3 Reasons Your AI Deals Aren't Closing

Why isn't your AI offering generating revenue?

Three structural problems sit between the strategy and the revenue.

Many BPO leaders have looked at the trajectory of this industry, where AI is taking transaction volumes and where enterprise buyers are heading, and they've made the call: we're building a Human+Technology offering.

That's not a small thing. Not everyone has gotten there yet. Some operators are still running the quiet calculation: "If I help clients automate, I'm selling away my own seats." That fear is rational in the short term, even if it's a losing long game, the same false sense of security that comes from volume staying steady right up until it doesn't.

But this piece isn't for them. It's for the leaders who've already decided, who are actively trying to sell AI-enabled offerings but aren't seeing the pipeline results they expected.

In working with BPOs at all stages, we've uncovered three structural problems behind the gap between strategy and revenue, and they're hard to see from the inside.

Why is your comp plan working against you?

It still pays for seats.

Most BPOs use a standard commission structure: a percentage of contract revenue that steps down each year. It's an industry norm, and it made sense when every deal was an outsourcing contract.

Run it against a tech-enabled deal and see what it produces. A traditional seat-based outsourcing contract is large: many seats, billed by the hour, so the commission on it is substantial. Your new AI-enabled offering, priced per agent per month, is worth a fraction of that in annual value. Same commission structure, and the same rep earns far less.

You've engineered a plan that pays your salespeople far less to sell your most strategic offering. You will get the behavior you incentivize, and right now you're rewarding selling seats. If the comp structure hasn't changed, the sales behavior won't either.

Fixing this isn't complicated in concept: you need a different compensation structure for tech-enabled deals, with higher commission rates, accelerators for new offering categories, or SPIFs tied to AI deal milestones. The specifics will vary. But leaving the old structure untouched while expecting a different sales outcome is the definition of wishful thinking.

Why doesn't your sales process fit an AI sale?

It was never built for a multi-stakeholder purchase.

Here's something that's true of most BPOs, and there's no judgment in saying it: your salespeople don't have a formal sales process. They never really needed one.

BPO sales has traditionally been a relationship-driven, instinct-guided motion. Your rep knows the VP of Operations. They get the call when the prospect is considering changing providers. That skill set built real businesses. But it breaks down completely when you try to sell AI.

Selling outsourcing services typically involves one department, a limited number of stakeholders, and a buyer who already understands what they're buying. Selling AI is a fundamentally different animal: a high-value, high-risk transaction with a longer cycle and multiple decision-makers. Suddenly you're navigating IT, Finance, Operations and sometimes Legal, all at once, and each stakeholder has different criteria. A CTO needs security and integration answers. A CFO wants a detailed ROI analysis. An operations leader needs to understand workflow impact.

Your rep needs to identify the economic buyer, cultivate a champion who'll advocate when they're not in the room, and know when to bring in technical resources, and critically, when not to. Showing a demo before requirements are fully understood is one of the fastest ways to kill a deal that has real potential.

The fix is a playbook. Not a binder no one reads, but a clear, practical guide that defines the stages of your AI sales process, the stakeholders your reps should map at each account, and the triggers that determine when to accelerate or slow down. Without that foundation, you're sending experienced salespeople into a fundamentally new kind of fight without a game plan.

Why hasn't your team been trained to sell it?

Selling BPO services and selling technology are different skill sets.

Assuming your salespeople will naturally make that transition because they're talented and experienced is a well-intentioned but costly assumption.

Consultative selling, which AI sales requires, means deeply understanding a prospect's business context, asking the right diagnostic questions, and positioning yourself as a trusted advisor, not a vendor pitching features. It's the discovery discipline that keeps a quote from becoming a spec-sheet comparison. That's a learned discipline, and this is where the industry falls short: the standard "enablement" for a new AI offering in BPO is a vendor briefing or two and some marketing collateral.

That's not training. That's exposure.

Real enablement means scenario-based practice, not slide decks. It means ongoing reinforcement, not a single session. And it means reps who can handle the specific objections that come up in AI conversations, about job displacement, data security and integration complexity, with confidence and credibility.

How do the three problems compound?

A rep who isn't paid to care, has no process and wasn't trained doesn't have a pipeline problem. They have a structural one.

Compensation, process and training are the three most common blockers on an AI deal, but they rarely show up alone. It's a problem that compounds further once you bring account managers into the picture, and structural problems don't get fixed by telling people to work harder or believe in the vision more.