AI Outbound is a thing right now. It’s not hard to understand why: the promise of agentic, infinitely scalable MQLs can be a heady exlir, indeed. So when AI SDR tools enter the outreach scrum of web leads, social engagement, cold email, and cold calling, the data can be hard to connect.
A 31% open rate. A 2.1% reply rate. Nine meetings booked. These numbers look like progress. They're not pipeline. Every one of them lives at Tier 1 of the Metric Hierarchy — activity metrics — and none of them connects to a sales conversation that's going anywhere.
If your monthly report doesn't include pipeline influence its a problem.
The failure rate is structural, not incidental
Between 50% and 70% of AI SDR tools churn within a year. Only 2% of companies implement them in a way that sticks. A SaaStr survey found that 7% of companies called outsourced SDR programs "highly successful." The remaining 67% reported they didn't work at all.
The most instructive case study is 11x.ai — $74 million in VC funding, losing 70–80% of its customers within months. The best-capitalized AI SDR company in the market couldn't keep ZoomInfo. Worth sitting with before you sign a contract.
The math on a typical campaign
Here's what a 2.1% reply rate produces: send 1,000 emails, get 21 replies. Strip out the unsubscribes, the out-of-office responses, the polite rejections. You're down to eight or ten real conversations. From those, you might book four meetings.
Industry benchmarks put meeting-to-qualified-opportunity conversion at 25–40% for cold-sourced outreach. Cold-sourced opportunities close at 10–25%. End to end, you're looking at roughly one closed deal per 500 to 1,000 delivered emails — on a good campaign.
At $3,500 a month, that math works out to $350–$875 per meeting depending on show rate. The cost per closed deal, if any close, is a number most vendors have never calculated for a client, because nobody asked.
Meetings booked is where the funnel looks best. That's why it leads the report.
Volume only works at scale — and most firms aren't at that scale
Jason Lemkin replaced his GTM team with 20 AI agents, generated 10 times the email volume, and described the results as "better than a mid-pack SDR, not better than top performers."
It works because Lemkin has the RevOps, the CRM, the sales team capacity to absorb 10x the volume.
For a 20-person staffing firm serving mid-market manufacturers in a defined region, the math is different. Your total addressable market might be 400 companies. A volume-first AI outbound campaign will move through a meaningful portion of that list in 90 days. The firms that don't reply aren't neutral — they've now associated your name with cold outreach they didn't want. Volume-first is built for markets measured in tens of thousands. Most professional services firms are not that market.
There's a version that works — it's just not what most vendors are selling
Signal-based campaigns — triggered by a leadership change, a funding announcement, a hiring spike — achieve reply rates of 5–18%, versus 1–3% for generic sequences.
The tool isn't the issue. The targeting infrastructure behind it is. Most vendors selling to $1M–$5M professional services firms aren't delivering the signal-based version. They're delivering templates with variable fields and calling it personalization.
What I'm reading:
AI SDR Tools Compared — The best breakdown of the category I've seen, including the 11x story. Worth reading before any vendor demo.
Cold Email Benchmark Report 2026 — Instantly's platform-wide data on reply rates. Useful because it's their own data arguing against the volume-first approach.
Best Outsourced SDR Services in 2026 — Where the SaaStr 7% stat comes from. The cost-per-meeting math alone justifies reading it.
Until next time,
Harry