The SDR seat is splitting in four

The SDR role is changing quickly but there is much ado about how AI will change the role. Most articles on this topic start with a verdict. I’m not going to do that. Let’s work through it together, the way I would in a planning doc, before I tell you where I’ve landed.

Here’s the conversations I keep getting pulled into:

• Sales leadership asks whether two open SDR seats really need backfilling.

• Finance asks why outbound headcount keeps climbing while reply rates don’t.

You give an answer. You’re not sure it’s the right one.

This post is a sponsored deep dive with 11x. The framework here is mine, pulled from RevOps conversations I’m having every week, and the perspective on what’s actually shipping at the enterprise level came out of a long conversation with their CEO Prabhav Jain.

I’ve been there. Multiple times. The last time was at a startup where leadership set the new fiscal year’s target at 275% of the previous year. By June we were pacing at 30%. Half the year done. Less than a third of the way there.

I pulled the fire 🚨 alarm.

Different problem, same shape. When the math underneath the SDR motion stops working, RevOps gets the call. Right now, the math is shifting again.

What I’m actually seeing

Pull an SDR’s calendar from a year ago. Put it next to last week. List building. Contact enrichment. First-touch emails. Follow-up cadences. Inbound triage. A year ago, that filled most of the week.

Now? A real chunk of it runs somewhere else.

The hallway feeling matches the data, but I’d encourage you to pull your own numbers before you take anyone else’s at face value. I’ve seen too many AI SDR pitches built on suspiciously clean industry benchmarks. Look at your own funnel. Compare Q1 to Q4 of last year. Compare per-rep volume now to twelve months ago. Compare your reply rates by segment. The story will tell itself.

When I do that exercise with operator friends, the same pattern keeps showing up:

• Per-rep outbound volume is up. A lot.

• Reply rates are down or flat. The volume isn’t translating one-to-one.

• Junior SDR seats are getting harder to justify. Senior SDR and account specialist seats are getting easier.

• New roles are showing up that didn’t exist 18 months ago. GTM Engineer. Pipeline Operations Analyst. RevOps Architect.

The question underneath the headcount question is this. What should the human revenue team look like once the execution layer can run on its own?

That’s what I want to work through. Four functions I see emerging, an honest read on whether your org should move this quarter or wait, and a 90-day plan if the answer is yes.

The execution layer is becoming infrastructure

Let me give the current model its due. Two SDRs. A solid cadence tool. A clean ICP. A manager running call reviews. That machine generated real pipeline for a decade. A lot of us got promoted on the back of it.

Nothing here is a claim the model was wrong.

What’s changed sits underneath the model. The cost of the execution layer.

Think back to when marketing teams adopted marketing automation around 2014. The repetitive send-and-nurture work stopped being headcount. It became infrastructure. You stopped counting it in heads and started counting it in throughput.

11x’s take on this is more specific than most. They’re not trying to be a better cadence tool. They’re trying to rebuild the execution layer from scratch, with the building blocks each owned by an agent. Lead finding, enrichment, scoring, research, message writing, CRM integration. Across phone, email, chat, SMS, even WhatsApp. The pitch isn’t to “replace your SDR,” it’s “give your team an orchestration layer that didn’t exist before.”

Whether that pitch holds at scale is something each operator has to test. But the framing matters. It’s the difference between buying a tool and adopting infrastructure.

Same move. Sales development now.

Here’s the mechanic. When the execution layer runs as infrastructure, the marginal cost of one more touch drops close to zero. 11x is one of the few I’ve seen actually shipping this at enterprise scale, with customers like Xerox, Sage, and Checkr running it in production. The math underneath outbound changes.

A few caveats I want on the table before we go further, because I’ve watched too many operators get burned by skipping these.

One. Fully autonomous AI SDRs have not replaced human teams at any scale I’ve seen. The deployments that actually work keep humans in the loop. The ones that pulled humans out entirely mostly got canceled. The vendors who pretend otherwise are lying to you.

Two. Vendors like 11x have caught up to this. Prabhav, CEO at 11x, put it to me like this: “AI didn’t take their job, it took the worst parts of it.” That framing matters because it sets the right expectation. Their outbound worker Alice runs the execution layer. Their inbound worker Julian handles voice and chat qualification. The humans run the parts that need humans.

Three. The conversation about AI SDRs gets framed wrong almost every time. People ask “what’s the reply rate” when the real question is “what does this unlock that I couldn’t do before.” AI SDRs or AI inbound sales agents (e.g. 11x’s Alice and Julian) handle outbound execution end to end and inbound qualification across every channel. The win isn’t a better email, but rather coverage your team couldn’t build any other way.

Here’s the reframe I’d offer.

We’ve long treated pipeline generation as a headcount problem. Reasonable instinct. It’s how the job worked. But what if it’s closer to the data warehouse now? Something you architect, staff thinly, and measure?

That reframe changes which humans we actually need above it.

Four functions emerging from one role

If the execution layer becomes infrastructure, the human work above it sorts into four jobs. Let me list them, then we’ll pressure-test each.

1. Pipeline architecture and orchestration

2. High-stakes account ownership

3. Inbound conversion and edge-case handling

4. Pipeline quality and feedback-loop ownership

Two notes before we dig in. These are functions, not titles. Most teams running this redesign today fold two of them into a single role, and grow into the rest. And plenty of teams keep the SDR title and just change what sits behind it. The functional split holds up better than any title taxonomy you’ll see on LinkedIn this quarter.

Pipeline architecture and orchestration

This function designs the system that produces pipeline. It owns the orchestration layer. Data sources, intent signals, AI agents, CRM, the wiring between them. It does not run a floor of SDRs.

There are fewer SDRs to run.

The titles vary. GTM Engineer. RevOps Architect. Pipeline Systems Manager. The work is system design.

The skill profile leans technical. SQL. A no-code workflow tool, or light Python. Deep CRM fluency. Signal taxonomy. Enough prompt literacy to direct an agent. Deliverability fundamentals.

Most people doing this came from RevOps, marketing ops, or sales engineering. Some came from an SDR seat. Specifically the ones who were already drifting toward systems and away from the dialer.

Comp lands in senior IC or first-line manager bands. Often comparable to a senior AE, without the swing.

One thing to watch. It’s tempting to slot your best SDR manager here and call it done. They earned the look. The work is genuinely different though. An SDR manager coaches humans through pipeline reviews. This function diagnoses why segment three of the ICP stopped converting and rewrites the targeting logic. Different muscle.

High-stakes account ownership

Here’s the part of the human SDR job that holds up well. Account ownership at low volume and high specificity.

A small set of named accounts. Three per rep. Sometimes ten. Rarely more.

Real research. Multithreading. Reading a buying committee.

The seat looks closer to a junior AE than a senior SDR. Senior tier roles I’m seeing posted increasingly carry titles like Strategic Account Development or Named Account ADR, with small account loads and AE shadowing built into the ramp.

The skill profile is sales. Account research. Executive conversation. Deal psychology. Light technical depth depending on the category. The human still owns this because no current AI agent reads committee politics well.

Comp shifts toward higher base, lighter commission, faster path to AE. Some teams skip the SDR-to-AE jump entirely and hire straight into a hybrid junior AE role.

One thing to watch. The instinct is to make this the reward seat for your longest-tenured SDR. Filter for strategic instinct instead. Years in seat won’t tell you who has it. A great list builder is not automatically a great account owner.

Inbound conversion and edge-case handling

Speed-to-lead used to be the whole inbound game. We built rotations and SLAs around it. That work mattered. It moved real numbers.

The bottleneck has shifted.

Voice agents now call inbound leads instantly. 11x’s inbound worker Julian has twenty-second response time across phone, chat, SMS, WhatsApp, and email. The system adapts to whichever channel the prospect actually engages on. Since speed is handled at the infrastructure layer, what’s left is the conversation itself. The human seat doesn’t go away and instead moves to the conversations the agent decided needed judgment.

This function owns the handoff moment. The ninety seconds after a prospect picks up. Read the energy. Find the unstated need. Book or advance the next step. It also takes the conversations the agent flagged as messy. Compliance-heavy deals. Multi-product cross-sell. Anything off-script.

The skill profile is discovery instinct and fast pattern matching. Comfort reading a conversation that arrived as a transcript or chat log. Most people here came from senior SDR or junior AE seats.

One misread to flag. This is not a queue worker watching dashboards. Most inbound volume clears before a human sees it. Every conversation that does reach the human is one the agent already decided needs judgment.

The stakes per conversation go up.

Pipeline quality and feedback-loop ownership

This function exists because of volume. An agent books 40 meetings a month. 12 convert. Someone has to own the other 28. Find the pattern. Feed it back into the system.

Half analyst, half coach.

The role wasn’t viable before because the volume wasn’t there to do statistics on.

What the function does, plainly. Audits agent output for false positives. Finds the ICP segments where the agent is missing and adjusts targeting. Watches for messaging drift. The slow slide to generic tone when a self-learning loop optimizes reply rate over reply quality. Runs sequence and persona tests. Owns the loop between sales and whoever configures the agent.

The skill profile. Comfortable in data, opinionated about copy, mechanical about funnel diagnosis. Often a RevOps or marketing ops background. Or the SDR who always liked the analytics more than the calls.

Comp lands at senior IC. Sometimes a management track. Sits in RevOps, or in a new AI operations group some companies are standing up to oversee agents across sales, marketing, and CS.

One misread here too. People hear “quality” and picture a QA reviewer. This role looks forward. The job is making the system better. Writing up yesterday’s misses is a small part of it.

Five gates to clear before we redesign anything

Before we take any of this to sales leadership, we owe ourselves an honest check. This redesign isn’t a gradient you slide along. There are gates.

Miss one and the productivity unlock doesn’t show up. You lose the SDRs you had. You pay for an agent that misses. That’s a worse spot than where you started.

Gate 1. Data quality. A CRM full of stale contacts and mismapped accounts breaks any agent. Garbage in, garbage out. Same rule we’ve had since SugarCRM. The agent writes to the wrong people at the wrong companies with the wrong context. Quick test. Can you name the percentage of your top 500 accounts with clean primary contact data? If you can’t, that’s the read I’d bring to leadership. We’re not ready yet.

Gate 2. Deliverability. Volume capacity means nothing if your domains get flagged. Teams running AI SDR seriously run mailbox infrastructure too. Warm pools. Rotating domains. Per-inbox throttling. A single domain handling both CS notifications and outbound lands you in spam before it lands a meeting. This line item gets forgotten in the budget. Don’t let it.

Gate 3. ICP clarity. Agents underperform hard on ICPs with high persona variance. “Decision makers at architecture firms” is not an ICP. It’s a wish. A one-paragraph ICP with three personas is ready. “Anyone who might need our software” is not.

Gate 4. Volume. The four-function split needs scale to make sense. An agent booking eight meetings a month doesn’t need a quality lead. You should be sitting in those eight meetings yourself. The split earns its keep once the patterns are statistically real. Most operators I’ve compared notes with put the line around 80 to 100 meetings a month from AI and human pods combined.

Gate 5. Motion type. This is the gate people miss. The split fits outbound-heavy and inbound-heavy mid-market and enterprise motions. It fits awkwardly on PLG-led motions, where the real work starts after a user is already inside the product. It also fits awkwardly on high-ACV, low-volume enterprise, where an agent’s marginal contribution is small next to the AE’s relationship work. A mostly product-led motion with light sales assist needs a different redesign. Roles built around activation and expansion.

Honest scoring, the way I’d do it for myself.

• Fail one gate. That’s a quarter of work.

• Fail two. That’s closer to a year.

• Fail three. This isn’t an SDR redesign conversation. It’s a RevOps foundations conversation, and the AI SDR pitch is pulling focus from the real one.

---

Where to start: a 90-day plan

When we bring this to leadership, the first move surprises people. The first move is an audit. The hiring comes later.

Days 1 to 30. Data and ICP. Pull a sample of your top 500 accounts. Score them on completeness. Contact, role, email validity, recent activity. If fewer than 80% pass, freeze the AI SDR conversation until the foundation is fixed.

In parallel, rewrite the ICP. Three paragraphs. Three personas. Named triggers. Named disqualifiers. Most teams fail this on the first try. That failure is useful information. Don’t paper over it.

Days 31 to 60. A constrained pilot. One ICP segment. One channel. One persona. Set baselines first. Meeting volume in that segment last quarter. Conversion to opportunity. Cost per opportunity. If you can’t answer those cleanly, pause and answer them.

The pilot has one job. Show what your team can actually operate.

Days 61 to 90. Hire the quality function first. Of the four, pipeline quality is the cheapest to stand up and the fastest to pay back. One senior IC. Audits agent output, diagnoses ICP fit, owns the feedback loop with whoever configures the agent.

On most teams I’ve watched run this, the hire covers itself inside a quarter.

Architecture, account ownership, and inbound conversion come after. In that order. As volume and infrastructure mature.

One more piece. The hard one.

The conversation with the SDR team you have today. The technology and the hiring are the easy parts. Sitting down with your reps is not. The honest version sounds like this.

Some of you move into the new functions. Most of you need new skills to get there. A few of you move up into junior AE seats. A few of you will leave.

That’s heavy. The instinct is to defer it. Deferring it just turns it into attrition and surprise, which is the worst version for everyone in the room.

The teams that handle this well start early, say it straight, and put real skilling paths in front of the people who want them.

That’s the advice I’d give my own sales leadership. It’s the advice worth carrying into yours.

Bringing it back

Let’s return to where we started. Two open SDR seats. A question you couldn’t fully answer.

Here’s where I’ve landed. I’d genuinely like to hear if you see it differently.

The seat is becoming four seats. Most of them won’t carry the SDR title.

Nobody has to make all four hires this year. Most teams won’t. The realistic path is one role that folds two functions together, usually starting with quality, then growing into the rest as the volume justifies it.

We don’t need this solved by Q3.

What we can do now. Run the audit. Get honest about the five gates. Walk into the next planning conversation with a real read instead of a guess.

That’s a strong place for RevOps to be standing. We’ve done the harder version of this before. Every time the GTM stack shifted under us. We can do this one too.

If you run the data audit this month, tell me what you find.

I’ll be running mine.

---

If you want help mapping any of this to your own GTM motion, the 11x team is open to that conversation. Tell them the RevOps Impact Newsletter sent you.