Average handle time (AHT) is the average time an agent spends on one customer interaction, including talk time, hold time and after-call work. It is one of the most tracked support metrics, and one of the most misused.
Support teams on Shopify stores and in call centers alike treat AHT as a target and push agents to close conversations faster. But speed without resolution brings the customer back, so total handling time goes up.
This guide explains what AHT measures, how to calculate it and what a good AHT looks like by industry. It then gives seven ways to lower it by removing friction, not by pressuring agents.
If you run a Shopify store, one way to lower handle time is to let AI take the routine chats. Your team then gets only the conversations that need a person. That is where Chatty fits.
Chatty is the AI sales agent for Shopify stores. AI chatbot and live chat are two standout features of Chatty. The AI chatbot answers from your own Shopify data, and with live chat your team takes over from the AI and closes the sale.
92% of product answers were right on price, stock and details when we compared 300 real Chatty answers with each store's website in October 2026.
Accuracy matters for handle time too, because a wrong AI answer turns into a longer conversation for your team later.
- Pushing agents to go faster often increases total handle time, not cuts it.
When agents skip steps to close conversations quickly, customers come back with the same problem. That creates more contacts and pushes total handling time up.
- AHT is a measure of effort spent, not a measure of resolution quality.
A short handle time does not confirm the issue was solved well. A long one does not mean something went wrong, because the metric captures time, not outcome.
- After-call work is the hidden inflation source in average handle time calculations.
AHT combines talk time, hold time and after-call work such as notes and tagging. If agents skip wrap-up fields, the numbers in your system will not reflect reality.
- Rising AHT usually points to a process or knowledge problem, not agent speed.
A gradual rise over a week usually follows a confusing product change, an unclear policy or a missing knowledge base article. Look there before you coach agents to go faster.
- Customers give up quickly, so a rising AHT turns into lost contacts.
Call Centre Helper reports that most customers who abandon a call do so within 30 to 60 seconds. When AHT rises and staffing stays the same, queues grow and more of those customers leave.
Overview of average handle time
Before you try to lower AHT, be clear on what the number includes and how your tools calculate it.
What is average handle time (AHT)?
Average handle time covers an interaction from the moment it starts until the agent is ready for the next one. It is easy to measure and easy to misread.
AHT measures handling effort, not quality. A short handle time doesn't mean the issue was solved well, and a long one doesn't mean something went wrong.
Many teams still treat a lower AHT as better by default. When you push agents to close conversations faster, they skip steps: clarifying the issue, confirming the fix or answering the related question. The customer comes back with the same problem, and total handling time goes up.
So the goal is not the lowest possible AHT. It is the right balance between speed and resolution quality for your business and your customers.
How do you calculate it?
Average handle time is made up of three parts:
- Talk time: the conversation with the customer.
- Hold time: any time the customer waits while the agent researches or consults.
- After-call work (ACW): notes, tagging, follow-ups and system updates after the conversation ends.
Put the pieces together and the formula is simple:
AHT = (total talk time + total hold time + total after-call work) ÷ number of interactions. For example, take 40 hours of talk time, 5 hours of hold and 5 hours of after-call work across 500 conversations. That is 3,000 minutes ÷ 500, an AHT of 6 minutes.
The tricky part is not the math. It is capturing each part consistently. If your system doesn't log hold time separately, or agents skip wrap-up fields, your AHT will not reflect reality.
So audit how your tools calculate AHT before you use it as a performance benchmark. The metric is only as good as its inputs.
Why average handle time matters to support teams
AHT touches three things a support lead answers for: cost, wait time and early warning of problems.
Operational efficiency and cost control
AHT sets how many conversations your team can handle. A lower AHT means more capacity without adding headcount.
That matters because people are the biggest cost. NICE puts labor at about 60% to 70% of a contact center's annual cost, covering wages, training and attrition. That is why AHT sits under staffing models, hiring plans and budget forecasts.
Customer wait time and availability
When AHT rises and staffing stays the same, queue times rise with it. Customers do not wait long. Call Centre Helper reports that most customers who abandon a call do so within 30 to 60 seconds.
In a 2017 Arise survey reported by ICMI, almost two-thirds of customers said they would wait two minutes or less on hold. Frustrated customers make conversations harder, which pushes AHT higher again. Keeping AHT stable breaks this cycle.
Early signal of friction
Rising AHT usually points upstream, to confusing product changes, unclear policies or missing answers in the knowledge base. It is rarely about agent speed.
Track trends, not spikes. A gradual increase over a week signals a systemic issue worth investigating.
Resolution also matters more than a few seconds of speed. SQM Group links every 1% gain in first call resolution to a 1% gain in customer satisfaction.
The same Call Centre Helper article cites one organization that found almost no difference in dissatisfaction between a 20-second and a 60-second wait. Used this way, AHT becomes a diagnostic tool, not a scorecard.
7 proven ways to reduce average handle time without hurting quality
The seven steps below run in order, from finding where time goes to measuring the result.
1. Apply AI across the support workflow
A lot of handle time goes to context switching. Agents jump between tabs to search the knowledge base, copy order numbers and draft replies from scratch. Each switch adds seconds across hundreds of tickets.
AI keeps agents in one place. Instead of searching, they see the relevant article. Instead of typing from scratch, they edit a suggested reply.
An NBER study of one company's support agents measured the effect. An AI assistant raised issues resolved per hour by 14% on average, and by 34% for newer, less skilled agents. Google Cloud says its Agent Assist helps service reps handle 28% more conversations, though that is the vendor's own figure.
The gain comes from putting AI inside live work, not from a chatbot that only deflects easy questions. Here's how to start:
- Audit where agents spend the most time first. Searching, writing or switching tools? The answer decides which AI capability to prioritize.
- Pick a tool that fits your platform. For enterprise help desks, start with the AI add-ons in your existing ticketing system.
- Start narrow. Turn on AI for your top 3 issue types only, and track resolution rate and CSAT before expanding, because accuracy builds trust.
- Scale gradually. Once AI handles routine questions reliably, move agents to the complex cases where human judgment matters.
On a Shopify store, this is what it looks like in Chatty.
When the AI hands a chat to your team, it adds a summary of the conversation. The agent reads the issue instead of asking the shopper to repeat it. In the inbox, Chat summary and Improve writing draft a summary or a cleaner reply that the agent edits before sending.
2. Diagnose AHT by components before changing anything
Before you optimize, find out what is actually slow. AHT has three parts (talk time, hold time and after-call work), and each one points to a different problem.
High hold time in billing usually means slow system lookups. Long wrap-up time in returns often means too many required fields. Each needs a different fix, so generic "speed up" guidance does not help.
Here's how to diagnose properly:
- Export AHT data broken down by talk time, hold time and after-call work.
- Segment by issue type (billing, returns, technical) and channel (phone, chat, email).
- Flag any component running 2x or more above average. That is your first target.
- Wait to message agents until you can tell them exactly which part is slow and why.
3. Redesign workflows to remove step friction
Once you know where time goes, redesign the workflow. Most AHT problems come from unnecessary steps, not slow agents.
Extra verification, low-value approvals and unclear escalation paths add minutes to every interaction. Training agents to move faster through a slow process only goes so far.
Redesign the workflow in four moves:
- Map your top 5 issue types step by step and time each stage. Find the steps that take a disproportionate share of the time.
- Remove verification steps that repeat information you already have (for example, reconfirming the email after login).
- Set approval thresholds: auto-approve refunds under $50, standard exchanges and routine credits. Keep human review for edge cases.
- Create a one-page escalation guide so agents don't waste time figuring out who handles what.
4. Improve first-pass diagnosis to prevent backtracking
Even with simpler workflows, agents lose time when they misdiagnose an issue early. They start down the wrong path, notice halfway through, then backtrack by asking again, switching workflows or transferring the customer.
This happens more than most teams realize. SQM Group puts the call center industry average first contact resolution rate at 71%. That means 29% of customers have to call back about the same problem, often because the first interaction missed something.
The fix is to build diagnosis into the workflow itself:
- Create a 3 to 4 question intake: What's the issue? When did it start? What have you tried? What outcome do you need?
- Tag the issue type within the first 30 seconds of every conversation, before agents start solving.
- Implement skills-based routing so billing issues go to billing-trained agents, not a general queue.
- Review transfers weekly. Each transfer is a missed diagnosis, so track the patterns.
5. Strengthen knowledge so agents find answers fast
Diagnosis gets agents to the right issue. They still need to find the right answer, and that is where knowledge access matters.
When agents stop trusting the knowledge base, they ping colleagues on Slack or escalate to be safe. Both add minutes.
A knowledge base agents trust can improve team productivity. Agents only reach for knowledge bases they trust, and trust comes from accuracy and freshness.
Here's how to make knowledge usable:
- Assign an owner to each of your top 20 articles, someone accountable for accuracy.
- Set refresh cycles: policies monthly, product info quarterly, troubleshooting guides after each release.
- Restructure articles for scanning mid-conversation: the answer in the first line, steps in bullets, edge cases at the bottom.
- Track article usage. Low traffic on articles about common issues means agents can't find them.
6. Reduce after-call work with standardization first, then automation
After the conversation ends, agents still have notes, tags and follow-ups to do. This work varies widely: some agents finish in 30 seconds, others take two minutes.
The difference is usually format, not speed. With a shared structure, everyone works the same way.
It is tempting to jump straight to automation such as auto-tagging and AI summaries. But automation works best with consistent inputs, so standardize first.
Here's the sequence that works:
- Define what a good note includes: issue type, root cause, resolution and whether follow-up is needed.
- Build templates for your top 10 issue types with pre-filled fields agents can select.
- Limit free-text notes to exceptions, and use dropdowns and checkboxes for common scenarios.
- Add auto-tagging and summary generation once 80% or more of notes follow the standard format.
In Chatty, auto-tagging labels each conversation by topic from your own tag list. It runs when the conversation is resolved or transferred to a person.
7. Set segmented targets and manage with trends
How you measure AHT matters as much as how you reduce it. A single target across all issue types creates the wrong incentives.
Password resets and billing disputes need different handling, and holding both to the same number invites gaming. Agents rush complex cases or avoid hard queues to hit it.
The fix is segmentation and trend-based management:
- Group issues by complexity: simple (password reset, order status), moderate (returns, billing questions) and complex (disputes, technical troubleshooting).
- Set separate AHT targets for each tier and channel. A 3-minute target makes sense for order status, not for technical troubleshooting.
- Don't rely on averages alone. Most interactions may be quick while a few hard cases cause most of the delay. Look at both the typical handle time and the slowest cases.
- Focus coaching on segments with consistently high AHT, not on individuals who had one bad day.
What is a good average handle time? (Industry benchmarks)
There's no universal answer. AHT depends on what your team handles, how complex the issues are and how much verification or diagnosis each one needs. Zendesk puts a good AHT across industries at around six minutes.
The ranges below are a directional guide, not an audited benchmark. Published estimates differ, so compare your own trend over time first.
Retail and e-commerce
A typical range is 4 to 6 minutes, and Zendesk's estimate for retail is lower, at 3 to 4. Most volume comes from order status, returns and shipping questions, which have clear answers and well-defined workflows.
AHT rises when product launches create edge cases or policies change mid-season. It drops when self-service handles the easy questions and agents focus on exceptions. If your AHT stays above this range, check whether agents are waiting on system lookups or working through too many steps.
Banking and financial services
Teams typically see 4 to 7 minutes, and Zendesk puts banking at 4 to 6. Security verification adds time upfront, and compliance means agents can't skip steps. Account issues often need several systems checked, which extends hold time.
Accuracy matters more than speed here. A rushed interaction that leads to a disputed transaction or a compliance issue costs far more than an extra minute of verification.
A longer AHT is acceptable when it protects the customer and the institution. It becomes a problem when agents wait on slow internal tools or escalation paths are unclear.
IT and technical support
Handle times usually run from 7 to 10 minutes, and Zendesk's estimate is 8 to 10. Troubleshooting drives the length: agents need to diagnose before they can resolve, and that takes back-and-forth. Complex issues may need remote access or a handoff to a specialist.
A longer AHT is expected when it leads to first contact resolution. It becomes a concern when agents repeat diagnostics that should have been captured earlier. The same goes for escalating issues better documentation would have solved.
Final thought
Average handle time works best as a signal, not a target. It shows where the support experience feels heavy or slow, and it is not a number to push agents to chase.
Start by understanding what is behind the number. Break AHT down by talk time, hold time and after-call work, then look at it again by issue type. That shows where time is actually lost.
From there, the fixes get clearer. Add AI where it removes friction and simplify workflows that have grown too complex. Keep knowledge easy to find, and diagnose the issue right the first time.
Keep resolution quality in view the whole way. When the work gets easier and issues are solved the first time, handle time takes care of itself.
If your team answers shoppers on a Shopify store, start with the routine chats. Let AI take those and hand the rest to your team with the summary already written.
FAQs
Not always. Lowering AHT makes sense when you remove unnecessary steps, improve knowledge access or cut waits on slow systems. In those cases, a shorter handle time is a real efficiency gain.
Pushing AHT down without fixing the cause backfires. Agents rush verification, skip follow-up questions or close tickets before the issue is solved. Repeat contacts rise, and total handling time per issue goes up.
Ask where time is wasted and where it is well spent. A billing dispute solved for good in 8 minutes is worth more than a 4-minute call that leads to a callback.
Not necessarily. AHT varies by channel, issue type and customer context.
A technical support queue runs longer than a password reset queue. A financial services team with heavy compliance spends more time on verification than a retail team handling shipping questions.
It varies a lot by channel:
- Phone and video support tend to have the highest AHT. Conversations happen in real time, agents handle one at a time and customers expect an immediate back-and-forth.
- Chat sits in the middle. An agent can run a few chats at once, but each still needs quick replies.
- Email and other async channels usually show a lower handle time per interaction, though total resolution time can be longer.




