Why dialer mode matters for recruitment agencies
Outbound calling remains the fastest way for agencies to reach passive candidates, schedule interviews, and close placements. The choice between predictive, progressive, and preview dialing directly influences three business outcomes: call success rate, regulatory exposure, and per‑call cost. In a market where Ofcom now treats abandoned and silent calls as misuse, the wrong dialer mode can generate fines of up to £2 million under the Communications Act 2003.
Regulatory backdrop (2026)
Ofcom’s January 2026 refresher class removed the historic "soft benchmark" of a 3 % abandoned‑call rate. The regulator now expects a de‑facto zero‑tolerance threshold for both abandoned calls (recorded messages played when no agent is available) and silent calls (no audible response). Penalties are tiered by severity, but any breach triggers a mandatory compliance review and can damage an agency’s reputation among candidates.
Predictive dialing in an AI‑driven environment
Predictive dialers were designed for human agents. By dialing at a ratio greater than one call per available seat, they keep agents busy and reduce idle time. In a fully AI‑driven estate, however, the over‑dialing logic creates two new failure modes:
- Abandoned calls: the system dials a candidate, the call is answered, but no AI session is ready, so a pre‑recorded message plays.
- Silent calls: the call connects, the AI has not yet streamed audio, and the recipient hears nothing.
Both outcomes are classified as "misuse" by Ofcom. The traditional cost‑saving benefit of predictive dialing is therefore outweighed by compliance risk when AI sessions are provisioned on demand rather than pre‑allocated.
Progressive dialing: a compliance‑first alternative
Progressive dialers place exactly one call for each AI session that is ready. This eliminates the over‑dial that generates most abandoned calls. The remaining risk is false‑positive answer‑machine detection (AMD), which can still produce silent calls if the system misclassifies a human answer as a machine.
Dialr Voice’s default progressive mode uses a two‑second information‑message window. Within those two seconds the caller must identify the organisation and provide an opt‑out number, satisfying Ofcom’s requirement for transparent outbound calls.
Preview dialing for high‑value outreach
Preview mode inserts a human‑review step before the call is placed. Recruiters can verify candidate data, add personalised notes, and confirm compliance flags. The trade‑off is lower volume, but the approach is ideal for senior‑executive searches, regulated sectors, or any outreach where a single mis‑dial could have outsized reputational impact.
Elastic AI concurrency and its economic impact
McKinsey’s 2025 AI adoption report shows that 88 % of enterprises use AI, and 33 % run it in production. When AI sessions can be spun up on demand, the scarcity that predictive dialing once solved (idle human agents) disappears. However, concurrency is bounded by model‑provider limits, carrier rate‑limits, and burst‑traffic caps. These constraints become a scheduling problem rather than a compliance problem.
Comparative table of dialer modes
| Dialer Mode | Over‑dial Ratio | Abandoned‑Call Risk | Typical Use Case | Compliance Note |
|---|---|---|---|---|
| Predictive | >1 (e.g., 1.2‑1.5) | High – abandoned calls common | High‑volume, low‑value outreach | Requires strict monitoring; Ofcom zero‑tolerance |
| Progressive | 1 (one call per ready AI session) | Low – abandoned calls rare | Bulk candidate outreach, compliance‑sensitive campaigns | AMD must be tuned; false positives monitored |
| Preview | 1 (human‑initiated) | Very low – human validation | Senior‑executive, regulated, or high‑value contacts | Manual step adds cost but eliminates compliance risk |
Recommended dialer strategy for recruitment agencies
- Default to progressive dialing for all bulk candidate outreach. This aligns with Ofcom’s zero‑tolerance stance while keeping AI utilisation high.
- Reserve preview mode for senior‑level or compliance‑sensitive contacts where a human check adds value.
- Configure AMD conservatively or disable it entirely. If AMD is used, set the false‑positive threshold to below 0.5 % and review metrics weekly.
- Implement real‑time observability through call‑level tracing, abandoned‑call counters, and silent‑call alerts. Dashboards should flag any metric above 0.1 % for immediate investigation.
- Synchronise CRM data (Salesforce, HubSpot, etc.) with the dialer to feed up‑to‑date Do Not and suppression lists. Avoid repeat calls within 72 hours; Ofcom treats such repeats as aggravating factors.
Observability and metric tracking
Effective monitoring requires three data streams:
- Call‑level logs (timestamp, candidate ID, dialer mode, outcome).
- Compliance flags (abandoned, silent, opt‑out received).
- AI‑session utilisation (concurrency, latency, error rate).
Integrating these streams into a unified dashboard enables a "zero‑tolerance" view. For example, a spike in silent calls often correlates with a carrier rate‑limit breach; the dashboard can trigger an automatic throttling rule.
CRM integration best practices
Recruitment agencies typically store candidate status, consent, and DNC preferences in a CRM. A bi‑directional integration should:
- Push new candidate records to the dialer in near‑real time.
- Pull call outcomes back into the CRM for pipeline updates.
- Apply suppression rules automatically (e.g., no call within 72 hours after a negative response).
Most modern CRMs expose REST endpoints; a lightweight middleware can map those endpoints to the dialer’s API without custom code.
Cost considerations in an elastic AI estate
Predictive dialing historically reduced per‑call cost by keeping human agents busy. With AI, the cost model shifts to "session‑minute" pricing. Progressive dialing maximises session utilisation because each session is guaranteed a call. Preview mode adds a human‑review cost but reduces wasted AI minutes on high‑risk contacts.
Scenario analysis (2025 pricing):
- Predictive: 1.4 calls per AI session, 0.2 % abandoned, £0.012 per minute → £0.016 effective cost per call.
- Progressive: 1 call per session, 0.05 % abandoned, £0.012 per minute → £0.012 per call.
- Preview: 1 call per session, 0 % abandoned, £0.012 per minute + £0.003 human review → £0.015 per call.
While progressive is cheapest, preview may be justified for senior‑level placements where a single successful call yields a high‑value fee.
Implementation checklist
- Audit current outbound volume and compliance incidents.
- Map candidate segments to dialer modes (bulk vs senior).
- Configure AMD thresholds; run a 2‑week pilot and record false‑positive rate.
- Set up real‑time dashboards for abandoned and silent call metrics.
- Validate CRM‑dialer sync with a test batch of 500 records.
- Document escalation procedures for compliance alerts.
Future‑proofing the outbound strategy
As AI model providers increase concurrency limits, agencies can scale progressive dialing without risking compliance. However, the regulatory landscape may evolve. Maintaining a compliance‑first mindset—zero abandoned calls, transparent opt‑out, and up‑to‑date DNC lists—will keep agencies regardless of technology changes.
Internal link suggestion: AI voice overview
Internal link suggestion: Outbound compliance checklist
Planning to improve your business calling operations? Get in touch with ProTalk Dialler to discuss your requirements and evaluate the right approach.
Internal link suggestion: contact centre and calling guides
Internal link suggestion: ProTalk Dialler pricing and plans
