SkillsSkill 4 of 5
Scale outbound without spending the signal
Judge outbound by opportunities per 1,000 emails and by complaints, and scale only the targeting that works.
Worked example
The AI SDR got more replies. Should you triple its volume?
Replies are the cheapest thing outbound produces. Follow them to meetings, pipeline and the sending reputation they cost.
The question
Ledgerwick, a fictional accounts-payable automation platform, ran an AI SDR alongside its human SDRs for one quarter, on separate sending domains. Sales development wants to triple the AI’s volume next quarter.
| AI SDR | Human SDRs | |
|---|---|---|
| Emails sent | 60000 | 6000 |
| Replies | 900 | 240 |
| Positive replies | 180 | 96 |
| Meetings held | 90 | 60 |
| Qualified opportunities | 18 | 21 |
| Spam complaints | 210 | 3 |
Invented data. Reply rate 1.5% for the AI SDR and 4.0% for human SDRs; meetings held per 1,000 emails 1.5 and 10.0.
| Emails sent | Qualified opportunities | Spam complaints | |
|---|---|---|---|
| Triggered: new finance leader or ERP change in the last 90 days | 8000 | 12 | 8 |
| List-based: no trigger | 52000 | 6 | 202 |
Invented data. The two rows add up to the AI SDR column above: 60,000 emails, 18 opportunities, 210 complaints.
“The AI SDR sent 60,000 emails and got 900 replies; human SDRs sent 6,000 and got 240. Would you triple the AI’s volume next quarter?”
Your turn
Say your answer out loud, or jot notes below. Both work. Or skip to the worked answer.
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Follow replies to pipeline
What would you do here? Open the move you’d make.
The AI got 3.75 times the replies, so it’s already winning.
Tempting, but it misses something. Replies aren’t pipeline. The AI produced 18 qualified opportunities from 60,000 emails; humans produced 21 from 6,000.
Follow replies to opportunities: 18 against 21, or 0.3 per 1,000 emails against 3.5.
The strong move. Opportunities are what sales needs; replies are the cheapest step.
Compare meetings: the AI held 90 to the humans’ 60, so scale it.
Tempting, but it misses something. AI meetings turned into opportunities 18 times out of 90, against 21 out of 60 for humans. More meetings, less pipeline.
See how a strong candidate says it
“The AI got 3.75 times the replies, but 18 qualified opportunities to the humans’ 21. Per 1,000 emails, that’s 0.3 opportunities against 3.5.”
Why it works: Replies are the cheapest step; opportunities are what sales needs.
Step 2: Find where the AI actually works
What would you do here? Open the move you’d make.
Judge the AI SDR as one channel with one result.
Tempting, but it misses something. Its 18 opportunities split sharply: 12 from 8,000 triggered emails and 6 from 52,000 list-based ones.
Conclude the AI can’t sell and shut it down.
Tempting, but it misses something. Triggered AI sends produced 1.5 opportunities per 1,000 emails. The problem is the list, not the channel.
Split by targeting: triggered sends were 8,000 emails but 12 of the 18 opportunities, 1.5 per 1,000.
The strong move. It separates the channel from the targeting, so the AI isn’t blamed for a list problem.
See how a strong candidate says it
“Triggered sends were 8,000 of 60,000 emails but 12 of the 18 opportunities: 1.5 per 1,000, against about 0.12 for list-based sends.”
Why it works: It separates the channel from the targeting, so the AI isn’t blamed for a list problem.
Step 3: Count what volume costs
What would you do here? Open the move you’d make.
Check complaints: 210, 0.35% of AI email against 0.05% for humans, and 202 came from list-based sends.
The strong move. Deliverability is shared, and tripling list-based volume would spend it for everyone.
Ignore complaints; they’re a small share of 60,000 emails.
Tempting, but it misses something. 0.35% is seven times the human rate, and the sending reputation carries every SDR’s email.
Buy more sending domains so complaints don’t matter.
Tempting, but it misses something. New domains spread the damage without fixing the cause: 202 of 210 complaints came from list-based sends.
See how a strong candidate says it
“The AI drew 210 spam complaints, 0.35% of its email, against 0.05% for humans, and 202 came from list-based sends. Tripling that puts our sending reputation at risk.”
Why it works: Deliverability is shared, and one channel can spend it for everyone.
Step 4: Make a scoped call
What would you do here? Open the move you’d make.
Triple volume as planned and review results next quarter.
Tempting, but it misses something. At the list-based rate, most extra email would add complaints, not opportunities: 52,000 list-based sends produced 6.
Keep the AI on triggered accounts, stop list-based sends, and grow only while opportunities per 1,000 and complaints hold.
The strong move. It reaches a decision with a guardrail and names what would let volume grow.
Hand all outbound back to human SDRs.
Tempting, but it misses something. Triggered AI sends produced 12 opportunities from 8,000 emails, a channel worth keeping where the targeting is right.
See how a strong candidate says it
“I wouldn’t triple volume. I’d keep the AI on triggered accounts, stop list-based sends, review each week what it got wrong and what needed a person, and grow volume only while opportunities per 1,000 emails hold and complaints stay low.”
Why it works: It reaches a decision with a guardrail and names what would let volume grow.
The answer, in short
- Decision
- Don’t triple AI volume. Keep the AI SDR on triggered accounts, stop list-based sends, and grow only while opportunities per 1,000 emails and complaint rates hold.
- Because
- List-based AI email was 52,000 of 60,000 sends but produced only 6 of 18 opportunities and 202 of 210 spam complaints.
- What would change it
- List-based sends producing opportunities near the triggered rate with few complaints, or a larger pool of triggered accounts to grow into.
Common traps
- Celebrating reply volume.
- Comparing channels without a per-email unit.
- Ignoring the sending reputation that volume spends.
- Blaming the AI for a targeting problem.
Check your answer
- Did I follow replies to meetings and opportunities?
- Did I compare per 1,000 emails?
- Did I separate targeting from the channel?
- Did I account for deliverability before scaling?
Want feedback on how you’d answer? Practice “Scale outbound without spending the signal” with an AI interviewer on Socratify.
Last reviewed Nov 9, 2026