How long does a good support response take — and how long should it? The 2026 benchmark

What FRT is, and why it still matters

First Response Time (FRT) is the time between a customer raising a request and the first substantive answer from an agent or from AI. Automatic acknowledgements do not count — the customer is still waiting for the case to be solved, not for a receipt.

FRT is the most closely tracked customer service KPI, because it shapes the first impression and feeds straight into CSAT. According to the Zendesk CX Trends 2026 report, 63% of customers rank response speed as the number one factor in service — ahead of resolution speed (57%) and channel availability (49%).

Response time benchmarks for 2026, by channel

There is no single good response time — it depends entirely on the channel. A customer will accept four hours of silence by email, but will walk away from a chat after three minutes. The industry benchmarks for 2026 are below (sources: Zendesk Benchmark, HubSpot State of Service, Lorikeet CX, GreetNow):

Email — the widest gap between expectation and reality

The average company replies to email in 8.5 hours. Top performers come in under one hour. A B2C customer will wait patiently for four — after that comes frustration, and the risk they abandon the case altogether. 62% of customers abandon the request if they get no answer within 48 hours.

Chat — the harshest threshold of all

Chat has the least tolerance of any channel. The benchmark is 40 seconds, and satisfaction falls in a straight line from there: every extra minute of waiting costs 2–3 CSAT points. After three minutes 53% of users close the window — and they do not switch to email, they switch to a competitor.

Phone — the 80/20 rule and the ceiling of patience

The classic call centre benchmark is 80% of calls answered within 20 seconds. Most companies miss it: the average wait in 2026 is 5 minutes 12 seconds, while customers tolerate at most 2 minutes 37 seconds. Beyond that, 34% of calls end in a hang-up.

Social media — where the risk is public

A customer writing to a company on X, Facebook or LinkedIn expects an answer within the hour. Companies take four to five hours on average. Crucially, silence beyond four hours correlates with public escalation — the customer starts posting, tagging the company, or leaving negative reviews elsewhere.

Customer expectations versus reality

The trend that matters most: 88% of customers expect faster answers than they did a year ago. What passed as acceptable in 2024 no longer does. The standard moves whether or not your internal operations keep up.

B2B is no longer just a slower B2C

The old assumption that business customers will put up with slower service no longer holds. 67% of B2B buyers say they have switched supplier over slow response times. Buyers under 40 expect a reply within the hour, whether they are shopping for themselves or for the company.

What slow support actually costs

A slow response is not merely a soft quality problem. It has a measurable financial cost:

• Lost sales leads: getting back within five minutes makes conversion 21 times likelier than getting back after thirty (source: InsideSales/MIT).
• Customer churn: 52% of customers leave a company after a slow support experience. For a SaaS business with 1 m PLN of ARR, 10% churn from that cause means roughly 100 000 PLN gone every year.
• Public escalation: a customer left waiting four hours on social media starts posting — and negative opinion travels further than positive.
• The hidden cost to your team: a customer kept waiting sends follow-ups asking whether anyone is there, and the backlog grows. Every ticket then takes longer to process.

FRT is not everything: the empty-reply trap

The commonest mistake when chasing FRT benchmarks: the agent fires off an empty reply saying they will look into it, so the metric turns green. On paper it looks excellent. For the customer nothing has changed — they are still waiting for a real answer, only now with less patience, because they have officially been seen.

This is why more and more teams split the metric in three:

• Time to acknowledgement — when the customer knows somebody is reading.
• Time to first meaningful response — when they receive something of substance.
• Time to resolution — when the case is closed.

The three say entirely different things about a team. The first can be fully automated. The second demands an understanding of context. The third often lies outside support altogether — waiting on the product, a supplier, the warehouse.

Five levers that genuinely shorten FRT

Counter-intuitively, FRT is above all a queue and routing problem, not a typing-speed problem. According to Lorikeet CX, optimising the queue and adding AI triage improves FRT five to ten times more than optimising how agents work. Here are five levers, in order of impact:

1. Routing by category and skill

Billing questions go to the billing person, technical ones to the technical person. Without that sorting, every ticket passes through a general queue and needs a handoff, which eats anything from tens of minutes to hours.

2. SLAs and priorities tied to the type of customer

An enterprise customer with a 24-hour SLA should not be queueing behind an anonymous free-trial ticket. Obvious — and still not implemented in plenty of companies.

3. A knowledge base and response templates

Seventy per cent of tickets are repeat questions. A good knowledge base plus snippets and macros cuts response time from eight minutes to ninety seconds. There is a second benefit: consistency — different agents stop answering the same question in different words.

4. AI at first contact: deflection and drafting

AI resolves 60–80% of routine questions in under three seconds. For tickets that do need an agent, it drafts a reply from the customer's history and the knowledge base. According to the Freshworks 2025 report, AI cuts average FRT from over six hours to under four minutes. More on this in: customer service in 2026 — ticketing, voicebots and AI.

5. Staffing by the data, not by intuition

Look at ticket volume hour by hour. Most SaaS teams peak between 10:00 and 14:00, Monday to Thursday. Adding one agent inside that window improves FRT more than adding one for the whole day.

Setting SLAs and priorities in practice

A good SLA is not a single number. It should have layers, matched to the customer and the urgency of the case:

The key to making it work: every request gets one priority, assigned automatically by rules (customer, category, keywords), and an agent may only override it with a documented reason. Without that, priorities become a matter of opinion and everything ends up as a P1.

When AI actually helps — and when it does not

AI in customer service is not magic. It is a specific tool with a specific range. It works superbly in three scenarios:

• Deflecting simple questions — where the parcel is, how to reset a password, what the opening hours are. With access to a knowledge base, AI answers correctly 60–80% of the time, in under three seconds.
• Triage and routing — AI reads the request, classifies it by category, priority and language, and sends it to the right person. That removes the dead time between the request arriving and an agent seeing it.
• Drafting a reply for the agent — AI produces a draft from the customer's history and similar cases. The agent corrects it and sends, cutting writing time from five minutes to one.

AI not work well in these cases:

• Cases that need empathy and emotional context — lost luggage, a medical error.
• Commercial negotiation, and decisions on discounts or refunds outside the standard procedure.
• Legal escalation, GDPR, and complaints seen from a compliance angle.

What to measure weekly, so you stop guessing

Without measurement, every lever above is just theory. The minimum set of metrics worth reviewing weekly:

• Median FRT (not the mean — outliers inflate it)
• 90th-percentile FRT — it shows how badly the tail is served
• FRT per channel — email, chat, phone, social
• Time to first meaningful response — excluding auto-confirmations
• Resolution time per request category
• CSAT after resolution plus the written comment
• Ticket volume by hour — to find the peak hours
• Recurring topics — the ten biggest categories each week

Those eight numbers on a weekly dashboard tell you exactly what to work on next. 82% of service leaders who review FRT weekly report year-on-year gains in both speed and satisfaction (Salesforce State of Service 2025).

All of these can be configured in the Debesis helpdesk on a single dashboard, with threshold alerts.

Summary and checklist

Ile should a good support response take in 2026? It depends on the channel:

• Email — under 4 hours (ideally under 1)
• Chat — under 40 seconds (satisfaction collapses at 3 minutes)
• Phone — under 20 seconds (patience runs out at 2 min 37 s)
• Social media — under 1 hour (past 4, you risk a public escalation)

Ile actually does the average response take? Usually two to three times longer than the benchmark. The good news: the gap between the industry average and the top 5% is eight hours on email and ninety seconds on chat — and both are distances you can genuinely close in 60 to 90 days.