The classic SDR job description asks one person to research accounts, write messages, make calls, handle objections, manage a sequence tool, and keep the CRM clean. That worked when each of those was shallow. As each got deeper, the role started splitting.
Three jobs are now visibly distinct.
1. The researcher
Owns: who gets contacted and in what order. Account selection, list building, signal interpretation, prioritisation.
Good at: pattern recognition and knowing when to stop digging. Comfortable being judged on a leading indicator rather than a meeting count.
Measured by: the quality of the list downstream, which shows as connect rate and reply rate on accounts they selected, not by their own activity.
Why it is a separate job: the skill is judgment under time pressure, and mixing it with calling means it always loses. Research is the thing that gets skipped when the meeting count is behind.
2. The conversationalist
Owns: live human interaction. Calls, replies, objections, and the first real conversation.
Good at: staying the same person under resistance, listening, and asking one more question. This is the least automatable skill in the function and increasingly the scarcest.
Measured by: connect-to-conversation and conversation-to-meeting, not by dials.
Why it is a separate job: it needs volume of live reps to improve, and a person splitting attention across list building never gets enough of them.
3. The operator
Owns: the machinery. Sequences, deliverability, data hygiene, routing, integrations, and the reporting the other two run on.
Good at: systems thinking, and caring about a bounce rate.
Measured by: deliverability, data accuracy, and cycle time, none of which are meeting counts.
Why it is a separate job: it is the one where a single mistake damages everything. A domain reputation problem costs months, and it is nobody's priority when everybody's target is meetings.
Why this matters more than it looks
Hiring. One job description produces one profile: usually an energetic generalist who is genuinely good at none of the three. Ramp fails, and the diagnosis blames onboarding.
Metrics. A single meeting-count target applied across all three punishes the researcher and the operator for doing their actual jobs well. This is the most common structural mistake in the function.
Career path. Three jobs means three ladders. The conversationalist becomes an AE. The researcher becomes a market or demand analyst. The operator becomes RevOps. A single ladder into AE loses two of your three best people.
What small teams should do
You cannot hire three people for a team of four. But you can name the three jobs and assign them, even part time, and measure each against its own number. The failure is not lack of headcount. It is pretending one metric describes three jobs.
Where this sits
Researcher is Market and Signal, conversationalist is Message, operator is Motion and Measurement, in the Tenbound Pipeline Architecture Standard. The maturity model describes when the split becomes necessary: around Orchestrated, the operator role stops being optional.
Related: building an SDR team and where SDRs should report.