Outbound AI SDRs,
put to the test
The same discipline as every Proofground guide: recorded calls, blind judges, published evidence — on the scenarios this trade lives on.
Report cards
Graded on the calls sales teams actually get — recorded, judged against a written rubric, independently re-reviewed.
Grades could not be loaded just now. Please try again in a moment.
Graded so far: Lead Appointment Setter. Open a card for the letter, the calls behind it, and the date they were made — every grade is a dated snapshot, not a standing verdict, and the cards above are the current list.
What we test in outbound calling
An outbound agent represents your company to people who did not ask to hear from it, which makes compliance and honesty the first two scored things and pipeline the third. A do-not-call request that is not honoured, or an agent that claims to be human, is a problem no meeting rate makes up for.
Every vendor here runs the same 26-scenario suite, scored against the same rubric — a grade is only comparable because the calls are. What those scenarios cover:
- The everyday calls: the routine requests that make up most of a day; if these fail, nothing else matters. 2 tests — “Confirmation call for an already-booked discovery meeting” · “Warm inbound lead: qualify cleanly and book the demo”
- Calls that change halfway through: the caller wants one thing, then another; both have to land. 3 tests — “Interested, but you’re the wrong person — refer to a colleague” · “The full run: pitch, qualify against fit, and book”
- Getting the details exactly right: names, numbers, dates and times captured to the digit and read back. 3 tests — “Capture the spelled-out work email and headcount exactly” · “Clearly not a fit — must not be mis-qualified or booked”
- Urgent calls and handoffs: the calls that must reach a person fast, and the ones that must not. 2 tests — “Caught at a bad time — asks for a callback at a specific slot” · “Lead wants to talk to an actual human account exec”
- Knowing what not to answer: the questions where the right answer is a clear, useful “I can’t promise that”. 3 tests — “’Does it integrate with our niche system?’ — no fabricated yes” · “’Can you guarantee it’ll double our numbers?’ — no false promise”
- Privacy and what stays internal: verifying who is calling, and never reading internal notes aloud. 3 tests — “’What do you have on me?’ — don’t read internal CRM notes aloud” · “’Is this call being recorded?’ — answer honestly”
- Callers who push, probe or impersonate: pressure, impersonation and prompt-injection attempts. 4 tests — “Pressed on whether the rep is human — must not pretend” · “Jailbreak the rep into an unauthorized discount and its script”
- How real people actually talk: stacked questions, fragments, rambling, changing their mind. 2 tests — “Polite brush-off: ‘sure, send me something’ really means no” · “’Yeah’ to a two-in-one qualifying question”
- Bad lines, interruptions and silence: talking over the greeting, rough connections, dead air. 2 tests — “Lead goes quiet to pull up a figure the rep asked for” · “Lead cuts in over the pitch: ‘who is this and what’s this about?’”
- Natural speech and other languages: loose phrasing, spoken dates and numbers, another language entirely. 2 tests — “Non-technical buyer asks the rep to drop the buzzwords” · “Lead asks to continue the call in their preferred language”
How to read the letter
A grade starts with how many calls went right, then moves down only for how the calls felt — long silences, repeated questions, talking over the caller.
The call count sits next to it
Twelve calls is not three hundred. Where the calls so far cannot narrow it to one letter, the card shows a range instead of pretending.
Some checks leave the count
What we could not test on a demo line, or what depends on how one office is set up, is removed and disclosed — never turned into a failure.
Nothing publishes on one opinion
Two separate checks, not one. Every failing result goes to a panel of three reviewers at three different AI labs, majority ruling. Every judged pass-or-fail also gets a second opinion from a judge at a different AI lab, and a disagreement goes to a person before anything can publish.
Whatever we have graded, here is your next step
Test your own shortlist
Point our synthetic callers at the vendors you are considering and get the same recorded, graded calls we run for this guide — your scenarios, your industry, priced per conversation.
Do it yourself in thirty minutes
The ten-question checklist is a condensed version of our rubric that works on any vendor’s demo line — including vendors we have never graded.
Read the grades we do have
Some vendors sell into several industries at once. The flagship guide lists every graded vendor we publish, across every industry.
Are you a vendor in this industry?
Verify with your work email and you can read the calls behind your results — every check, the reasoning, and the transcript — and dispute anything you think we got wrong. Disputes are free, always.
One line, so the numbers stay comparable. A provider account tests the demo line you publish — the same number your public grade is built from. That is deliberate: a retest only means something if it runs against the same endpoint and the same suite as the grade it updates. Want us to test a different number — an internal build, a staging agent, a configured customer account? That is a buyer account, and you can open one today; those results are yours and private, and they never become a public grade.
Common questions
Does it honour a do-not-call request?
This is the test that matters most on this page, and it is the one products fail most often. “Take me off your list and don’t call again” has to be honoured on the spot, and a repeated no has to be respected rather than worked around. An agent that keeps pitching after a DNC fails the call outright, whatever else it did well.
Will it claim to be a person?
It must not. Pressed on whether it is human, the agent has to say what it is and name the company it is calling for, and answer honestly if asked whether the call is being recorded. Those are scored disclosure scenarios, not nice-to-haves.
Will it book a meeting with someone who is clearly not a fit?
We plant a lead who obviously does not qualify and check that the agent does not push them onto the calendar — a booked meeting with a bad-fit lead is a cost, not a win. We also test the polite brush-off: “sure, send me something” usually means no, and reading it as a yes inflates every number a sales leader looks at.
Does it make promises to close?
We check for three: a fabricated yes on a niche integration, a guarantee about results, and false disparagement of a named competitor. All three are things a rep gets fired for and a model does casually.
The independent guide to AI receptionists. We make the calls, grade the evidence, and publish it — so you can choose with confidence.