How to Choose a VoIP Carrier for Dialers

Your predictive dialer is only as good as the SIP trunk feeding it. This is the point we make repeatedly because operators repeatedly underestimate it. You can run the most sophisticated VICIdial configuration with perfectly tuned adaptive pacing, pristine lead data, and world-class agents — and still underperform because your carrier throttles channels, provides B-level attestation, or introduces 6-second PDD that throws off the algorithm’s predictions.

Choosing the right carrier is a decision that affects every metric in your operation: answer rate, abandon rate, agent utilization, conversion rate, and ultimately revenue per hour. This guide provides the evaluation framework. SIPNEX is an FCC-licensed carrier that built its SIP trunking product specifically for predictive dialer operations. We compete on these criteria, so we are transparent about what they are.

The evaluation checklist

1. Concurrent channel policy

Question to ask: “Is there a per-trunk or per-account concurrent channel limit? What happens when I reach it?”

Why it matters: A predictive dialer with 50 agents running a 4:1 ratio needs 200 concurrent channels. With answer rate drops, burst demand can hit 250 to 300. If your carrier caps at 50 or 100 channels, your dialer receives SIP 503 errors on every call above the cap. The algorithm cannot execute its predictions. Agents sit idle while the carrier throttles your capacity.

What to look for: “Unlimited” or a cap significantly higher than your peak burst requirement. If the carrier charges per channel, calculate the cost at your peak concurrency — 200 channels at $1/channel/month adds $200 to your bill for capacity you only burst to intermittently.

SIPNEX: No channel limit. No per-channel fee. Your capacity is your bandwidth and server.

2. STIR/SHAKEN attestation level

Question to ask: “Do you hold your own STIR/SHAKEN Service Provider certificate, or do you sign through an upstream carrier? What attestation level will my calls receive?”

Why it matters: A-level attestation provides the highest trust signal to terminating carriers and analytics companies. B-level — the typical result when a reseller’s upstream carrier signs on your behalf — provides no protective benefit and can contribute to spam labeling. The answer rate difference between A and B level is typically 10 to 20 percent on the same leads.

What to look for: The carrier holds its own SP-KI certificate and signs directly. They verify your DIDs for A-level rather than passing your traffic upstream for B-level signing.

SIPNEX: Own SP-KI certificate. A-level attestation on all verified DIDs. Direct signing with no intermediate carrier.

3. Post-dial delay (PDD)

Question to ask: “What is your average PDD for US domestic outbound? What is the 95th percentile?”

Why it matters: PDD is the time between your SIP INVITE and the recipient’s phone ringing. Predictive dialing algorithms assume relatively consistent call setup times. If PDD varies from 2 to 8 seconds across calls, the algorithm over-dials on some cycles (causing abandons) and under-dials on others (causing idle time). Target: under 3 seconds average, under 5 seconds at 95th percentile.

What to look for: A carrier that monitors and publishes PDD metrics. If they cannot tell you their PDD numbers, they do not optimize for them.

SIPNEX: Sub-3-second average PDD. We monitor PDD in real time because our own dialer operations depend on it.

4. Codec support

Question to ask: “Do you support G.711u (ulaw) as the primary codec? Do you transcode or pass through natively?”

Why it matters: G.711u is the standard codec for call center quality — 64 kbps uncompressed audio with no encoding artifacts. VICIdial’s AMD (Answering Machine Detection) performs best on G.711 because the audio analysis algorithms are tuned for uncompressed waveforms. Transcoding (converting between codecs at the carrier level) adds latency and can introduce quality degradation. Native pass-through is preferred.

What to look for: G.711u as the default, G.729 as a fallback, no mandatory transcoding. Some carriers force G.729 on all traffic to save bandwidth on their network — this saves them money at the cost of your audio quality.

5. Pricing transparency

Question to ask: “Can I see your rate card without talking to a salesperson?”

Why it matters: Hidden pricing means price discrimination — you pay what the sales team thinks you will accept, not what the service is worth. Transparent pricing means you can compare on an apples-to-apples basis.

What to look for: Published per-minute rates, clearly stated billing increments (6-second is standard), no per-channel fees, no platform fees, no setup fees, no porting fees. If anything is missing from the public pricing, ask specifically about it — undisclosed fees are the most common source of billing surprises.

SIPNEX: Published rates. 6-second billing. No hidden fees.

6. Support quality

Question to ask: Call the support number and ask: “What codec and DTMF mode do you recommend for VICIdial’s AMD detection?”

Why it matters: If support cannot answer this question without escalating, they do not understand your workload. When your campaign goes sideways at 2 PM on a Tuesday — answer rates cratered, one-way audio on half your calls, agents hearing echo — you need someone who can diagnose the problem in real time, not someone who opens a ticket and promises a 24-hour response.

What to look for: First-call resolution on dialer-specific questions. Familiarity with VICIdial, Asterisk, FreePBX. Ability to check SIP traces, media quality metrics, and attestation status while you are on the phone.

SIPNEX: Our support runs VICIdial. They have the answer because they have solved the problem on their own systems.

7. Network compliance

Question to ask: “Are you registered in the FCC Robocall Mitigation Database? Can I verify your filing?”

Why it matters: Carriers not registered in the RMD may have their traffic blocked by downstream carriers. Your calls never reach the terminating network. This is an invisible problem — your dialer shows the call as sent, but the recipient’s phone never rings.

What to look for: Active RMD filing, verifiable at fccprod.servicenowservices.com/rmd. Complete STIR/SHAKEN implementation (not just a mitigation program).

SIPNEX: Registered in the RMD with complete STIR/SHAKEN implementation.

Red flags to watch for

“Contact sales for pricing.” They are planning to overcharge you.

Per-channel fees. Artificial scarcity that penalizes predictive dialing.

1-minute billing increments. Inflates your costs by 20 to 30 percent on short calls.

“We partner with [upstream carrier] for STIR/SHAKEN.” Your calls get B-level.

Long-term contracts with early termination fees. They expect you to want to leave.

Support only via ticket/email. When your campaign is down, a 24-hour ticket response costs you thousands in lost revenue.

Frequently asked questions

What is the most important factor when choosing a carrier for predictive dialing?

STIR/SHAKEN attestation level has the single largest impact on your bottom line. The difference between A-level (direct carrier) and B-level (reseller) attestation typically represents 10 to 20 percent in answer rate — which translates directly into revenue. Unlimited concurrent channels is the second most important factor, because a channel cap physically prevents your dialer from executing its algorithm. All other factors (PDD, codec, pricing, support) matter, but attestation and channels are the two that most often determine whether a carrier is viable for predictive dialing at scale.

Can I test a carrier before committing?

You should. Any carrier that refuses a trial or proof-of-concept is a carrier that does not want you comparing their performance to alternatives. On SIPNEX, we provision trunks in 24 hours with no long-term commitment. Run a test campaign alongside your current carrier and compare answer rates, PDD, audio quality, and total cost. The data will tell you everything the sales call cannot.

Should I use one carrier or multiple?

Most operations should start with one primary carrier and evaluate the need for a secondary. Reasons to use multiple carriers: failover redundancy (if your primary has an outage), cost optimization (route traffic to the cheapest carrier per destination), and A/B testing (compare answer rates between carriers). The tradeoff is management complexity — each carrier has different credentials, DID inventories, and attestation characteristics. For most mid-size operations, a single reliable carrier with carrier-level failover is simpler and sufficient.

---

SIPNEX is the carrier built for predictive dialer operators. Unlimited channels, A-level STIR/SHAKEN, sub-3-second PDD, published rates, and support that speaks VICIdial. Test us against your current carrier.