AI Services Companies

AI Customer Service Statistics 2026

Published November 16, 2026

Our analysts compiled figures on AI in customer service from Gartner's surveys of customers and service leaders, Salesforce's State of Service report, Zendesk's CX Trends report, Intercom's published resolution rates for its Fin AI agent, Klarna's disclosures about its AI assistant, and the National Bureau of Economic Research study of 5,179 support agents, to show how far AI has gone in customer support as of Q4 2026.

Key Findings

  • Service teams estimate that AI handles 30% of customer service cases today and project 50% by 2027, according to a Salesforce survey of 6,500 service professionals.6
  • Intercom says its Fin AI agent resolves an average of 76% of conversations across more than 12,000 customers, up from 23% at launch in 2023.8,9
  • 64% of customers would prefer that companies did not use AI for customer service, and 53% would consider switching to a competitor if they found out a company was going to, according to Gartner.3
  • Support agents with access to a generative AI assistant resolved 14% more issues per hour on average, and novice and low-skilled agents resolved 34% more.11
  • Only 21% of service and support leaders say their organizations have reduced human agent headcount because of AI, and 7% have already rehired workers after earlier cuts, according to Gartner's 2026 survey.4

Customer service was one of the first business functions where companies put generative AI in front of customers, and it is where results are easiest to count: cases resolved, minutes per contact, and cost per transaction. In 2026, 91% of service leaders said they faced executive pressure to adopt AI, according to Gartner, while only 4% said they had achieved the return they expected.5 Vendor claims, company earnings reports, and independent studies often point in different directions.

Our analysts read Gartner's press releases on customer attitudes (July 2024), its agentic AI prediction (March 2025), and its 2026 workforce survey, plus CMSWire's report from Gartner's November 2026 Customer Service and Support Conference. We used Salesforce's 7th State of Service report (November 2025), Zendesk's 2025 CX Trends report, Intercom's published Fin figures, Klarna's announcements and earnings figures as reported by CX Dive, and the Brynjolfsson, Li, and Raymond study published by NBER and the Quarterly Journal of Economics. All sources were opened on November 16, 2026.

In the following sections, we compare AI adoption in customer service, AI resolution rates, customer attitudes, cost per contact, the productivity of agents working with AI, and the effect of AI on customer service jobs.

Service Teams Estimate AI Handles 30% of Cases and Expect 50% by 2027

Surveys of service teams are the main source for how much customer service work AI already does. We collected the adoption figures and forecasts from Salesforce, Gartner, and Zendesk.

AI Adoption in Customer Service
MeasurePeriodValuePublisher
Share of service cases handled by AI, as estimated by service teams202530%Salesforce
Projected share of service cases handled by AI202750%Salesforce
Rank of AI among service leaders' priorities2024 to 2025Rose from 10th to 2ndSalesforce
Service leaders saying security concerns delayed or limited AI initiatives202551%Salesforce
Service leaders under pressure to adopt AI2023 survey60%Gartner, 2024
Service leaders facing executive pressure to implement AI202691%Gartner, via CMSWire
Common customer service issues resolved by agentic AI without human help, prediction202980%, with a 30% cut in operational costsGartner, 2025
CX leaders expecting 80% of interactions to be resolved without a human in the next few years202475%Zendesk

Source: AI Services Companies analysis of Salesforce, State of Service, Gartner, Gartner, CMSWire, and Zendesk, CX Trends 2025 data, conducted November 2026. Data: customer-service-cases-handled-by-ai-2025-2027.csv

Waffle chart of the share of customer service cases handled by AI: 30 of 100 squares filled for 2025 and 50 of 100 squares filled for the 2027 projection

Source: AI Services Companies analysis of Salesforce data, conducted November 2026.

Key Findings

  • The Salesforce figures are estimates by service professionals, not counts of cases. They came from a survey of 6,500 service professionals and decision makers in 40 countries conducted from April 25 to June 6, 2025.6
  • AI rose from the 10th to the 2nd priority for service leaders in a year. Improving the customer experience stayed first.6
  • Gartner predicts that by 2029 agentic AI will resolve 80% of common customer service issues without human help, cutting operational costs by 30%. The prediction covers common issues, so it does not mean 80% of all contacts.2

AI Agents Now Resolve Most Simple Conversations at the Companies That Publish Rates

We collected published AI resolution rates from Intercom, Klarna, and a Zendesk customer to see how often AI agents close conversations on their own. Resolution rate, the share of conversations an AI agent closes without handing off to a person, is the most quoted performance figure for AI in customer service, and only a few vendors and companies publish it.

AI Resolution and Deflection Rates Reported by Vendors and Companies
Company or productMeasureValueDate reported
Intercom FinAverage resolution rate at launch23%2023
Intercom FinAverage resolution rate at the Fin 2 launch52%October 2024
Intercom FinAverage resolution rate across 6,000+ customers66%October 2025
Intercom FinAverage resolution rate across 12,000+ customers76%November 2026
Intercom FinShare of customers with resolution rates above 80%More than 20%October 2025
Klarna AI assistantShare of customer service chats handled in the first monthTwo-thirds (2.3 million conversations)February 2024
Klarna AI assistantDrop in repeat inquiries25%February 2024
Vagaro, using Zendesk AIShare of incoming requests resolved by AI44%November 2024

Source: AI Services Companies analysis of Intercom Fin, Intercom, Fin Ideas, Intercom blog, Klarna, and Zendesk data, conducted November 2026. Data: intercom-fin-resolution-rate-2023-2026.csv

Line chart of Intercom Fin average resolution rate: 23% at launch in 2023, 52% at the Fin 2 launch in October 2024, 66% in October 2025, and 76% in November 2026

Source: AI Services Companies analysis of Intercom data, conducted November 2026.

Key Findings

  • Intercom's average resolution rate rose 53 percentage points from launch to November 2026, by our analysis of its published averages. Intercom said the rate climbed by about one percentage point a month after the Fin 2 launch.1,8,9
  • Intercom itself notes that a 66% resolution rate does not mean Fin does 66% of the work, because the conversations it resolves are often the simplest ones.10
  • Vendor resolution rates are self-reported and defined by each vendor. Quote them with the company name and the date, since Intercom's customer base grew from thousands to more than 12,000 over the period.8,9

Most Customers Would Prefer Companies Not Use AI for Service, but Many Will Delegate Simple Tasks

Customer surveys show both resistance to AI in customer service and willingness to use it for routine tasks. We collected the consumer findings from Gartner and Zendesk.

Customer Attitudes Toward AI in Customer Service
FindingShare of consumersPublisher and survey
Would prefer that companies did not use AI for customer service64%Gartner, 5,728 customers, December 2023
Would consider switching to a competitor if a company was going to use AI for customer service53%Gartner, December 2023
Top concern about AI in customer serviceIt will get harder to reach a personGartner, December 2023
More likely to trust AI agents that are friendly and empathetic64%Zendesk, nearly 5,100 consumers, June to July 2024
Ready to delegate tasks such as order tracking to AI67%Zendesk, 2024
Expect AI interactions to feel tailored to them61%Zendesk, 2024
Have used voice AIHalfZendesk, 2024
Would switch to a competitor after one bad experience63%Zendesk, 2024
Customers who use the phone in a service interactionOne thirdGartner, 2026, via CMSWire

Source: AI Services Companies analysis of Gartner, Zendesk, CX Trends 2025, and CMSWire data, conducted November 2026. Data: customer-attitudes-ai-customer-service.csv

Horizontal bar chart of customer attitudes: 64% prefer companies not use AI for service, 53% would consider switching, 63% would switch after one bad experience, 67% ready to delegate tasks to AI, 64% more likely to trust friendly AI agents, 61% expect tailored AI interactions, and 50% have used voice AI

Source: AI Services Companies analysis of Gartner and Zendesk data, conducted November 2026.

Key Findings

  • Customers' top concern about AI in customer service is that it will become harder to reach a person, followed by AI displacing jobs and AI giving wrong answers, according to Gartner.3
  • The Gartner and Zendesk figures are not in conflict. Gartner asked about AI in customer service in general, and Zendesk asked about specific tasks and traits. Customers accept AI for order tracking and similar errands while resisting it as a barrier to a person.3,7
  • One third of customers use the phone in a customer service interaction, and only the website is used more, according to Gartner analyst Emily Potosky.5

Klarna Cut Its Customer Service Cost per Transaction by 40% in Two Years

Few companies publish their cost per customer contact, so Klarna's disclosures are the most cited real-world figures on AI and service costs. We collected them alongside Gartner's data on AI spending and returns.13,14

Customer Service Costs and Returns From AI
MeasurePeriodValuePublisher
Klarna customer service cost per transactionQ1 2023$0.32Klarna, via CX Dive
Klarna customer service cost per transactionQ1 2025$0.19 (down 40%)Klarna, via CX Dive
Klarna estimated profit improvement from its AI assistant2024$40 millionKlarna, February 2024
Klarna savings attributed to its AI assistantAs of Q3 2025$60 millionKlarna CEO, via CX Dive
Klarna customer service and operations costQ1 2024 to Q1 2025$57 million to $51 millionKlarna, via CX Dive
Klarna customer service and operations costQ3 2024 to Q3 2025$42 million to $50 millionKlarna, via CX Dive
Growth in AI spending by customer service leaders202638% (overall service budgets grew 2%)Gartner, via CMSWire
Service organizations that achieved the ROI they expected from AI20264%Gartner, via CMSWire
Service organizations reporting any positive ROI from AI202624%Gartner, via CMSWire

Source: AI Services Companies analysis of CX Dive, May 2025, CX Dive, November 2025, Klarna, and CMSWire data, conducted November 2026. Data: klarna-customer-service-cost-and-resolution-time.csv

Two-panel column chart of Klarna customer service: cost per transaction fell from $0.32 in Q1 2023 to $0.19 in Q1 2025, and average time to resolve an errand fell from 11 minutes to under 2 minutes with its AI assistant

Source: AI Services Companies analysis of Klarna data, conducted November 2026.

Key Findings

  • Klarna said in February 2024 that customers resolved errands in under 2 minutes with its AI assistant, compared with 11 minutes before, and that the assistant was doing the work of 700 full-time agents. By November 2025 it put that figure at 853 agents.12,14
  • Klarna's customer service and operations cost rose to $50 million in Q3 2025 from $42 million a year earlier, even as it reported $60 million in AI savings. The company grew to 114 million active users over the same period, so cost per transaction and total cost moved in different directions.14
  • Only 4% of service organizations have achieved the return they expected from AI, and 24% report any positive return, according to Gartner. Gartner analyst Christopher Sladdin pointed to rising AI costs and the difficulty of turning time savings into financial savings.5

AI Assistance Raised Support Agent Productivity by 14%, and by 34% for Newer Agents

The strongest evidence on AI and agent productivity comes from a study of a real deployment at a large software company's support operation. We collected its findings with survey data from Salesforce and Zendesk on agents who use AI.

Productivity of Customer Support Agents Using AI
MeasureValueSource
Change in issues resolved per hour, all agents with AI assistantUp 14% on averageBrynjolfsson, Li, and Raymond, NBER
Change in issues resolved per hour, novice and low-skilled agentsUp 34%Brynjolfsson, Li, and Raymond, NBER
Change for experienced and highly skilled agentsMinimalBrynjolfsson, Li, and Raymond, NBER
Agents in the study5,179 customer support agentsBrynjolfsson, Li, and Raymond, NBER
Less time spent on routine cases by reps using AI20% (about four hours a week)Salesforce, 2025
Reps with AI who say it creates growth opportunities71%Salesforce, 2025
Reps with AI who have developed new skills86%Salesforce, 2025
Agents who believe an AI copilot would help them do their job better73%Zendesk, 2024

Source: AI Services Companies analysis of Brynjolfsson, Li, and Raymond, Generative AI at Work (NBER Working Paper 31161), Salesforce, State of Service, and Zendesk, CX Trends 2025 data, conducted November 2026. Data: ai-agent-assist-productivity.csv

Horizontal bar chart of productivity effects of AI assistance: issues resolved per hour up 14% for all agents and 34% for novice and low-skilled agents, and 20% less time on routine cases for service reps using AI

Source: AI Services Companies analysis of NBER and Salesforce data, conducted November 2026.

Key Findings

  • The study found that the AI assistant spread the practices of the most able agents to newer ones and helped them move down the experience curve faster. It also found improved customer sentiment and higher employee retention.11
  • The study was first released as an NBER working paper in April 2023 and published in the Quarterly Journal of Economics in 2025, volume 140, issue 2.11
  • Salesforce's productivity figures are self-reported by survey respondents. Salesforce noted that the optimism of agentic AI users may reflect selection bias among early adopters.6

Only 21% of Service Leaders Have Cut Agent Headcount Because of AI

We collected Gartner's 2026 findings on how service organizations are changing their workforce, to see whether forecasts that AI would quickly shrink contact centers show up in what leaders report.

Customer Service Workforce Changes Because of AI, 2026
MeasureShare of service and support leaders
Shifting human agents into different roles within service and support71%
Hiring freeze45%
Shifting human agents into other business units39%
Reduced human agent headcount because of AI21%
Using forced reductions as a workforce strategy20%
Plan to rehire human workers within the next 18 months11%
Have rehired human workers after previous reductions7%
Average headcount reduction among organizations that cut, previous 18 months16.7% of the workforce
Hired specialized staff to support AI initiativesMore than half

Source: Gartner, Gartner Survey Finds 21% of Service and Support Leaders Have Reduced Human Agent Headcount Because of AI, November 5, 2026. Data: customer-service-workforce-changes-ai-2026.csv

Lollipop chart of workforce changes among service leaders in 2026: 71% shifting agents into different roles, 45% hiring freeze, 39% shifting agents to other units, 21% reduced headcount because of AI, 20% forced reductions, 11% plan to rehire, and 7% have rehired

Source: AI Services Companies analysis of Gartner data, conducted November 2026.

Key Findings

  • Gartner surveyed 275 service and support leaders in July and August 2026. Among the 21% whose organizations cut agents because of AI, the average reduction was 16.7% of the workforce over 18 months.4
  • Gartner predicts that half of companies that cut customer service staff because of AI will rehire by 2027. The 7% that have already rehired and the 11% that plan to are the first signs of that.4,5
  • In April 2026, 31% of service leaders had implemented or were planning AI-driven frontline layoffs, according to Gartner, so plans ran ahead of the cuts that took place.5

What AI in Customer Service Means Going Into 2027

The numbers show AI doing a growing share of simple customer service work. Service teams estimate it handles 30% of cases, the vendors that publish rates report averages of two-thirds or more, and companies such as Klarna report lower cost per transaction and faster resolutions. The independent evidence on agents is also positive, with the largest gains for newer staff.

The figures on returns and jobs are more modest. Few service organizations have reached the returns they expected, most have not cut agents, and some that did are hiring again. Customers still call, and most say they would prefer companies not use AI for service at all.

So if you cite a figure for AI in customer service, say whether it is a vendor's self-reported rate, a company disclosure, a survey estimate, or an independent study, and give the date, because figures from these four kinds of sources can differ widely for the same measure.

Data Collection and Methodology

All sources were opened on November 16, 2026. Survey figures come from Gartner press releases (a survey of 5,728 customers in December 2023, published July 2024; a prediction published March 2025; and a survey of 275 service and support leaders in July and August 2026, published November 5, 2026), CMSWire's report from Gartner's Customer Service and Support Conference on November 4 and 5, 2026, Salesforce's 7th State of Service report (6,500 service professionals, April to June 2025), and Zendesk's 2025 CX Trends report (nearly 5,100 consumers and 5,400 service leaders, agents, and buyers, June to July 2024). Figures attributed to Gartner via CMSWire were reported by CMSWire from conference sessions. Gartner also announced the 4% ROI figure in a November 4, 2026 press release, which we did not open.

Resolution rates for Intercom Fin are Intercom's own averages across its customers, taken from its October 2025 posts and its product site; the launch year is 2023 and the customer counts differ by date. Klarna figures come from its February 27, 2024 press release and from its earnings figures as reported by CX Dive in May and November 2025. Productivity figures come from Brynjolfsson, Li, and Raymond, Generative AI at Work (NBER Working Paper 31161; Quarterly Journal of Economics, 2025), which measured productivity as issues resolved per hour. In the Klarna chart, under 2 minutes is drawn at 2 minutes. We calculated differences and ratios from the published values (for example, 76% minus 23% for the change in Fin's resolution rate and $0.19 against $0.32 for Klarna's cost per transaction).

Sources (14)
  1. AI Services Companies analysis of AI customer service data from Gartner, Salesforce, Zendesk, Intercom, Klarna, and NBER, 2023 to 2026, conducted November 2026.
  2. Gartner, Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029 (March 5, 2025)
  3. Gartner, Gartner Survey Finds 64% of Customers Would Prefer That Companies Didn't Use AI For Customer Service (July 9, 2024)
  4. Gartner, Gartner Survey Finds 21% of Service and Support Leaders Have Reduced Human Agent Headcount Because of AI (November 5, 2026)
  5. CMSWire, Just 4% of Customer Service Leaders Hit Their Expected Automation Returns (November 10, 2026)
  6. Salesforce, AI Expected to Resolve Half of Service Cases by 2027, Data Shows (State of Service, 7th edition, November 13, 2025)
  7. Zendesk, 2025 CX Trends Report: Human-Centric AI Drives Loyalty (November 20, 2024)
  8. Intercom, Fin product site (opened November 16, 2026)
  9. Intercom, Fin Ideas: The next leap forward for Fin (October 17, 2025)
  10. Intercom, What's new with Fin 3 (October 15, 2025)
  11. Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond, Generative AI at Work, NBER Working Paper 31161 (Quarterly Journal of Economics, 2025)
  12. Klarna, Klarna AI assistant handles two-thirds of customer service chats in its first month (February 27, 2024)
  13. CX Dive, Klarna credits AI for slashing customer service costs (May 21, 2025)
  14. CX Dive, Klarna says its AI agent is doing the work of 853 employees (November 20, 2025)