatch priority is the actual protection.

What an SLA is: A formal, measurable commitment with financial consequences for missed performance. A signal that the carrier has made you a priority. A basis for claiming credits when the carrier fails.

What an SLA is not: Insurance for your business. Compensation for the damage caused by downtime. A guarantee of zero downtime. A complete list of everything that could go wrong.

The four SLA metrics that matter

1. Uptime percentage

Uptime is usually expressed as a percentage: 99.9%, 99.5%, 99%, etc. It's supposed to represent the percentage of hours in a month that your service is available and working.

Here's the math: 99.9% uptime means 0.1% downtime allowed. Over a month (730 hours), that's 43 minutes of downtime per month, or 8.7 hours per year. Business fiber typically comes with 99.9% SLAs. More premium tiers offer 99.99% (4 minutes per month).

The hidden details: How does the carrier measure uptime? Some measure it from the carrier's demarcation point (the modem on your end). Others measure it from their network core. These measurements differ significantly. If the modem reboots, is that downtime? Some SLAs say yes. Others say no. Get clarity on how uptime is measured before you sign.

Also, what's excluded? Most SLAs have carve-outs for scheduled maintenance, customer-caused outages, and equipment the customer owns. A carrier can often exclude 8 hours per month for "scheduled maintenance" and that time doesn't count against the SLA. That 99.9% SLA suddenly becomes 99.4% once you account for the exclusions.

2. Mean Time To Repair (MTTR)

MTTR is the average time the carrier has to respond to your repair request. A typical SLA might specify "4-hour MTTR for critical circuits." This usually means 4 hours to first contact — a technician on the phone — not necessarily 4 hours to full restoration.

The distinction matters. A 4-hour MTTR SLA means a technician has to call you within 4 hours. It doesn't mean your service is restored within 4 hours. Restoring a fiber cut might take 12 hours. The carrier still meets the SLA if they called within 4 hours. The real value is in the dispatch: a circuit with an MTTR SLA gets a truck roll immediately. A circuit without an SLA goes into the queue and might not get a truck for days.

On-site MTTR (where the technician must physically arrive at your location within the window) is rarer and more valuable, but more expensive. Negotiate for on-site MTTR if your business can't tolerate the delay of waiting for a first contact.

3. Latency

Latency is the round-trip delay — the time it takes for data to travel from your device to the carrier's network and back. Measured in milliseconds (ms). Business applications are generally sensitive to latency above 50ms. Voice calls become noticeably delayed above 150ms. Video becomes choppy above 100ms.

A typical business internet SLA might specify "average latency under 40ms, peak latency under 80ms." That's a measurable standard. MPLS circuits often come with tighter latency guarantees (under 30ms) because that's part of why you're paying for MPLS instead of regular internet.

The trap: latency is easy to measure technically but hard for a customer to prove if the carrier disputes it. Latency issues are often intermittent. You need good monitoring on your end to capture proof. Make sure your SLA also specifies how latency is measured (from where to where?) and gives you visibility into the carrier's measurements.

4. Jitter

Jitter is the variation in latency. If your latency is 40ms on one packet and 45ms on the next and 38ms on the third, you have jitter. For real-time applications like voice and video, jitter matters as much as the absolute latency. High jitter causes choppy, delayed conversations.

Jitter is the least commonly specified SLA metric, which is a mistake. Voice providers usually care about jitter, but basic internet SLAs often don't mention it. If you're running VoIP on your circuit, make sure the SLA specifies jitter limits. Typical targets are "jitter under 10ms average, under 20ms peak."

How credits work and why they miss the point

When a carrier misses an SLA, they owe you a credit. The structure varies but typically looks like this:

• Standard: 5% of the monthly recurring charge per incident, capped at 30% per month
• Negotiated better: 10% per incident, uncapped
• Negotiated best: 15% per incident, uncapped, plus expedited repair SLA

Here's the reality: these credits rarely compensate for actual business impact. If your internet goes down for 6 hours and costs you $25,000 in lost orders, a 10% monthly credit (maybe $500) doesn't even touch the real damage.

So why do SLAs have credits at all? Three reasons: (1) They create accountability. A carrier with no SLA and no credits doesn't care if you're down. A carrier with SLA credits has financial pressure to avoid it. (2) They're better than nothing. A $500 credit is $500 you wouldn't have gotten without the SLA. (3) They create a paper trail. If you ever litigate or escalate to a regulator, documented SLA misses and missed credits build your case.

The bigger value of the SLA credit mechanism is that it forces carriers to track their own performance. If a carrier has to pay credits when they miss SLAs, they monitor their network more closely, staff their NOC better, and prioritize customer repairs more aggressively. All that happens because of the SLA, not because of the credit amount.

When negotiating SLAs, push for: (1) Higher credit percentages. 10% minimum. (2) Uncapped monthly credits. Don't let the carrier cap their exposure. (3) Multiple incident credits. Each missed SLA = a separate credit, not one monthly total.

Negotiating better SLAs

Carriers ship standard SLAs that favor them. You can negotiate better. Here's what carriers will usually move on:

Uptime threshold: Most standard is 99.9%. If you need higher, push for 99.95% or 99.99%. You'll pay more, but business-critical services justify it.

MTTR window: Standard is 4-hour first contact. Push for 2-hour if your business is truly critical. Carriers resist this because it requires more staffing, but some will do it for an upcharge.

Latency/jitter specs: If they're not in the SLA, add them. "Average latency under 40ms, peak under 80ms, jitter under 15ms average." Most carriers will accept this as long as it's reasonable for the service type.

Credit multiplier: Instead of negotiating percentage, sometimes you can negotiate a multiplier for repeated offenses. "First miss: 5%. Second miss in same month: 10%. Third miss: 15%." This incentivizes carriers to fix systemic problems quickly.

Escalation engineer: For large contracts, push for a dedicated escalation contact — a senior NOC engineer who handles your tickets. Worth every penny for critical services.

Residential-grade vs. business-grade SLAs

The biggest SLA difference you'll encounter is between residential and business service. They're night and day.

Residential grade (cable internet for the home, residential fiber): No SLA at all. Zero. The carrier has no contractual obligation to meet any performance standard. If your service goes down, they have 4 hours to "respond" (usually meaning take a trouble ticket). No guaranteed repair time. No credits. If you call at 2 PM and they respond at 6 PM saying "a tech will come by next Tuesday," that's compliant.

Business grade (cable internet for business, business fiber, MPLS, Ethernet): Has an SLA. Usually 99.9% uptime, 4-hour MTTR, latency/jitter specs, and credits for misses.

The physical service might be identical (same fiber, same backhaul), but the SLA and support model are completely different. Business customers get priority dispatch. Residential customers wait in a queue. This is why business-class service costs more — you're not just paying for bandwidth, you're paying for priority.

MPLS vs SD-WAN vs broadband SLA comparison

MPLS is the SLA gold standard because carriers manage the entire path end-to-end. They control everything and can make hard commitments. Dedicated fiber is close second. SD-WAN can meet business needs but relies on multiple carriers or paths, which complicates SLA commitments. Residential broadband is no-SLA territory and shouldn't be used for anything business-critical.

What to do when your carrier misses SLA

Document everything. When does the outage start? When does the carrier acknowledge it? When does repair begin? When is service restored? Screenshot everything. Save carrier emails. Build a timeline. That's your proof.

Notify the carrier immediately. Call the NOC, report the outage, and ask them to log it as an SLA-eligible incident. In writing is better than verbally. Email the account manager same day saying "Circuit XYZ went down at [time], restored at [time]. This is an SLA-reportable incident."

File a formal SLA claim. After the incident is resolved and you have all the data, submit a formal SLA credit claim to the carrier. Include dates, times, duration, proof of outage, and calculation of the credit due. Most carriers have a claims process. Use it. If they deny the claim without good reason, escalate.

Escalate if they push back. Carriers sometimes deny SLA claims by arguing the outage was caused by something excluded in the SLA (scheduled maintenance, customer equipment, etc.). If that's wrong, push back. Cite the contract language. Provide your proof. If they still refuse, consider escalating to your broker or to state telecom regulators.

For repeated misses, demand remediation. If a carrier misses the same SLA metric repeatedly, that's systemic. Demand that they fix it or offer a service replacement. A carrier that can't maintain their SLA isn't meeting your business needs, and you may have grounds to terminate early without penalty.

Frequently asked questions