TL;DR
Most businesses buy a SIP trunk labeled “unlimited calling” and assume capacity is a solved problem. It isn’t. “Unlimited” almost always refers to minutes and per-minute billing — not the number of calls you can have in progress at the same time. That concurrency limit is set by channels, and when every channel is busy, the next caller gets a busy signal or a dead-air drop, regardless of how much “unlimited” bandwidth is left on the contract. The real fix isn’t buying more minutes — it’s sizing channels to your actual busy-hour call volume and knowing exactly what happens the moment you exceed it.
The Status Quo Trap: “We’re on an unlimited plan, so we’re covered”
Ask most office managers or IT leads why they haven’t revisited their SIP trunk sizing since it was installed, and the answer is usually: “it’s unlimited calling, so capacity isn’t something we think about.” That’s the trap. “Unlimited” is a billing term, not a capacity guarantee — it tells you how your minutes are metered, not how many conversations your system can hold open at once.
Here’s the mechanic most buyers never get walked through at the point of sale. A SIP trunk is provisioned with a fixed number of channels, and each channel supports exactly one call. A trunk sized for 20 channels can carry 20 simultaneous calls — the 21st caller doesn’t get charged extra, they get rejected, unless the plan has bursting enabled. “Unlimited” describes what happens to calls 1 through 20 on the bill. It says nothing about what happens to call 21.
This gap stays invisible for years because average daily call volume looks fine on every report. The problem only shows up at the busy hour — open enrollment week, a product recall, a marketing campaign that actually works, a weather event that floods every location’s main line at once. Capacity planning has to be built around peak concurrency, not average volume, and the industry’s own grade-of-service standard reflects that: trunk sizing is typically targeted to a 1% call-blocking probability at the busiest hour of the busiest day, not at the average hour of an average day. A business running 500 calls a day with short average handle times might realistically need only 10-15 channels — but if all 500 ever tried to land in the same 60 minutes, that same trunk would choke instantly. Most businesses have never actually measured which scenario they’re exposed to.
The deeper issue is that “unlimited” was never the number that mattered. Channel count was. And nobody re-runs that math until the day a campaign, an outage reroute, or a seasonal spike proves it wrong in front of customers.
The Tele Data Guru Framework: The Busy-Hour Capacity Matrix
Before renewing or resizing a SIP trunk, map your actual call profile against the channel ratio and failure mode that applies to it. Most contracts are sized once, at initial install, against whatever ratio the sales rep assumed — not the ratio your specific call pattern actually requires.
| Call Profile | Recommended Channel Ratio | What Happens at the Ceiling | Overflow Behavior to Verify in Writing |
|---|---|---|---|
| General office / reception line | Roughly 1 channel per 3-4 employees | Busy signal or voicemail bounce for the next inbound caller | Does the plan burst automatically, and at what per-minute premium? |
| Call center / support desk | Closer to 1 channel per agent (1:1) | Calls queue past acceptable hold time or abandon entirely | Is queuing built into the PBX, or does the trunk simply reject the call? |
| Marketing campaign / outbound dialer | Sized to campaign peak, not daily average | Dialer throttles or outbound calls silently fail to connect | Is burst capacity available same-day, or does it require a provisioning request? |
| Disaster recovery / failover trunk | Sized to 100% of a single location’s peak, assuming total reroute | The “backup” trunk hits its own ceiling during the exact event it exists for | Has the full-reroute scenario ever been load-tested, or only assumed? |
| Multi-location consolidated trunk | Sized to the sum of overlapping busy hours across sites, not each site individually | One location’s peak silently starves channel availability for every other site | Does the provider report per-location concurrency, or only trunk-wide totals? |
Most businesses only ever evaluate the first row of this matrix — because that’s the scenario the original installer sized for. The rows that actually cause outages are the ones nobody tested: the failover trunk during the real disaster, the dialer during the real campaign, the consolidated trunk during the real regional event.
The Channel Shortfall Cost Formula
Channel Shortfall Cost = (Peak Concurrent Call Demand − Provisioned Channels) × Average Call Value × Blocked Calls per Busy-Hour Event
Example: a 60-agent support operation is provisioned for 40 channels on the assumption that “unlimited calling” meant capacity was handled. During a product issue, peak concurrent demand hits 58 calls — an 18-channel shortfall. If the average resolved call is worth $150 in retained revenue or avoided churn, and 18 calls per hour are blocked for a 6-hour event, that’s 108 blocked calls × $150 = $16,200 in a single afternoon — before factoring in the reputational cost of customers who got a busy signal from a company that advertises 24/7 support. Run this formula against your worst historical call-volume day, not your average day, and the “unlimited” plan’s real ceiling becomes a budget line instead of a surprise.
Commercial Realities & Vendor Pitfalls
- “Unlimited” is a minutes claim, not a concurrency claim. It typically means no per-minute billing on domestic calls within your channel count — it does not mean unlimited simultaneous calls. Get the channel count in writing, separate from the word “unlimited” on the quote.
- Bursting isn’t free, and it isn’t automatic by default on every plan. Overflow capacity is frequently billed at a steep premium over the base per-channel rate once you exceed your committed count, and some providers require it to be explicitly enabled rather than assuming it as a built-in safety net.
- Your failover trunk is sized for the outage you can imagine, not the one that happens. A backup trunk provisioned for “normal operations at the backup site” routinely gets overwhelmed the moment it’s actually asked to absorb a full regional reroute — because nobody load-tested the real-world failover scenario, only the paperwork version of it.
- Codec choice quietly changes your real bandwidth exposure. A higher-fidelity codec can consume meaningfully more bandwidth per call than a compressed one once protocol overhead is included, which matters when concurrent call volume competes with data traffic on the same circuit during a peak event.
- Call completion and caller ID trust depend on compliance you may not have verified. Carrier-grade call completion and caller ID trust increasingly depend on STIR/SHAKEN compliance further up the call path — worth confirming with any provider, especially one priced purely on a per-channel basis.
Implementation Checklist
- Pull 12 months of call detail records and identify your actual busy-hour peak concurrent call count — not average daily volume.
- Compare that peak against your provisioned channel count and current staffing ratio (roughly 1:3-4 for general office, closer to 1:1 for call centers).
- Get written confirmation of what “unlimited” means on your current contract: minutes, channels, or both.
- Confirm whether bursting is automatic or requires enrollment, and get the overage rate in writing before you need it.
- Load-test your disaster recovery or failover trunk against a full-reroute scenario, not just a partial one.
- For multi-location trunks, request per-location concurrency reporting, not just trunk-wide totals.
- Re-run the Channel Shortfall Cost formula against your worst historical call-volume day, and revisit sizing before the next campaign, seasonal peak, or office expansion — not at the next renewal date by default.