The statistics library

612 sourced numbers on customer support

The figures cited across reports, landscape guides, benchmarks, and dispatches — each traceable to its linked source and context. Repeated figures stay separate when their publication context differs. Copy a citation or share a number straight to LinkedIn.

427 numbersClear search and filters

2 Aug 2026

date the EU AI Act transparency rules for covered interactive AI systems begin to apply

European Commission, AI Act policy and implementation timeline Insight: The EU AI Act chatbot deadline is an operating change, not a footer update

Article 50

provision requiring covered AI systems that interact directly with people to inform them that they are interacting with AI

Regulation (EU) 2024/1689, official text Insight: The EU AI Act chatbot deadline is an operating change, not a footer update

10 Jun 2026

publication date of the EU Code of Practice on transparency of AI-generated content

European Commission, AI Office Insight: The EU AI Act chatbot deadline is an operating change, not a footer update

implementation questions the Commission says its Article 50 guidance will clarify: scope, concepts, practical application, and demonstrating compliance

European Commission, Article 50 transparency guidance FAQ Insight: The EU AI Act chatbot deadline is an operating change, not a footer update

70%

Average first-call resolution rate across call-center industries — 'good' is 70-79%, world-class is 80%+

SQM Group, Call Center Industry Standards / KPI Benchmarks Insight: FCR is the most gamed metric in support

7 minutes

Industry benchmark average handle time (AHT) for call centers

SQM Group, Call Center Industry Standards / KPI Benchmarks Insight: FCR is the most gamed metric in support

63%

of consumers say they'll switch to a competitor after just one bad experience — up 9% year-on-year

Zendesk CX Trends 2025 Report Insight: FCR is the most gamed metric in support

88%

2023 industry benchmark internal quality score (IQS) from QA conversation reviews

Zendesk QA (Klaus), Customer Service Quality Benchmark Report 2023 (cited on Zendesk blog) Insight: FCR is the most gamed metric in support

14%

of customer-service issues customers fully resolve in self-service — only 36% even for 'very simple' issues

Gartner survey, Aug 2024 (reported by destinationCRM) Insight: FCR is the most gamed metric in support

30-50

Tags in the sweet spot for a support taxonomy agents can actually use — past this, tagging accuracy and reporting both degrade

SentiSum, "Best practice for building a tagging taxonomy" (Kirsty Pinner) Insight: A tagging taxonomy that survives past six months

~1 in 3

Support organizations whose staff do not use the predefined ticket categories as they were intended, in a survey of 461 organizations

HDI SupportWorld, ticket categorization survey Insight: A tagging taxonomy that survives past six months

57%

Support organizations that do not measure whether tickets are categorized correctly at all

HDI SupportWorld, ticket categorization survey (461 organizations) Insight: A tagging taxonomy that survives past six months

67%

Data professionals who do not fully trust the data their organization uses for decision-making

Drexel University LeBow College of Business & Precisely, Data Integrity Trends and Insights 2025 (reported by diginomica) Insight: A tagging taxonomy that survives past six months

$12.9M

Average amount poor data quality costs an organization per year

Gartner, "Data Quality: Why It Matters and How to Achieve It" (2020 research) Insight: A tagging taxonomy that survives past six months

12%

Organizations that say their data is of sufficient quality and accessibility for AI initiatives

Drexel University LeBow College of Business & Precisely, Data Integrity Trends and Insights 2025 (reported by diginomica) Insight: A tagging taxonomy that survives past six months

54%

Organizations whose most recent significant, serious or severe outage cost more than $100,000 — with 16% reporting more than $1 million

Uptime Institute, Annual Outage Analysis 2024 (executive summary) Insight: When production breaks, support pays the bill

73%

Consumers who will switch to a competitor after multiple bad experiences — which an outage delivers to the whole customer base in one afternoon

Zendesk Benchmark data, customer service statistics 2026 Insight: When production breaks, support pays the bill

85%

Share of human-error outages that stem from staff failing to follow procedures — or from flaws in the procedures themselves

Uptime Institute, Annual Outage Analysis 2025 press release Insight: When production breaks, support pays the bill

88%

Customers who expect faster response times than they did just one year ago — a patience window that shrinks further mid-incident

Zendesk CX Trends 2026 Insight: When production breaks, support pays the bill

88%

Customers who expect faster response times than they did just one year ago

Zendesk CX Trends 2026 Insight: Backlog triage: digging out without burning the team

>90%

Occupancy level above which agent burnout becomes the most obvious consequence

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: Backlog triage: digging out without burning the team

83.3%

Average maximum occupancy target professionals enter into the Erlang staffing calculator, across 160,000+ calculations

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: Backlog triage: digging out without burning the team

25%

Backlog reduction in the first 10 weeks after Ping Identity moved from tiered support to swarming

Consortium for Service Innovation, Intelligent Swarming Case Study: Ping Identity, 2024 Insight: Backlog triage: digging out without burning the team

81%

Reduction in time to resolution on the most complicated cases after Ping Identity adopted intelligent swarming

Consortium for Service Innovation, Intelligent Swarming Case Study: Ping Identity, 2024 Insight: Backlog triage: digging out without burning the team

88%

Customers who expect faster response times than they did just one year ago

Zendesk CX Trends 2026 Insight: SLAs that protect customers, not dashboards

85%

CX leaders who say customers will drop a brand over an unresolved issue, even on the first contact

Zendesk CX Trends 2026 Insight: SLAs that protect customers, not dashboards

>90%

Occupancy level above which agent burnout becomes the most obvious consequence, per Erlang staffing guidance

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: SLAs that protect customers, not dashboards

75%

Improvement in time to first response in Ping Identity's first 10 weeks after replacing tiered escalation with swarming

Consortium for Service Innovation, Intelligent Swarming Case Study: Ping Identity, 2024 Insight: SLAs that protect customers, not dashboards

81%

Reduction in time to resolution on Ping Identity's most complicated cases after adopting intelligent swarming

Consortium for Service Innovation, Intelligent Swarming Case Study: Ping Identity, 2024 Insight: SLAs that protect customers, not dashboards

25%

Reduction in support backlog in Ping Identity's first 10 weeks on swarming, with SLA adherence up 13%

Consortium for Service Innovation, Intelligent Swarming Case Study: Ping Identity, 2024 Insight: SLAs that protect customers, not dashboards

80/20

The near-universal service-level target — answer 80% of contacts within 20 seconds — has no research behind it; industry veterans admit no one knows where it came from.

Call Centre Helper, Is 80/20 Still a Reasonable Service Level?, 2019 Insight: Staffing to a service level

1917

Year Danish mathematician A.K. Erlang published the queueing formula (Erlang C) still used to convert a service-level target into a required agent count.

TechTarget, What is Erlang C and how is it used for call centers? Insight: Staffing to a service level

21 vs 28 agents

In a worked Erlang example for the same call load, 21 agents answer just 32% of contacts within 20 seconds while 28 answer 97% — the staffing curve is steep, then flattens.

SWPP (Society of Workforce Planning Professionals), Calculating Call Center Staff Insight: Staffing to a service level

83.3%

Average maximum occupancy target contact-center professionals set in the Erlang staffing calculator across 160,000+ calculations — the practical ceiling before quality and retention suffer.

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: Staffing to a service level

20–35%

Share of paid agent time lost to breaks, training, meetings and absence in most centers; divide your Erlang agent count by (1 − shrinkage) to get the schedules you actually need to hire.

SWPP, Calculating Call Center Staff Insight: Staffing to a service level

8% / 7% / 3%

Average abandonment for voice, synchronous and async channels — but the benchmark warns these are context, not a target: service level is the driver, abandonment the outcome.

ACXPA, Abandonment Rate: Contact Centre Guide & Benchmarks Insight: Staffing to a service level

88%

of customers now expect faster response times than they did just a year ago

Zendesk CX Trends 2026 Insight: Forecasting ticket volume without a data science team

83.3%

average maximum occupancy target contact-center professionals enter into the Erlang staffing calculator, across 160,000+ calculations

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: Forecasting ticket volume without a data science team

>90%

occupancy level above which agent burnout becomes the most obvious consequence, per Erlang staffing guidance

Call Centre Helper, How to Calculate Occupancy in the Contact Centre, 2024 Insight: Forecasting ticket volume without a data science team

38%

average annual call-center agent turnover in 2022, which SQM ties directly to declining FCR and CSAT

SQM Group press release, Call Center Attrition Rate: It Is Now the Most Important KPI, 2023 Insight: Forecasting ticket volume without a data science team

74%

of consumers now expect customer service to be available 24/7 — a coverage floor forecasting has to respect

Zendesk CX Trends 2026 Insight: Forecasting ticket volume without a data science team

700 agents

worth of work Klarna said its AI assistant did in its first month (Feb 2024) — two-thirds of all chats, with a projected $40M lift to 2024 profit

Klarna, February 2024 Insight: The 2024 deflection wave, revisited

14%

of customer service issues are fully resolved in self-service — even though 73% of customers start there

Gartner, 2024 Insight: The 2024 deflection wave, revisited

76%

Fin's marketed average resolution rate — counted by Intercom's own rule, where a customer who stops replying within 24 hours is scored as resolved

Intercom, 'From resolutions to outcomes' Insight: The 2024 deflection wave, revisited

50%

of organizations that expected to significantly cut their customer service workforce will abandon those plans, Gartner predicted in June 2025

Gartner, June 2025 Insight: The 2024 deflection wave, revisited

20%

of customer service leaders reported any AI-driven headcount reduction in a Gartner survey of 321 leaders (Oct 2025); 55% held staffing flat while handling more volume

Gartner, December 2025 Insight: The 2024 deflection wave, revisited

Half

of companies that cut customer service staff because of AI will rehire for those roles by 2027, Gartner predicts

Gartner, February 2026 Insight: The 2024 deflection wave, revisited

14%

average increase in issues resolved per hour when support agents get a generative-AI assistant

NBER Working Paper 31161, "Generative AI at Work" (Brynjolfsson, Li, Raymond) 2023 Insight: Agent assist vs. autonomous: choosing the seam

The Support Brief

The most useful new numbers, sourced in one monthly issue.

Monthly. No ads. Unsubscribe anytime. Privacy