The Inbound SDR Agent Playbook

The short version

An inbound SDR agent can reliably do four things: respond to new enquiries within seconds, qualify against your criteria, book meetings into the right calendar, and keep the CRM accurate. It should not negotiate, quote bespoke pricing, or handle complex objections. Treat it as a speed-to-lead and hygiene machine, and the numbers work; treat it as a replacement seller, and they do not.

Why speed-to-lead is the whole business case

Response latency is the single most controllable variable in inbound conversion. Enquiries answered within minutes convert materially better than those answered hours later, and most teams cannot staff instant response across time zones and weekends. An agent that replies in under a minute, every hour of every day, captures value that is currently lost to the calendar.

What the agent does, step by step

1. Acknowledge immediately. A personalised reply referencing the specific enquiry, sent within seconds.
2. Enrich. Look up the company, size, sector and existing CRM history before asking anything.
3. Qualify conversationally. Three to five criteria maximum — need, timing, scale, decision role. Never interrogate.
4. Route. Qualified leads to a booking link on the right rep's calendar; unqualified to nurture with a genuinely useful resource; existing customers to their account owner.
5. Log. Write the transcript, qualification answers and outcome to the CRM as structured fields, not a note blob.
6. Follow up. Two or three spaced follow-ups on no response, then stop. Hard cap enforced in code.

Qualification design

Write your criteria as explicit, checkable statements before building anything: for example, company size above a threshold, a named use case, a timeframe within two quarters, and a contact who is a decision maker or direct influencer. Score each as met, unmet or unknown. Route on the score; never let the model invent its own criteria. Review the boundary cases weekly for the first month and adjust thresholds — that review loop is where most of the accuracy gain comes from.

Guardrails that are not optional

• Never quote prices outside a published list.
• Never commit to delivery dates, contractual terms or discounts.
• Disclose that it is an automated assistant when asked, and follow local disclosure rules.
• Hand off to a human on any request for one, any complaint, and any legal or security question.
• Hard caps on outbound message volume per contact and per hour.
• Suppression lists honoured before any send; consent status checked against your CRM.

Metrics that matter

• Median response time — should be under 60 seconds.
• Enquiry-to-meeting-booked rate, versus the pre-agent baseline.
• Meeting show rate — the honest test of qualification quality.
• Qualification accuracy — sampled human review of 30 conversations a week.
• Handoff rate and the reasons behind it.
• CRM field completeness on agent-created records.

Watch show rate closely. An agent that books more meetings with worse fit looks like success in week one and a problem in week six.

A worked example

A B2B services firm received roughly 300 inbound enquiries a month with a median first response of four hours and a 12% enquiry-to-meeting rate. After deploying an inbound agent with four qualification criteria and calendar routing, median response fell to 40 seconds and the meeting rate rose to 19%. Show rate held at 71%, within two points of the human baseline, and CRM completeness went from 54% to 96% of fields populated. The team did not shrink; the two SDRs moved to outbound, where the agent does not operate.

What the agent should not do

Cold outbound at scale invites deliverability and compliance problems and rarely reflects well on the brand. Complex discovery, multi-stakeholder navigation and pricing negotiation belong to humans. Renewal and escalation conversations belong to account owners. Keeping this boundary explicit is what preserves trust in the system internally.

Implementation sequence

Week one: instrument the current baseline. Week two: write qualification criteria and message templates, and get them approved. Weeks three to four: build the CRM and calendar integrations with a human-in-the-loop review of every conversation. Week five: release automatic sending for high-confidence cases only. Week six onward: widen scope as sampled quality holds.