AI SDRs vs intent-based outbound: which actually books meetings?
AI SDR companies raised $400M+ in 2025 on the pitch of 'replace your SDR team.' We measured the meetings booked. Here's what the numbers show.
2025 was the year AI SDR companies — 11x, Artisan, Regie, AISDR, and a long tail of others — raised hundreds of millions of dollars promising to replace your SDR org.
2026 is the year we have enough data to evaluate them honestly.
We pulled benchmarks across 40 GTM teams running AI SDRs, intent-based outbound, or both. Here's what we found.
What an AI SDR actually does
An AI SDR is, in practice, a high-throughput personalization-and-sending engine. It takes a list, enriches each contact, drafts a "personalized" first touch using an LLM, and sends thousands of these a week. Some include a basic LinkedIn touch; most are email-first.
It is fast and it is cheap per touch. Both true.
What intent-based outbound does differently
Intent-based outbound (Saava-style) sends fewer touches, but only after a signal indicates the recipient is in a near-term buying window. The signal is the precondition. The volume is intentionally low.
It is slow per dollar of pipeline at the start and faster per dollar of pipeline at scale. Less obviously true, but true.
The benchmark
For the same target ICP, same SDR pay rates, same sequencing tools, over 90 days:
| AI SDR | Intent-based | Notes | |
|---|---|---|---|
| Touches/week | 4,200 | 180 | 23x volume |
| Reply rate | 0.9% | 11% | 12x quality |
| Replies/week | 38 | 20 | AI wins by 2x |
| Meetings booked/week | 4 | 11 | Intent wins by 2.75x |
| Cost per touch | $0.18 | $4.50 | AI wins by 25x |
| Cost per meeting | $189 | $73 | Intent wins by 2.6x |
| Close rate from meeting | 8% | 24% | Intent wins by 3x |
| Cost per closed-won | $2,360 | $305 | Intent wins by 7.7x |
The reply-rate gap is structural. AI-generated cold touches are getting more sophisticated, but inboxes are responding by getting more skeptical faster than the touches are getting better. The reply rate is trending down for AI SDR, not up.
The close-rate gap is even more structural. AI SDR meetings are with prospects who responded to a clever email. Intent-based meetings are with prospects who were already researching the space. Same demo. Wildly different conversion.
When AI SDRs do work
There are three cases:
- Pure top-of-funnel volume — you sell into massive markets (e.g., SMB, prosumer) where the long tail of small accounts is too big to qualify and your unit economics tolerate a $2K CAC.
- You're already at signal saturation — your intent stack has surfaced everything it can, and AI SDRs work the next 30% of the addressable market that doesn't leave intent signals.
- You have a strong inbound flywheel and just need to reactivate dormant accounts in bulk.
When intent-based always wins
- Enterprise / mid-market deals (ACV > $25K)
- Markets with concentrated decision-makers (PE, F500 ops leaders, founders)
- Categories where the buyer leaves a public research trail (most B2B SaaS, all GTM tooling)
- Teams below $5M ARR — every meeting matters too much to fill the calendar with bad-fit prospects
The honest middle
The teams winning in 2026 don't pick one. They run intent-based for the top 5% of their ICP (where the unit economics demand it) and AI SDR or volume outbound for the long tail (where the cost-per-touch math works).
The mistake is putting your AE-quality ICP through an AI SDR. You burn relationships you can't rebuild, and you waste signal you can't recover.
How to evaluate AI SDR vendors honestly
Three questions to ask:
- What's your customer's average reply rate after 90 days? (Vendors will quote launch numbers — those decay.)
- What % of replies become qualified meetings? (Reply rate without quality is vanity.)
- What does your churn look like at 6 months? (If it's >50%, the product is not generating durable pipeline.)
We've seen AI SDR vendors quietly drop reply rates from 3% in pilot to 0.6% in production. The drop is the truth.