AI is lifting customer service performance - without mass layoffs, says Gartner
AI isn't wiping out support jobs. It's making them faster and sharper. A new Gartner survey shows most teams are using AI to handle more volume with the same headcount - and in many cases, to create new roles.
The study, run in October 2025 with 321 customer service and support leaders, found that only a minority cut staff after adopting AI. Most kept teams steady while productivity went up.
"Customer service and support leaders should avoid framing AI initiatives solely around headcount reduction," said Melissa Fletcher, Senior Principal, Research in the Gartner Customer Service & Support practice. "Instead, focus on incremental transformation and workforce augmentation. Leaders should plan for new roles, leverage central resources, and communicate transparently about AI's impact to manage expectations effectively."
Key numbers you can use
- 20% of organizations reduced headcount due to AI.
- 55% held staffing flat while handling more customer volume.
- 42% are hiring new roles to support AI (AI strategists, conversational AI designers, automation analysts).
- By 2027, half of the organizations planning major AI-driven cuts are expected to abandon those plans as fully "agentless" service proves tough to pull off.
Source: Gartner Customer Service & Support
What this means for support leaders
- Position AI as assistive tech, not a headcount slasher. Make that clear early and often.
- Stand up new specialties: conversation design, AI operations, prompt standards, data quality, and automation governance.
- Build a central enablement hub for models, prompts, knowledge, and QA so teams don't rebuild the same thing five times.
- Set a change playbook: training, comms, feedback loops, and clear career paths tied to AI skills.
Where AI drives real value right now
- Intake and routing: summarize customer intent, tag issues, and route to the right queue instantly.
- Agent assist: suggest replies, surface knowledge, and fill forms while the agent stays in the conversation.
- Knowledge upkeep: draft articles from resolved cases and flag outdated content automatically.
- Case wrap-up: generate summaries, next steps, and dispositions to reduce after-call work.
- Quality and compliance: score interactions against policy and highlight coaching moments.
- Self-service: improve deflection with clearer FAQs and safer, supervised chat automation.
Metrics to track (so you can prove it works)
- Average Handle Time (AHT) and After-Call Work (ACW)
- First Contact Resolution (FCR) and Escalation Rate
- CSAT and Customer Effort Score (CES)
- Deflection/containment for self-service flows
- Agent utilization, time to proficiency, and QA scores
- Knowledge freshness and article adoption
Why "agentless" is harder than it looks
Customer intent is messy, policies change, and edge cases matter. Fully automated service struggles with ambiguity and risk. That's why Gartner expects many firms to rethink big layoff plans - supervised AI with skilled agents in the loop is proving more reliable.
Implications for outsourcing and shared services
- Differentiate with AI-enabled QA, knowledge workflows, and agent co-pilots rather than pure labor arbitrage.
- Offer outcome-based pricing tied to AHT, CSAT, and deflection improvements.
- Stand up shared AI operations and governance for clients who don't have the talent in-house.
- Upskill agents into high-skill work: complex resolution, retention, and consultative support.
A practical 90-day plan
- Pick two high-volume workflows (e.g., password resets, refunds). Document the happy path and top edge cases.
- Pilot agent assist first. Measure AHT, CSAT, and QA before/after. Share wins weekly.
- Upgrade knowledge: clean source content, add structured fields, and set a refresh cadence.
- Create lightweight governance: prompt standards, review checkpoints, and policy guardrails.
- Communicate career paths: new allowances for AI skills, badges, and stretch projects.
Level up your team
If you're building skills for agent assist, conversation design, or automation ops, explore targeted learning paths for support roles here: AI courses by job.
Your membership also unlocks:
