These programs almost never end in an argument. They end in a calendar invite that nobody reschedules.
Month one is setup. Month two produces a few meetings and everybody relaxes. Month three is flat, and the agency has an explanation that sounds reasonable, because it usually is reasonable. Month four is quieter. By month five somebody in the finance meeting asks what exactly this line item is producing, and nobody in the room has a good answer.
Here is the short version of why. Five things kill these programs, and three of them were already true on the day the contract was signed. Only one of the five is really about the agency's skill at sending email.
I have been the second agency on an account more than once. The first call contains the same sentence nearly every time: "we have tried this before and it did not work."
The number everyone quotes tells you nothing useful
If you have researched this at all, you have met the statistic. A big round percentage of outsourced SDR programs did not work, the exact figure depending on whose blog you landed on. I went after the version that travels furthest, and every trail ended at one blog quoting another, so I am not adding another link to that chain.
Keep the shape and throw the decimal away. Plenty of these programs do fail, more than the category likes to admit out loud. What no percentage will tell you is which ones, or why. The failure is five specific mechanisms, and once you can name them you can watch for them in your own program.
Nor does any of it argue for hiring instead. The failures below hit a first in-house hire almost as hard.
Nobody had proven the offer before somebody sold volume against it
This is the biggest one and the least discussed, because admitting it is uncomfortable for both sides.
Outbound scales whatever you already have. If people are buying your thing at a normal rate through referrals and inbound, an outbound program pours more of the same people into the top. If nobody has bought it yet, or the offer changes depending on who is asking, outbound tells you that faster and more expensively than any other channel.
So the honest disqualifier is this: a company with no proven offer should not be buying an SDR program at all. My own filter, applied before any contract, rules out companies trying to build the business itself from zero, because no amount of sending fixes an unsold product. Agencies that skip that filter get paid for six months and then get blamed, and both of those things are fair.
If you are unsure which side of the line you are on, the question is answered in who outbound actually works for rather than in a proposal.
Meetings booked was the only number in the report
Pick a metric and you get it. Pick meetings booked and you get meetings booked, including the ones that were never going to buy.
A program measured purely on volume of calendar entries drifts toward whoever will say yes to a call, which is not the same population as whoever will say yes to an invoice. Then month four arrives with a full calendar and no pipeline, and the client concludes that outbound does not work for their market. It might. They just paid for the wrong output.
There is a second version of this, which is worse. Companies with a real sales problem often buy an SDR program to avoid looking at it. Adding a hundred conversations to a process that already loses the ones it has will produce a bigger, more expensive version of the same result.
The fix lives in the contract. Define what a meeting has to be before it counts, then report held and advanced alongside booked. Those three numbers together are a diagnosis. Booked on its own is a receipt.
Nobody on your side was free to take the calls
I turn down work over this, and I would rather say it plainly than be polite about it.
An outbound program produces conversations at a rate your calendar has to absorb. If the person who takes the calls is also the person who delivers the work, then a busy month means the calls get pushed, the replies sit for four days, and the campaign gets paused "just until things calm down".
A paused campaign in month three is a dead campaign in month five. Sending has memory; you do not get to switch it back on where you left it.
The other half of this is the inbox. Replies need answering the same day, by someone who can hold a conversation, and on most stalled programs that job has quietly become nobody's job.
Before you sign anything, decide who takes the calls and who answers the replies, by name. If both answers are the founder and the founder is already at capacity, the program will fail and it will not be the agency's fault. First 90 days of outbound sets out the client-side work month by month.
The person who learned your business left
Outbound gets better the longer someone does it for you. They learn which title actually replies, which industry never does, which objection means no and which one means not yet. That knowledge lives in a person before it lives in a document.
Then the person leaves. Sales development runs 35 to 40% turnover a year, on an average stint of 14 to 16 months, by the reckoning of one outsourcing provider writing about its own market. Argue with the decimals if you want. Nobody argues with the direction. Whoever is on your account today is unlikely to be there in eighteen months, whether they sit in your office or someone else's.
The difference is what survives them. Ask, in month one, where the ICP definition, the messaging, the list criteria and the reply history are written down, and who owns that file when the contract ends. A program that lives in one person's head is a program with a resignation letter in its future.
The timeline came from a proposal instead of from the work
Buying the function is quicker to start than building it. One provider's own breakdown puts an in-house hire at 30 to 90 days to recruit and another 60 to 120 days to ramp, against an outsourced campaign that can be live in weeks. That is a vendor describing its own advantage, so read it as a shape rather than a promise.
The trouble is what that shape does to expectations. A month becomes "we should see revenue by the second one", and a first meeting in week six gets treated as a disappointment. Nothing broke. The clock was wrong before it started.
Why founders end up on their third agency
Put those five together and you get the pattern I meet most often: a founder who has hired three agencies and is now sure the whole category is a scam.
Almost always they were overpromised and under-delivered to, in that order. The promise was a number of meetings by a date. The delivery was a real program running against an unproven offer, measured on the wrong metric, feeding a calendar nobody had cleared, staffed by somebody who has since moved on.
The uncomfortable half of that sentence is that two of those five failures belong to the buyer. Agency three often works, not because it is better, but because by then the founder has an offer that sells, a person on replies and a definition of a qualified meeting. The thing that changed was on their side of the table.
The checks that catch it before month four
These are for a program that is already running.
Week two: open the list and read fifty rows. If you cannot tell why those companies are on it, nothing downstream saves you.
Week four: read twenty sent messages and twenty replies, in full. This is the single fastest way to know whether a human is running your account.
Week six: count held meetings, not booked ones, and ask what happened to the gap.
Week eight: ask what the agency has learned about your market that they did not know in week one. A good answer is specific and slightly annoying. Silence here is the strongest early signal there is.
Month three: check who owns the data, the domains and the list. That answer only becomes urgent on the day the contract ends, which is exactly when nobody wants to open the question.
Before you hire anyone there is a longer list of questions worth asking, and most of them are about exactly these five failures.
Programs that survive month four tend to share one unglamorous feature. Somebody on the client side has been reading the replies since week one, and they can tell you what the market said.