How OneSource Cloud built a $6M proposal pipeline with a two-person team

OneSource Cloud sells GPU compute infrastructure and private AI deployment to enterprise buyers. The deals are large, the buying committees are wide, and the cycle runs two to three quarters at minimum. It is exactly the kind of motion where pipeline gets built slowly or not at all, because the people best equipped to build it are the same people closing.

Mark McNeill is Head of Global Sales. His team is small and every rep is full-cycle. That is the constraint the whole story turns on.

The problem was never effort. It was arithmetic.

Before Aurium, prospecting was outsourced. OneSource Cloud had contracted the same outbound lead generation agency for five years, running multiple campaigns across multiple service lines. It was, as McNeill puts it, “all people powered.”

The agency’s process looked superficially like Aurium’s. Tell us about your campaign, give us collateral, let us learn your products. What came back was one suggested connection message and one suggested outbound email. Then the handoff.

The handoff was the whole problem. If the agency found someone who said this sounds interesting, McNeill’s description of what happened next is blunt: now you take it over. A connection request is not a qualified lead, and one email cannot qualify anyone. Everything between “sounds interesting” and a booked meeting landed on McNeill’s team, which is to say on his closers.

“I have a very small team. I really wanted them focused on real good opportunities they could close quickly, not doing SDR stuff and BDR stuff. We didn’t have time for that.”

Two other things bothered him. The email channel was not working at all. And the agency ran outreach from robo profiles, synthetic accounts unconnected to any real person.

His concern was not hypothetical: synthetic accounts get found eventually, and when they are shut down the account history goes with them.

The contract was also expensive: by McNeill’s estimate, five to six times the monthly cost of the equivalent Aurium seats.

What changed

McNeill had reason to be skeptical. He came to OneSource Cloud from the master agent technology distribution world, where generating leads for a supplier portfolio was the business. He had seen the category up close.

He had tried, by his own count, every people-powered and supposedly AI-driven lead generation service on the market, running as many as 11 at once, and rated none of them successful.

His specific objection was that most vendors were AI-assisted, not AI-driven. A model would draft a connection request, and that was the extent of it. What convinced him was that Aurium ran the whole loop.

“I will swear up and down on a stack of Bibles to any reference customer you bring me that this is truly AI driven at every single point. It helps you create your campaign. It creates the messaging. It does all the outreach. It identifies the right ICPs to go after.”

Three parts of how the program runs are worth separating out.

1) Conversations that hold, at a volume people cannot

The reason a small team can run wide outbound is that the conversations are not on the team. Across four active campaigns, McNeill is party to thousands of ongoing conversations at once.

“I am now conversing with over 4,000 people. Actively conversing with them right now, while I’m talking to you. There’s no way your team, no matter how big it is, could manage those kinds of back and forth conversations.”

The threads are not shallow. The typical warm-up takes four or five exchanges. Some take far longer, and Aurium carries them the whole way.

One ran to 75 exchanges before the prospect agreed to a meeting, handled end to end without a rep touching it.

Because outreach runs from real profiles rather than synthetic ones, the reputational exposure McNeill worried about does not apply. What he did not expect was that nobody would notice they were talking to software.

“My whole team and myself, we’ve had about 40,000 contacts and conversations going back and forth. Out of those 40,000 interactions, not one has come back to any of us and said, I actually think I’m dealing with AI. Not once did that ever happen.”

2) Qualification that arrives finished

The complaint about the old vendor was that leads showed up raw. McNeill’s read now is that the ICPs reaching his team are better qualified than what the team used to produce itself before a meeting.

Collateral goes out on its own, without a rep being asked. And the targeting compounds, because the system optimizes against meetings held rather than connections accepted, learning from the leads that make it all the way through rather than the ones that merely connected.

3) The feature nobody asked for

McNeill’s own takeaway was not the conversational engine or the acceptance rates. It was calendars.

Most wasted time in a GTM team, he argues, goes on pinning down a meeting slot that works for an entire buying committee, then rescheduling it when it slips.

In practice, prospects rarely take a scheduling link. They offer their own Google Meet or Teams, or name a time and ask for an invite. Aurium reads both calendars, coordinates, and sends the invite, then handles the reschedules and the added attendees.

“That is grunt work that no executive wants to do. Taking that off their plate has been an enormous time saver that we didn’t expect, and now it’s almost everyone’s favorite feature.”

Results

Across two seats and four active campaigns, Aurium sent 7,396 connection requests and booked more than 100 calls in six months. Connection acceptance runs at 22.6%, against 9% before Aurium.

More than half of those meetings advanced past a first conversation into gathered requirements and a requested proposal. The resulting proposal pipeline is over $6 million, including two large private-AI opportunities, one with a national government and one with a university.

The reps got their hours back. Where prospecting used to take two or more hours a day, it now takes none.

Asked what he tells GTM leaders who are on the fence, McNeill goes to the same place every time.

“Your team can never hold thousands of conversations simultaneously. You couldn’t hire enough people to do that.”