Key Takeaways
- The industry-average AHT across all channels is 6 to 10 minutes for routine contacts, with phone contacts consistently running the longest at a median of 7.5 minutes
- After-call work accounts for 15 to 25 percent of total AHT in well-run centers; centers where ACW exceeds 35 percent of AHT are leaving measurable capacity on the table
- Cutting AHT more than 20 percent below benchmark without equivalent FCR gains causes CSAT scores to drop by an average of 8 points - faster resolution that leaves the issue open is not resolution
- AI assist tools reduce handle time by 25 to 35 percent on routine contacts; knowledge base maturity cuts AHT by 20 to 40 percent depending on content quality and search integration
- Each 60-second reduction in AHT on a 100-agent team handling 1,000 daily calls frees roughly 1,000 minutes of capacity - equivalent to one additional full-time agent per shift
Why average handle time statistics matter in 2026
Average handle time is the clock that runs every support operation. How many contacts a given team can process in a shift, how much each contact costs to resolve, how fast queues build during volume spikes - all of it flows from AHT.
That relationship holds whether you run a 10-person in-house team or a 500-seat outsourced center. The arithmetic does not change. What changes is how AHT is measured, what drives it in each channel, and where the tradeoffs sit between speed and quality.
The data below covers AHT benchmarks by channel and industry, what AHT reduction does to customer satisfaction, how AI and knowledge base maturity affect handle time, the after-call work component most benchmark reports underweight, and the link between handle time and cost per contact.
For broader agent productivity context, see customer support agent productivity statistics for 2026. For the staffing model implications that flow from these benchmarks, see customer support staffing ratios statistics for 2026.
What AHT actually measures
Average handle time combines three components: talk time (or active response time), hold time, and after-call work (ACW).
Talk time is the period the agent is directly engaged with the customer - speaking on a call, typing a chat response, or composing an email reply.
Hold time is the period the customer waits while the agent retrieves information, escalates internally, or consults a colleague. On voice channels it is measured in real seconds. On asynchronous channels like email, the equivalent is the gap between agent reply sequences.
After-call work is the time spent completing the contact record after the interaction ends - updating the CRM, logging notes, categorizing the ticket, processing any follow-up tasks. ACW does not involve the customer but occupies the agent's capacity fully.
The formula is: AHT = Talk Time + Hold Time + After-Call Work Time.
The most common benchmarking error is using AHT as a synonym for talk time. Centers that track only talk time systematically undercount handle time by 20 to 40 percent, which causes their capacity models to overestimate how much work a given headcount can handle.
ICMI's 2025 Contact Center Benchmarking Report found that 31 percent of the centers surveyed measured AHT as talk time only, excluding hold time and ACW. Those centers showed a consistent pattern: they appeared to be outperforming AHT benchmarks while simultaneously running understaffed against their actual service level targets. The disconnect was the missing ACW.
AHT benchmarks by channel
Channel is the single largest driver of AHT variation. A phone call, a chat session, and an email response involve different cognitive loads, different pacing, and different tools - handle times reflect all of those differences.
| Channel | AHT (Routine Contacts) | AHT (Complex / Escalated) | Source |
|---|---|---|---|
| Voice (inbound phone) | 6 to 10 minutes | 18 to 30 minutes | ICMI Contact Center Benchmarking Report 2025 |
| Live chat (per session) | 8 to 12 minutes | 15 to 24 minutes | Zendesk Benchmark Report 2025 |
| Email (agent active time per ticket) | 4 to 6 minutes | 8 to 15 minutes | Salesforce State of Service 2025 |
| Social media DM | 5 to 9 minutes | 10 to 18 minutes | Salesforce State of Service 2025 |
| Messaging apps (SMS / WhatsApp) | 6 to 11 minutes | 12 to 20 minutes | Zendesk Benchmark Report 2025 |
| Callback (outbound to resolve inbound request) | 7 to 13 minutes | 20 to 28 minutes | ICMI Contact Center Benchmarking Report 2025 |
ICMI's 2025 benchmarking report - covering more than 400 North American contact centers - puts the voice AHT median at 7.5 minutes for routine contacts. That median has held within a one-minute band for the past three years, but the distribution has widened: the top decile by AHT is getting faster (under 5 minutes), while the bottom decile is getting slower (above 14 minutes). Automation is widening the gap between well-resourced and under-resourced operations.
Chat AHT runs longer than most operations expect. The 8 to 12 minute range from Zendesk's 2025 benchmark covers the time per session, which includes the gaps between messages while the customer types. Agents running concurrent sessions are not idle during those gaps, which is why chat productivity is measured in sessions per hour rather than handle time alone. Handle time comparisons across voice and chat without accounting for concurrency produce misleading results.
Email agent active time - the period the agent is actually working on the ticket, not the overall ticket lifecycle from submission to resolution - runs 4 to 6 minutes for routine contacts. That figure excludes the waiting time between the customer's submission and the agent's first response, which is a separate metric tracked as first response time. For that benchmark see average customer support response times.
AHT benchmarks by industry
Industry is the second major driver of handle time variation. Complexity per contact, regulatory requirements, account lookup depth, and the average technical sophistication of the customer base all push AHT up or down by material amounts.
| Industry | Average AHT (All Channels) | Phone AHT Benchmark | Source |
|---|---|---|---|
| Financial services and banking | 8 to 12 minutes | 9 to 14 minutes | ICMI Contact Center Benchmarking Report 2025 |
| Healthcare (patient services and billing) | 10 to 16 minutes | 12 to 18 minutes | Gartner Customer Service Survey 2025 |
| Technology / SaaS support | 7 to 11 minutes | 8 to 13 minutes | Zendesk Benchmark Report 2025 |
| Retail and e-commerce | 4 to 7 minutes | 5 to 8 minutes | Talkdesk Research 2025 |
| Telecom and utilities | 8 to 14 minutes | 9 to 16 minutes | ICMI Contact Center Benchmarking Report 2025 |
| Insurance | 9 to 15 minutes | 11 to 17 minutes | Gartner Customer Service Survey 2025 |
| Travel and hospitality | 6 to 10 minutes | 7 to 12 minutes | Salesforce State of Service 2025 |
| B2B enterprise software | 12 to 20 minutes | 14 to 22 minutes | HubSpot Customer Support Survey 2025 |
Financial services AHT is elevated by verification requirements. Regulatory compliance mandates identity verification before account access, which adds 60 to 90 seconds to every interaction before the substantive issue is even addressed. Gartner's 2025 analysis found that verification steps alone account for 15 to 22 percent of total call time in regulated financial institutions.
Healthcare handle times are the longest across the benchmark sources. HIPAA verification, insurance eligibility lookups, and the multi-party nature of healthcare billing mean a single contact often involves three or four separate data systems. Gartner found that healthcare agents spend an average of 4.2 minutes per call in active hold or lookup time alone.
Retail and e-commerce runs the shortest average AHT. Order status, return processing, and shipping inquiries are high-volume, low-complexity, and increasingly handled by agent-assist tools that surface the relevant order data before the agent picks up the conversation. Talkdesk's 2025 industry report found that retail operations using integrated order management overlays reduced AHT by 31 percent relative to teams relying on manual lookups.
B2B enterprise software has the longest handle times outside healthcare. HubSpot's 2025 customer support survey found median AHT of 15 minutes for enterprise tier contacts, driven by technical depth, multi-user account structures, and the integration complexity that comes with B2B software environments. Enterprise support teams with dedicated technical specialists consistently outperformed blended teams by 18 to 25 percent on first-contact resolution, though at higher per-agent cost.
The AHT vs CSAT tradeoff
Reducing AHT is one of the most common contact center objectives. It is also one of the most reliably mismanaged.
The speed-quality tradeoff is real, but it is not uniform. Small AHT reductions - achieved through better tooling, process improvements, or reduced hold time - can improve CSAT by cutting customer wait time without degrading resolution quality. Push AHT targets aggressively and the result tends to reverse.
ICMI's 2025 benchmarking report found that contact centers reducing AHT more than 20 percent below their channel benchmark - without corresponding first-contact resolution (FCR) gains - experienced an average CSAT score decline of 8.3 points on a 100-point scale. Agents wrap interactions before the underlying issue is resolved to hit their handle time target. Customers call back.
| AHT Reduction vs Benchmark | Typical CSAT Impact | FCR Impact | Source |
|---|---|---|---|
| 1 to 10% below benchmark (tool / process improvement) | Neutral to +2 points | Neutral | ICMI 2025 |
| 11 to 20% below benchmark | Neutral to -3 points | -2 to -4% | ICMI 2025 |
| 21 to 30% below benchmark (without process change) | -5 to -9 points | -5 to -9% | ICMI 2025 |
| Above 30% below benchmark | -10 to -15 points | -10 to -16% | Gartner Customer Service Survey 2025 |
Gartner's 2025 customer service survey found the CSAT penalty is concentrated in complex contacts. For simple, transactional contacts - order status checks, password resets, basic account lookups - AHT reductions of 15 to 25 percent show no CSAT impact because the interaction does not require conversational depth. The risk materializes when AHT targets are applied uniformly across a contact mix that includes billing disputes, technical troubleshooting, and complaints.
SQM Group's 2025 benchmarking study found that the top quartile of centers by customer satisfaction actually ran AHTs 12 to 18 percent above the industry median. They were spending more time per contact and producing better outcomes: FCR rates 11 percentage points above average and CSAT scores 14 points above average. The centers at the bottom of the CSAT distribution were, on average, 8 percent below median AHT.
HubSpot's 2025 customer support survey asked customers directly what drove dissatisfaction with support contacts. The top answer was not long wait time or long handle time - it was "I had to contact support again because the issue wasn't resolved." Contacts that run 2 minutes longer but resolve the issue produce better CSAT than contacts that run fast and reopen. That basic finding recurs across multiple independent studies.
Talkdesk's 2025 industry report frames the practical implication: AHT is a capacity metric. CSAT and FCR are quality metrics. Managing AHT as a quality metric produces neither efficiency nor quality.
After-call work time benchmarks
After-call work (ACW) - the post-interaction period agents spend completing records, logging notes, and processing follow-ups - is the least-tracked component of AHT and one of the most controllable.
| ACW Metric | Top Quartile Centers | Industry Average | Bottom Quartile | Source |
|---|---|---|---|---|
| ACW as % of total AHT | 15 to 20% | 20 to 28% | 30 to 40% | Talkdesk Research 2025 |
| Voice ACW absolute time (minutes) | 0.8 to 1.5 min | 1.5 to 2.5 min | 3.0 to 5.0 min | ICMI Contact Center Benchmarking Report 2025 |
| Chat ACW absolute time | 0.5 to 1.0 min | 1.0 to 2.0 min | 2.0 to 4.0 min | Zendesk Benchmark Report 2025 |
| % of centers with automated ACW tools | 54% | - | - | Gartner Customer Service Survey 2025 |
| ACW reduction from post-call AI summarization | 60 to 75% | - | - | Zendesk CX Trends 2025 |
Talkdesk's 2025 research covers 2,500 contact centers globally. Top-quartile centers hold ACW to 15 to 20 percent of total AHT through a combination of CRM integration (automatic logging of contact metadata during the call), structured disposition codes (replacing free-text notes with categorized fields), and post-call summarization tools that draft the case record automatically.
ICMI's 2025 data shows voice ACW averaging 1.5 to 2.5 minutes across the industry, with the bottom quartile at 3 to 5 minutes. On a team handling 1,000 calls per day, the difference between a 1.5-minute and a 3-minute ACW is 25 agent-hours of daily capacity. That is equivalent to three additional full-time positions on an 8-hour shift.
CRM-telephony integration is where the ACW gains are most consistent across the 2025 benchmark data. When the contact platform and CRM share the same data layer, the agent's authentication, queue assignment, and contact reason are pre-populated before the interaction starts. Agents arriving at a call with the customer's account already pulled spend less time in post-call documentation because the structured fields were filled during the call. ICMI found that centers with full CRM-telephony integration ran ACW 32 percent below centers relying on manual post-call entry.
Post-call AI summarization has moved into the mainstream. Gartner found 54 percent of enterprise contact centers have deployed some form of automated ACW tooling as of 2025, up from 34 percent in 2023. Centers using AI summarization - where the tool drafts the case notes and the agent reviews and approves - showed ACW reductions of 60 to 75 percent relative to fully manual post-call documentation. On a team handling 200 voice contacts per day at a 2-minute ACW, a 65 percent reduction saves over 4 hours of capacity daily.
Impact of AI and automation on AHT
AI assist tools deployed inside the agent desktop have become the primary driver of AHT reduction in well-funded contact centers. The category includes response suggestion copilots, real-time knowledge base surfacing, automated hold-time lookup, CRM auto-population, and post-call summarization.
| AI Feature | AHT Reduction (Routine Contacts) | AHT Reduction (Complex Contacts) | Source |
|---|---|---|---|
| Response suggestion / AI copilot | 25 to 35% | 10 to 18% | Zendesk CX Trends 2025 |
| Real-time knowledge base surfacing | 20 to 30% hold time reduction | 15 to 22% | Talkdesk Research 2025 |
| CRM auto-population (telephony integration) | 12 to 20% ACW reduction | 10 to 15% | Salesforce State of Service 2025 |
| Post-call AI summarization | 60 to 75% ACW reduction | 55 to 70% | Zendesk CX Trends 2025 |
| Automated identity verification | 45 to 60 seconds saved per call | Same | Gartner Customer Service Survey 2025 |
| AI-powered routing (reduces mis-routes) | 8 to 14% reduction in transfers | Same | Talkdesk Research 2025 |
Zendesk's 2025 CX Trends report, covering 100,000-plus agent-hours across enterprise contact centers, found agents using AI copilot tools closed routine tickets 35 percent faster than agents without them. For complex contacts, the gain narrowed to 12 to 18 percent because the AI suggestion required more editing and the agent's review of lower-confidence suggestions consumed time.
Talkdesk's 2025 research adds a useful channel split: AI assist gains are larger on chat than on voice. Chat agents using real-time suggestion tools showed 38 percent AHT reduction on routine contacts, compared to 27 percent for voice agents using the same tools. The difference reflects that chat agents can review and edit a suggestion before sending, while voice agents must interrupt their conversation to check a suggested response - so suggestion quality matters more in the voice context.
Gartner's 2025 survey found that automated identity verification - biometric voice matching, knowledge-based authentication automation, or device-based authentication - saves 45 to 60 seconds per call in regulated industries. For financial services teams where verification is mandatory on every contact, that is 12 to 16 percent of total AHT on a 7-minute call.
Salesforce's 2025 State of Service report surveyed 5,500 service professionals globally. Among organizations that had deployed AI assist tools, 78 percent reported that AHT on routine contacts decreased by at least 20 percent within 90 days of deployment. Among the 22 percent that saw no improvement or degradation, the two most common causes were inadequate agent training on the new tools and poor AI suggestion quality from insufficient training data - specifically, knowledge base content that was too thin or too outdated for the model to surface relevant suggestions.
The Forrester Customer Service Technology Index 2025 found 49 percent of enterprise contact centers have deployed at least one AI assist feature as of 2025, up from 31 percent in 2023. The 2-year adoption rate is the fastest in any contact center technology category since cloud telephony adoption in 2016 to 2018.
Impact of knowledge base quality on AHT
Hold time - the period customers wait while agents search for information - is one of the most addressable components of AHT. Knowledge base quality is what determines whether those searches succeed quickly or fail repeatedly.
| Knowledge Base Maturity Level | AHT Impact vs Baseline | Hold Time Impact | Source |
|---|---|---|---|
| No structured KB (informal notes, institutional memory) | Baseline | Baseline | SQM Group 2025 |
| Basic KB (centralized articles, minimal search optimization) | -5 to -10% AHT | -8 to -15% hold time | SQM Group 2025 |
| Mature KB (search-optimized, regularly maintained, role-specific) | -20 to -30% AHT | -25 to -35% hold time | Gartner Customer Service Survey 2025 |
| AI-integrated KB (contextual surfacing, auto-updated) | -30 to -40% AHT | -35 to -45% hold time | Gartner Customer Service Survey 2025 |
SQM Group's 2025 benchmarking study shows a step-function improvement in AHT as knowledge base maturity increases. Moving from no structured KB to a basic centralized repository cuts AHT by 5 to 10 percent. Moving from basic to a maintained, search-optimized KB cuts another 15 to 20 percent. AI-integrated knowledge bases, where the system surfaces relevant articles contextually based on what the customer said, add another 10 to 15 percent on top of that.
The ceiling matters: AI-integrated KBs show 30 to 40 percent AHT reduction versus the no-KB baseline. Against a well-maintained traditional KB, the AI layer adds roughly 12 to 15 percent. The incremental case for AI-integrated KB is real but not transformative unless the baseline content is solid.
The standard failure mode in knowledge base programs is content decay. Gartner's 2025 survey found the median enterprise KB has a 34 percent article staleness rate - more than one in three articles is outdated, inaccurate, or superseded by a product change. When staleness exceeds 25 percent, KB usage drops sharply: agents learn through repeated failures that the articles cannot be trusted and revert to memory or colleague queries. At that point, the KB investment is generating no AHT benefit while consuming maintenance budget.
HubSpot's 2025 customer support survey found that organizations with a designated KB manager - a named owner responsible for content accuracy - had 21-point lower staleness rates and 19 percent lower average AHT than organizations where KB maintenance was distributed across the team without ownership. The distinction between "everyone maintains it" and "a specific person owns it" is what the data consistently separates.
Talkdesk's 2025 research adds a practical AHT context: the industries with the longest handle times are also the industries with the lowest KB utilization rates. Healthcare (KB utilization: 41 percent of contacts), insurance (44 percent), and financial services (52 percent) trail retail (72 percent) and SaaS (68 percent). The gap is partly regulatory - some healthcare and insurance information cannot be pre-written in articles because it depends on individual eligibility - but Talkdesk found that even in regulated industries, increasing KB utilization from 40 to 60 percent reduced average hold time by 22 percent without regulatory risk.
AHT trend over time: 2020 to 2026
AHT is not falling across the industry despite widespread AI adoption. It is rising slightly on a blended basis because automation is changing the mix of contacts that reach agents.
| Year | Industry-Average Blended AHT (All Channels) | Primary Driver |
|---|---|---|
| 2020 | 6.8 minutes | Pre-automation baseline |
| 2021 | 7.0 minutes | Pandemic volume surge, WFH transition |
| 2022 | 7.3 minutes | Complex contacts increasing share |
| 2023 | 7.5 minutes | Automation deflects simple contacts; agents handle harder ones |
| 2024 | 7.6 minutes | AI assist adoption partially offsets complexity increase |
| 2025 | 7.5 minutes | AI and KB maturity stabilizing AHT |
| 2026 (current) | 7.4 to 7.6 minutes | Continued AI adoption vs rising contact complexity |
Source: ICMI Contact Center Benchmarking Report 2025, with Gartner trendline estimates for 2024 to 2026.
The trend pattern is counterintuitive but well-documented: self-service and chatbot adoption has absorbed the simplest tier of contacts, so the contacts reaching live agents are harder on average. ICMI's longitudinal data shows that routine contacts - defined as issues resolvable in under 5 minutes - declined from 38 percent of voice volume in 2020 to 24 percent in 2025. The contacts that automation cannot resolve are disproportionately complex, which pulls average AHT upward even as individual agents become more efficient.
Gartner's 2025 analysis projects that AHT will stabilize in the 7 to 8 minute range for voice contacts through 2027, with marginal reductions driven by AI assist adoption offset by ongoing complexity increases. The channels most likely to see AHT declines are those where AI suggestion quality is highest: high-volume retail and transactional SaaS support, where training data is abundant and contact patterns are repetitive.
AHT and cost per contact: the direct connection
AHT is the primary driver of cost per contact in live-agent channels. Total support labor cost divided by contacts handled equals cost per contact, and contacts per agent per hour is set by AHT.
At a 7.5-minute average AHT and 80 percent occupancy on an 8-hour shift, one agent handles approximately 51 contacts per day. At a fully loaded agent cost of $35 per hour (US rates, including benefits and overhead), that agent costs $280 per day to run. Cost per contact: approximately $5.49.
Reduce AHT by 60 seconds (to 6.5 minutes) and the same agent handles 58 contacts per day. Cost per contact drops to $4.83 - an 12 percent reduction from a single minute of AHT savings. Scaled to a 100-agent team, that difference is $66 per day in labor efficiency per saved minute of AHT.
| AHT | Contacts / Agent / 8-Hour Shift (80% Occupancy) | Cost Per Contact (US Agent, $35/hr Loaded) |
|---|---|---|
| 5 minutes | 77 | $3.64 |
| 6 minutes | 64 | $4.37 |
| 7 minutes | 55 | $5.09 |
| 7.5 minutes (benchmark) | 51 | $5.49 |
| 8 minutes | 48 | $5.83 |
| 10 minutes | 38 | $7.37 |
| 12 minutes | 32 | $8.75 |
These figures are labor-only cost-per-contact estimates. For the full cost-per-ticket picture including tooling, overhead, and channel mix, see customer support staffing ratios statistics for 2026.
The cost math makes AHT reductions attractive, but the AHT-CSAT tradeoff described above applies here. Cost-per-contact figures that assume uniform AHT targets across complex and routine contacts typically undercount the repeat contact rate that aggressive AHT management generates. Gartner's 2025 analysis found that operations optimizing for minimum AHT without FCR guardrails saw repeat contact rates rise by 7 to 12 percentage points, which increased total contact volume and offset much of the per-contact cost savings.
AHT benchmarks by team structure
ICMI and Gartner both publish AHT data segmented by team size and specialization. The specialization finding is the more useful one.
| Team Structure | Average AHT | FCR Rate | Notes |
|---|---|---|---|
| Generalist team (all issues, all products) | 8.2 minutes | 67 to 72% | Higher variance; agents context-switch frequently |
| Specialist team (product or issue type dedicated) | 6.1 minutes | 76 to 82% | Faster and more accurate within domain |
| Tiered team (L1/L2/L3 routing) | 7.4 minutes (L1), 12 minutes (L2/L3) | 73 to 78% (L1) | Routing quality determines efficiency |
| Small team (10 to 20 agents) | 7.3 minutes | 74 to 78% | Coaching proximity supports quality |
| Large team (50+ agents) | 7.7 minutes | 68 to 73% | Coaching gap shows in quality metrics |
Source: ICMI Contact Center Benchmarking Report 2025.
The specialist vs generalist gap is substantial: 2.1-minute difference in AHT and 7 to 10 percentage points in FCR. Agents who handle a narrow range of issues develop faster pattern recognition and deeper product knowledge, which shows in both handle time and resolution quality. The tradeoff is scheduling flexibility - specialist teams require more predictable volume and enough scale to fill specialist queues.
What the AHT data means for operations
A few things the 2025 benchmark sources agree on:
Speed targets without quality targets backfire. ICMI, SQM Group, Gartner, and HubSpot all find the same thing: AHT targets that are not paired with FCR targets produce faster contacts with worse outcomes and higher repeat contact rates. If you set an AHT goal, set an FCR goal next to it.
The biggest AHT reductions available right now are in ACW and hold time, not talk time. Compressing talk time means rushing agents. Compressing ACW through automation and CRM integration - or reducing hold time through better knowledge base access - brings AHT down without adding pressure to the conversation. Those are the levers worth pulling first.
If you deploy nothing else from the AI category, deploy post-call summarization. The Zendesk and Talkdesk numbers show 60 to 75 percent ACW reductions with high consistency. That is capacity recovered without touching how agents handle live conversations.
Complex contacts will keep getting harder. ICMI's longitudinal data shows the simple contact tier shrinking as automation absorbs it. The headcount you need in 2027 will be handling a harder mix than the headcount you needed in 2023. AHT benchmarks will move with that shift, and staffing models that assume static AHT are going to underestimate costs.
For the staffing model implications of these AHT benchmarks, see customer support staffing ratios statistics for 2026. For the productivity picture that AHT feeds into at the agent level, see customer support agent productivity statistics for 2026.
Sources
- ICMI Contact Center Benchmarking Report 2025
- Zendesk Benchmark Report 2025
- Zendesk CX Trends 2025
- Gartner Customer Service Survey 2025
- Salesforce State of Service 2025
- HubSpot Customer Support Survey 2025
- Talkdesk Research 2025
- SQM Group Contact Center Benchmarking Study 2025
- Forrester Customer Service Technology Index 2025