Is an AI SDR worth it?
Short answer
As a replacement for a person, usually not on the public record. It shows high reported churn, a named pilot that lost to human SDRs, and Gartner expecting fewer than 40% of sellers to report a productivity gain from agents. As a tool that researches and drafts for a person who owns the conversation, it can be.
The short version
It depends on which job you buy it for. Bought to replace a person, the public record is poor. Bought to do research and first drafts for a person who still owns the conversation, it has a much better case.
What the public record shows
There is not much of it, and most of it comes from the vendors. Three pieces stand out because someone put a name to them.
A named pilot that lost. ZoomInfo ran 11x for one month. Its verdict:
“11x’s product performed significantly worse than our SDR employees.”
ZoomInfo did not go ahead.
Churn that the vendor disputes. An 11x employee told TechCrunch the company was losing 70-80% of the customers that came in. 11x said its worst churn came from early cohorts in late 2023, and put its retention rate at 79%. We cannot settle that dispute, and we are not trying to.
A vendor’s own admission. The co-founder of Artisan, another AI SDR vendor, said first-generation AI SDRs get a low response rate and have relatively high churn.
Where it can pay
The tasks are a different story from the role. Salesforce reports that sellers expect agents to cut prospect research time by 34% and email drafting by 36%. Salesforce sells agents, and these are expectations rather than results. They still point at the same two tasks our own corpus flags: pre-call research was the most repeated automatable complaint in 1,157 practitioner posts.
Gartner adds a condition that matters more than the tool. Teams that fix their data, their automation and their user experience first are, on its forecast, five times more likely to see a return on AI than teams that buy a quick fix. An AI SDR on top of a messy CRM, in our reading, sends more email from the same bad list.
What to test before you buy
Split the list. Give the AI SDR half of one list and a person the other half. Compare meetings held, not emails sent or replies counted.
Ask for retention by cohort. A single retention figure hides the early cohorts. Ask how many customers from a year ago still pay.
Read what it sends. Pull twenty of its emails before they go out and read them as the prospect would. Our related tool grades a cold email line by line, which is a quick way to do that check.
Keep a person on the reply. The evidence for AI doing research and drafts is far stronger than the evidence for AI handling the conversation that follows.
What this is based on
-
ZoomInfo ran a one-month pilot of 11x, an AI SDR product, and said it 'performed significantly worse than our SDR employees'. ZoomInfo did not go ahead.
-
An 11x employee told TechCrunch: 'We were losing 70-80% of customers that came through the door.' 11x disputed this. It said its highest churn came from initial cohorts in late 2023, and that its retention rate was 79%.
TechCrunch, 24 March 2025. An anonymous source, disputed by the vendor. →
-
The co-founder of Artisan, another AI SDR vendor, said first-generation AI SDRs 'get a pretty low response rate' and have 'relatively high churn'. He also said: 'We had extremely bad hallucinations when we first launched.'
TechCrunch, 9 April 2025. A vendor describing its own category. →
-
By 2028, fewer than 40% of sellers will say AI agents have improved productivity. CSOs who overhaul data, automation and user experience will be five times more likely to gain ROI from AI than those who choose quick fixes.
-
Once fully in place, sellers expect agents to cut prospect research time by 34% and email drafting by 36%. These are expectations, not measured results.
-
Pre-call research is the most repeated automatable complaint in our corpus, across five practitioner subreddits.
What we don't know
- · We found no independent study that runs an AI SDR and a human SDR on the same list and compares meetings held. Every outcome figure we found is a vendor claim, a pilot verdict or an anonymous source.
- · The 11x churn figures are disputed, and the worst of them come from early cohorts. Products in this category change fast, so the record today may be better or worse.
- · We cannot give you a break-even point. Vendor pricing is rarely public and varies by contract, and the value of a meeting differs by deal size.
- · This page reports no measurement of our own of an AI SDR's results.
On the search results for this question: We measured the shorter query 'ai sdr worth it': forum threads rank and no vendor owns the answer. 'ai sdr' was one of only two queries in our 97-SERP sample with no AI Overview.
We have no affiliate relationship with any tool named on this page, and nothing here is sponsored.