An outbound dialer earns its keep on three questions: when to dial each contact, what number shows up on the recipient's phone, and what the calls actually sounded like afterward. Orbit's dialer answers the first with per-contact best-time scheduling, the second with a caller-ID coverage report that names the area codes your number pool leaves uncovered, and the third with post-call sentiment scoring on every connected call. None of this replaces the compliance decisions the operator makes — consent posture, quiet hours, and do-not-call enforcement stay tenant-owned controls the platform applies exactly as configured. This guide walks through each capability, then runs a collections campaign through all of them.
Outbound dialing in 2026: the operator owns the compliance posture
US outbound calling sits under the TCPA's federal 8 AM–9 PM recipient-local calling window, and several states layer stricter rules on top. The regulatory exposure lives with the organization placing the calls — so the meaningful question for any platform is whether enforcement is deterministic and configurable, not whether the platform absolved you of the decision.
In Orbit the dialing-attempt gates are tenant-owned settings the operator sets explicitly: the quiet-hours window, do-not-call list scope, consent-state requirements, and per-campaign dialing mode. Every dial attempt passes the configured gates at dispatch time, and the choice between automated modes (preview, progressive, predictive, agentless) and the genuinely manual no-autodialer mode is itself a compliance-relevant decision the operator makes per campaign, per the compliance controls reference. The platform's job is to apply the tenant's rule deterministically on every attempt; the tenant's job is to decide what that rule is. Everything below assumes that division of responsibility.
Best-time-to-dial: each contact gets a predicted answer window
A campaign without best-time scheduling dials its list top-to-bottom in upload order. The contact most likely to answer at 10 AM gets dialed at 8 PM because that is where the upload placed them, and the pacing engine burns attempts on people who were never going to pick up.
When a campaign opts into best-time-to-dial, list-upload enrollment stamps each contact's next dial time with the engagement engine's predicted highest-answer-probability window, resolved on the voice channel. Contacts with no engagement history get a timezone-aware local-morning fallback, and contacts whose history comes from SMS or email still get a meaningful window through the engine's cross-channel fallthrough. The existing claim order then dials each contact inside their predicted window instead of in list order.
Two properties matter for operations. First, the tenant's quiet-hours gate still evaluates every dial at dispatch time, so a predicted window can never push a call outside the configured envelope — the same recipient-local rule that governs send-time optimization for SMS. Second, the feature fails open: if the engagement profiler is unavailable, enrollment falls back to immediate scheduling, so an upload is never blocked because the optimization layer is down.
The opt-in is per campaign, which suits teams that want to measure the lift. Run one campaign in list order and one with best-time enabled against the same audience, then compare answer rates in the campaign analytics.
Caller-ID coverage: find the area codes your pool cannot present
Local presence — presenting a caller ID whose area code matches the recipient's — measurably lifts answer rates on North American dialing. Its silent failure mode is coverage: when the campaign's number pool contains no number in the recipient's area code, the dial falls back to the campaign's default caller ID without telling anyone, and the answer-rate lift you configured for never materializes.
Orbit's coverage report cross-references the campaign's recipient area-code distribution against the area codes the pool can actually present, layers the rolling answer rate per destination area code on top, and returns a ranked list of the highest-volume uncovered area codes. That turns "buy more local numbers" into a specific work order: these three area codes account for this share of the list, and a local number for each is available from Numbers in the Orbit dashboard. Non-North-American destinations are reported for totals but deliberately never recommended — there is no NANP area code to match.
This is also where caller-ID reputation belongs. A number presented across thousands of unwanted calls accumulates spam flags; a pool whose coverage tracks the actual recipient distribution spreads presentation across many numbers and keeps any single one from burning out.
Sentiment on outbound calls: what happened, at campaign level
Every originated attempt already flows through the post-call synthesis service, which scores the call's sentiment on a −1 to +1 scale and labels it positive, neutral, or negative. The campaign sentiment view aggregates those scores so an operator can read, per campaign: the share of connected calls that landed in each bucket, the average sentiment score across scored calls, and the sentiment breakdown by call outcome — answered, voicemail, no answer, and so on. Calls not yet scored by the post-call service show in their own unscored bucket rather than being silently dropped from the tally.
That last distinction matters more than it looks: a literal zero average would read as dead-neutral sentiment, so unscored attempts are excluded from the average entirely. The view answers the QA question a collections or reminders program actually asks — not "did we connect," but "how did the conversations go, and which outcomes carry the negative calls." The scoring model is the same one that scores messages across channels, so the vocabulary stays consistent between the dialer card and the inbox view.
Manual dialing and the carrier health gate
Two operational safeguards frame the AI-assisted modes.
Manual, no-autodialer dialing. Under the strictest reading of the TCPA, calling a wireless number without prior express consent requires a genuinely manual path — a human picks one number and clicks to dial it, with no automated equipment storing or sequencing numbers. Orbit ships that as a distinct path rather than a campaign mode: it originates a single operator-supplied number through the same click-to-call flow as the softphone, with every pre-flight guard — quiet hours, do-not-call, fraud-velocity, frequency caps, billing — applied unchanged. The call record is stamped as manually originated by a named operator, so the audit trail can prove the no-autodialer posture. Campaign dialing and manual dialing coexist deliberately; the mode choice is the tenant's compliance call.
The carrier health gate. Before originating batches, the dialer checks the tenant's recent attempt history against a sliding five-minute window. When more than half of recently terminated attempts failed at the transport layer — the softswitch rejected or never received the call — the gate blocks new origination with a 503 and resets the claimed contacts to pending, instead of burning the campaign's attempt budget on calls that could never connect. Attempts blocked on list quality (unreachable numbers) or compliance gates do not count toward the failure signal, and a minimum sample size keeps low-volume tenants from tripping false positives. When the carrier path recovers, the next claim cycle resumes normally.
Worked example: a collections campaign with sentiment QA
A collections team runs weekly dialing against a consent-scrubbed repayment list.
- Upload with best-time enabled. The repayment list enrolls with each contact's predicted answer window stamped at upload. Contacts with engagement history get their modeled window; cold contacts get the local-morning fallback. The list no longer dials in CSV order.
- Check coverage before launch. The coverage report on the campaign shows seven uncovered area codes carrying 14% of the list. The team buys one local number for each of the top three in Numbers and re-runs the report — coverage moves from list-wide to targeted on the highest-volume codes.
- Launch in progressive mode. The compliance gates the operator configured — the organization's quiet-hours window and do-not-call scope — evaluate every dial at dispatch. Best-time windows that would land outside the configured envelope are simply never dispatched outside it.
- Watch the health gate, not just the dashboard. Mid-campaign, the wholesale carrier path degrades. The health gate blocks origination inside minutes, the claimed contacts return to pending, and the campaign's attempt budget survives the outage instead of draining into failed originations.
- Read sentiment, not just connects. The campaign sentiment view shows answered calls trending negative on Thursdays — the day the script leads with the balance amount. The script gets revised and the following week's average score recovers, with the unscored bucket confirming scoring coverage rather than polluting the average.
Each step uses a control the operator opted into deliberately; the AI-assisted layers narrow down when and how to dial, while the tenant-owned compliance gates decide whether.
What to configure first
If you take one thing from this guide: turn on best-time-to-dial for one campaign and open its coverage report before the next upload. Those two steps convert outbound from "dial the list" into "dial each contact when they answer, from a number that matches," and everything else in the dialer — sentiment QA, the health gate, the manual path — compounds on top of that.
---
For the per-campaign opt-in, coverage report, and sentiment view, see the dialer section of the Orbit dashboard and the compliance send-gates reference. Outbound features are available on all Orbit plans.