You've built a call center for the cycle. Seats are full, the dialer is loaded, the script is approved. Then on launch day half the calls don't connect, agents are sitting idle, and your number pool is burning down anyway because the carrier is counting every failed attempt against you.
The script isn't the problem. The numbers aren't the problem yet. The problem is the pipe between your call center and the rest of the phone network — and most campaigns don't look at it until it's already broken.
This post is for the operator standing up a political call center for the next eight weeks. We'll cover what "connectivity" actually means in this context, the four places it usually breaks, and what to look for in a provider before you sign anything.
What connectivity means for a campaign call center
Your call center has a dialer (Five9, Genesys, an in-house build, doesn't matter). Your dialer needs to push calls out into the regular phone network. The thing connecting those two is a SIP trunk — basically a wholesale pipe that turns your internet-based dialer into actual phone calls people answer on their cell phones.
Connectivity is the quality of that pipe. Three things determine whether it holds up under campaign pressure:
- Capacity — how many calls can run through it at the same time without queueing or rejecting.
- Routing — which downstream carriers your provider hands calls to, and whether they pick the path your call is most likely to be answered on.
- Reliability — whether the trunk stays up when traffic spikes, when a downstream carrier has a partial outage, when you're calling at 7 PM on a Tuesday and so is everyone else.
Get any one of these wrong and your agents are idle. Get two wrong and your cycle is in trouble.
Where political call center connectivity dies
There are four common failure modes. We see all four every cycle, usually in the first three days after launch.
The trunk is undersized
A 50-seat predictive dialer doesn't need 50 channels. It needs 100 to 200 — because the dialer is calling ahead of agents, trying to find live answers and bridge them into open seats. If your trunk only allows 60 concurrent calls, the dialer hits its ceiling, agents wait, and the cost per contact triples.
Most providers under-quote channels at signup to win the deal. Then on launch day you find out you need to pay 3x to scale up. Get the answer in writing before you commit: how many concurrent channels can I burst to, and what does it cost.
The provider routes badly to certain regions
Carriers don't all play nice with each other. Some wholesale providers have great paths to AT&T and Verizon but poor paths to T-Mobile. Some have one bad carrier hop that adds 800ms of latency to half your traffic. You don't see it on a test call. You see it when the data comes back showing your answer rate is 18% in some states and 4% in others.
Ask your provider — directly — how they handle T-Mobile traffic and whether they have direct interconnections or hop through a transit network. If they can't tell you, that's the answer.
The firewall eats the audio
The most painful failure mode of all: signaling works, the call connects, the agent says hello, and the recipient hears nothing — or vice versa. Almost always this is a NAT or firewall problem at the call center side. The SIP packets (the "ring this phone" instructions) got through. The RTP packets (the actual voice) didn't.
The fix is boring and mechanical: open the UDP port range your provider uses (usually 10000-20000) on both directions of your firewall, and make sure any NAT box knows to pass media addresses through correctly. A 30-minute fix that takes three days to diagnose if nobody on the call center side knows the network.
There's no failover
Your provider has an outage. Maybe a regional one, maybe a full one. Maybe just one downstream carrier they use went down. Without a second trunk configured as failover on your SBC, your dialer just stops dialing. Hours of contact opportunity gone in a cycle where you can't get them back.
A second termination provider — even just routing 5–10% of normal traffic to keep it warm — is the cheapest insurance you can buy. When primary goes down, you flip the ratio and the operation keeps running at degraded capacity instead of grinding to zero.
What to look for in a termination provider before you sign
A short checklist you can walk through with any provider's sales rep in a 20-minute call. If they hedge on any of these, find another provider.
- Concurrent channels. What's my plan ceiling? What's the cost to burst above it? Can I burst the same day or do I need to file a ticket?
- Carrier routing. Do you have direct interconnects to AT&T, T-Mobile, and Verizon, or do you transit? What's your published answer rate by carrier?
- STIR/SHAKEN signing. Do you sign my outbound calls at attestation level A? (Required for legitimacy — if they sign at B or C, your calls get filtered.)
- Onboarding speed. When I sign today, how fast is my trunk live and ready to take traffic? (Should be hours, not days.)
- Failover. Can I configure a secondary trunk with you as well, in case my primary provider goes down? Or do I need to bring my own second provider?
- Support during launch week. Do I have a human reachable by phone in the first 72 hours, or am I filing tickets and waiting for next-day responses?
- Rate deck transparency. Is there a per-minute rate sheet I can see before I sign, or is it negotiated? (Both are fine — opacity is not.)
A real wholesale provider answers all seven of these directly. A reseller pretending to be one will dodge half of them.
A note on call quality vs. deliverability
Connectivity is about whether the call goes through at all. Deliverability — whether the recipient sees a clean local caller ID instead of "Spam Likely" — is a separate problem with a separate playbook. You need both. A perfectly delivered call that drops mid-conversation is worthless; a crystal-clear call that shows up flagged is also worthless.
The good news is they're solved by mostly different work. Connectivity is a one-time setup with your provider. Deliverability is an ongoing operational practice you maintain across the cycle.
What to do this week
If you're more than two weeks out from launch:
- Pick a primary termination provider and run a 100-call test load through them. Look at answer rate, drop rate, and audio quality.
- Pick a secondary provider and configure failover on your SBC, even at minimal traffic. Test the failover by manually flipping the ratio.
- Walk your call center's network through the firewall checklist — open the UDP media port range, document NAT behavior, verify with a quick test call.
- Get carrier routing answers in writing from your primary provider, especially for T-Mobile.
If you're less than two weeks out, all four of those are still doable — just compressed. The most expensive mistake is launching with the connectivity issues unsolved and burning your number pool against them in the first 72 hours.
Phones are the boring infrastructure under the political work. Get them right early and your team can focus on the part of the job that actually moves voters.