Most SIP trunk contracts get signed on the strength of a rate card and a sales call. The problems show up later: a channel cap you did not know existed, B-level attestation because the provider signs through someone else, a billing increment that quietly inflates every invoice. By then you are mid-campaign and switching costs are real.
The fix is asking the right questions before you sign — and knowing what a bad answer sounds like. This matters most for high-volume outbound operations, where the trunk is not a commodity input: it determines answer rate, abandon rate, agent utilization, and revenue per hour. A predictive dialer with 50 agents can be undone entirely by a SIP trunking contract that looked fine on paper.
Here are the seven questions, why each one matters, and the answers that should end the conversation.
The seven questions
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 provider caps at 50 or 100 channels, every call above the cap is rejected, often with a SIP 503 or 403 depending on the carrier. The algorithm cannot execute its predictions. Agents sit idle while the provider throttles your capacity.
What to look for: “Unlimited” or a cap significantly higher than your peak burst requirement. If the provider charges per channel, calculate the cost at your peak concurrency — at a hypothetical $1 per channel per month, 200 channels add $200 to your bill for capacity you only burst to intermittently.
2. STIR/SHAKEN attestation level
Question to ask: “Whose STIR/SHAKEN certificate signs my calls — yours or an upstream carrier’s? What attestation level will my calls receive, and why?”
Why it matters: A-level attestation is the highest trust signal a signing provider can send to terminating carriers and analytics companies. B-level is the usual result when the signer cannot vouch that you are authorized to use the calling number — common when a reseller hands your traffic to an upstream carrier. Analytics engines treat partial (B) attestation as a lower-trust signal; no carrier publishes a specific answer-rate gap.
The rules on who signs changed in 2025. Under the FCC’s third-party authentication rules (FCC 24-120, effective September 18, 2025), a provider with a STIR/SHAKEN obligation may pay a vendor to do the signing, but it must make the attestation decisions itself and sign with its own certificate, not the vendor’s.
What to look for: Calls signed under the provider’s own certificate, even if a vendor runs the signing platform, and A-level on DIDs the provider has verified you are authorized to use.
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). A reasonable rule of thumb: under 3 seconds on average, under 5 seconds at the 95th percentile.
How to measure it: Take it from your own SIP traces. Time each call from your INVITE to the first provisional response after 100 Trying — usually 180 Ringing or 183 Session Progress. RFC 6076 defines this as session request delay. Report the average and the 95th percentile separately.
What to look for: A provider that tracks PDD and will share its numbers. If it cannot, measure PDD yourself during a trial.
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 PCM audio with no low-bitrate compression. 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 providers 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: Published rates let you compare providers like for like before you spend time on sales calls. Unpublished pricing is not automatically a problem — many carriers quote custom rates for volume — but then get the full rate card, billing increment, and fee schedule in writing.
What to look for: Published per-minute rates, clearly stated billing increments (short increments such as 6/6 or 12/6 are common for dialer traffic), 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.
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.
7. Network compliance
Question to ask: “Are you registered in the FCC Robocall Mitigation Database? Can I verify your filing?”
Why it matters: Providers 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).
Red flags and follow-up questions
No public rate card. Not a deal-breaker on its own, but ask for rates, the billing increment, and every fee in writing before you sign.
Per-channel fees. Ask what your peak burst concurrency would cost, not your average.
1-minute billing increments. A 45-second call bills as 60 seconds on 60/60 and as 48 seconds on 12/6, so 60/60 bills 25 percent more time on that call. Run your own average call length through each provider’s increment before comparing per-minute rates.
A vague answer on STIR/SHAKEN. If a provider says it partners with another company for signing, ask whose certificate is on your calls and what attestation level they get. Since September 2025, a provider with a STIR/SHAKEN obligation must sign under its own certificate even when a vendor does the signing.
Long-term contracts with early termination fees. Ask whether you can start month to month while you test the route.
Support only via ticket/email. When your campaign is down, a 24-hour ticket response can cost you a full day of dialing.
An AUP that bans your traffic. A great rate means nothing if the acceptable-use policy prohibits autodialing — check the SIP providers that allow auto dialers breakdown before you sign.
How SIPNEX answers these questions
Ask us the same seven questions and compare the answers with any other provider’s. Point by point: no channel limit and no per-channel fee — your capacity is your bandwidth and server. Our own SP-KI certificate, A-level attestation on all verified DIDs, direct signing with no intermediate carrier. Native G.711u pass-through. For PDD, measure it on a test campaign the way question 3 describes.
The commercial answers: published rates with 12/6 billing — a 12-second minimum, then 6-second increments — and no hidden fees. For support, put question 6 to our team and judge the answer yourself. SIPNEX is registered in the RMD, with calls signed under our own certificate.
Channel, attestation, codec, and pricing details are on our predictive dialer SIP trunk carrier page.
Frequently asked questions
Which of these questions matters most?
For outbound at scale, attestation and channel policy usually decide whether a provider is viable. Analytics engines treat partial (B) attestation as a lower-trust signal; no carrier publishes a specific answer-rate gap, so compare providers on your own leads. A channel cap physically prevents a predictive dialer from executing its algorithm. PDD, codec, pricing, and support all matter, but attestation and channels are the two that most often rule a provider out.
Can I test a provider before committing?
You should. Ask for a trial or proof-of-concept before you sign; without one, you are comparing sales claims instead of results. Put both trunks on the same campaign for a week and let the CDRs decide — answer rate, PDD, audio quality, and effective cost per connected minute on identical lead lists. The data will tell you what a sales call cannot.
Should I use one provider or multiple?
Most operations should start with one primary provider and evaluate the need for a secondary. Reasons to use multiple: 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 providers). The tradeoff is management complexity — each provider has different credentials, DID inventories, and attestation characteristics. For most mid-size operations, a single reliable provider with carrier-level failover is simpler and sufficient.
Ask these seven questions of any provider you are evaluating, including us. If you would rather put them to a person, reach the SIPNEX team.
Keep reading.
The carrier built by operators, for operators.
FCC-licensed carrier with its own STIR/SHAKEN SP certificate. Operator-owned. SIP trunks built for operators who dial at volume.