Customer support is changing: From outsourcing to Results-as-a-Service

For a long time, people were the only ones to handle customer support. So, when businesses expanded and had to serve a larger customer base, their main instinct was to hire more staff. The logic is simple: when ticket volume rises, the queue backs up, and CSAT slips, you need larger teams to meet demand.

‍

However, this model didn’t work long-term. Support volume isn't stable, but the bills for the team you hired are. Not to mention that an increase in your team’s headcount does not always equal better ticket processing, quicker case closure, and higher CSAT. 

‍

HDI and MetricNet found that reducing a service desk's call abandonment rate from 8% to 4% required increasing headcount from 10 to 13 agents (a 30% staffing increase) yet produced "very little benefit in terms of higher customer satisfaction." This means most organizations are already operating in a zone where added headcount delivers negligible business improvement while costs climb linearly. It's one of many customer service statistics showing that more spending no longer automatically buys better service.

‍

Yet, buying more capacity doesn’t guarantee a desired outcome. Whether you build the team in-house or outsource it, you don’t know exactly what it will bring to your business (aside from more closed tickets).

Why buying capacity made sense for so long

For most of outsourcing's history, labor was almost the entire cost of service. Technology wasn’t so much involved in delivery, so the only lever to scale support was to add people, and the only way to cut costs was to hire them somewhere cheaper.

Where did the capacity-based pricing start?

The first deal of the kind happened between Kodak and IBM back in 1989. The companies signed a 10-year, $250 million contract that set the template still used today. Through the 1990s, a wave of partnership process offshoring swept business as companies chased Indian and Philippine wages, which were much lower than Western ones.

‍

Interesting to know: Stanford researchers put direct costs at $10,354 per Indian employee against $55,598 in the US in 2003, with billable labor around $3.10/hour offshore versus $21.50 onshore (see page 16). When the arbitrage runs that wide, and labor is basically the only deliverable, it makes sense to buy and price support by the agent or the hour.

‍

With no automation in the loop, the relationship between contact volume and staffing stayed close to linear. Adding agents was the only way to grow, because there was no layer underneath to absorb the extra load.

‍

That logic held as long as its two conditions did: labor as the dominant cost, and headcount as the only way to scale. Both are coming loose now, and that's why the capacity model is starting to break.

{{cta}}

Today, support is the last input-priced line item

Over the last few years, in addition to traditional human support, businesses have started using smart software tools to improve service delivery. And though at the start of this technological boom, buying a license or seat and paying whether or not you use it was popular, in 2026 it’s already fading away. The billing is now more feature-based, tied to what the tool actually delivers.

‍

This shift is most visible in the SaaS industry. Per-seat pricing is still common, but demand for it is falling fast. Specifically, Futurum's 1H 2026 survey of enterprise software buyers found that: 

‍

• Fewer than one in five still prefer classic per-user pricing
• 43% now want consumption-based billing
• 27% want outcome-based pricing. 

‍

It was also found that legacy BPOs stuck on seat-only pricing risk disqualification before they reach the shortlist.

‍

The change is driven by the broader AI innovation, which didn’t spare the support industry either. Popular AI agents running on top of the major help desks already bill by the outcome — resolved tickets. It's part of a broader shift toward the always-on resolution economy, where you pay for problems solved rather than seats filled:

‍

• Fin AI charges for a resolved conversation
• Zendesk AI — for a verified automated resolution
• Salesforce Agentforce — each AI action

The SaaS pricing trend in 2026

What AI did to the cost of a ticket

Behind the pricing shift is a plain change in unit cost. When a customer resolves their own question through AI self-service, it costs a business a median of $1.84, against $13.50 once an agent handles it, by Gartner's benchmark. Same customer, same question, yet about 7x the cost as soon as a person steps in.

‍

That changes how support scales. When the same AI can close routine tickets for a dollar or two, handling more of them no longer means hiring in proportion. But it’s not just the price that makes AI so alluring — it’s the way it drives business outcomes.

‍

Learn more about AI automation in customer service from a client case study.

Where the savings actually come from

Once AI handles a real share of the tickets, the cost of resolving each one drops sharply. That drop drives the savings. 

‍

According to one McKinsey report, AI-enabled customer service resulted in "a doubling to tripling of self-service channel use, a 40 to 50 percent reduction in service interactions, and a more than 20% reduction in cost-to-serve," with assisted-channel incidence ratios falling 20–30%. This example shows agent headcount isn't necessarily correlated with outcomes, as the same team serves far more customers (and at a lower cost) when following the human + AI support model.

‍

“You have to actually treat your AI agent as a part of your team. The way to go is just to treat your AI as a human, as you would train your customer support representative."
— Valentyna Dimova, VP of Customer Support, EverHelp, at the Tidio webinar

‍

Importantly, this cost collapse isn't driven primarily by headcount elimination, as is often associated with AI. In Gartner's October 2025 survey of 321 service leaders, only 20% had cut agent headcount because of AI, while 55% kept staffing steady while serving more customers. 

‍

AI isn't a free lever, though. Gartner expects rising generative AI costs to push per-resolution costs up over time. Even so, when you account for AI platform spend and the long tail of complex tickets that still require full-rate human handling, realistic net cost reduction across a whole support organization lands at 20–35% within 6–12 months.

Why "more agents” isn’t the right lever anymore

That's where paying per agent stops making sense for the business. Every ticket runs through a human, and every human ticket costs you $6–$15 — even the routine ones AI could easily close for a dollar or two. 

‍

You're paying human prices for work that can be done as well by AI. Worse, since providers usually bill by the agent, automating those tickets only shrinks their own invoice. So, you keep paying for headcount while the cheaper, and even more effective option goes unused.

‍

Outcome-based contracts fix that. Instead of buying hours, you're buying the outcome — the ticket resolved, the CSAT target hit. This setup is gaining more recognition, as 67% of organizations already use outcome-based outsourcing models, up from 45% two years earlier, per Deloitte's 2024 Global Outsourcing Survey.

{{cta}}

So, what is Results-as-a-Service?

Results-as-a-Service is a support model in which, instead of renting a set number of agents, you agree on the outcomes that matter for your business case, and the provider’s primary objective is to supply you with everything to hit them. 

‍

How they get there, with people, AI, or a mix of both, becomes their job to figure out and manage. Ideally, they do it through a structured CX optimization framework rather than trial and error. 

‍

That ownership is the real difference from plain outsourcing. Result-as-a-Service providers are the ones responsible for your business outcomes. So, when, for example, your goal is to save on cost-per-resolution, and the AI can close a ticket faster and cheaper, they have every reason to use it.

‍

Example: For one of our lean-ops clients, an outcome-focused, human + AI setup cut a 200-hour support wait down to a 4-minute first response, reaching 86% CSAT within a year of collaboration.

Capacity-for-hire vs Results-as-a-Service: comparison

‍

Choosing support this way changes what you look for in a provider. If you're comparing options, our vendor evaluation checklist helps you weigh them against the actual deliverables they bring to the table.

Ready to pay for outcomes instead of seats?

Adding agents stopped being the answer a while ago. So here's the question worth asking: when you pay your current support provider, what are you actually buying? 

‍

You can keep sizing a team to your busiest week and paying for it through the slow ones, or you can pay for what you really wanted all along — fully resolved tickets and loyal customers. That's the move from outsourcing to Results-as-a-Service, and it's already happened almost everywhere else in your software budget. Support is just the last line to catch up.

‍

At EverHelp, that's how we think about support: a hybrid of people and AI, measured on the outcomes you actually care about rather than the number of seats we fill. 

‍

Book a meeting, and we can discuss your perfect outcome-focused setup too.