Friday, 4:40 pm. A customer asks whether her parcel will arrive before her daughter's birthday. Your team finishes at 5 pm, the email stays unanswered. On Monday at 9:15 am it is the first one to be answered. For your team that feels fast: done first thing. For the customer it was around 64 hours of waiting.
Gaps like this are not bad luck but a pattern. An analysis by the helpdesk provider Gorgias shows: for the typical shop, 53 % of pre-purchase questions arrive outside Monday to Friday, 9 am to 6 pm. And the median wait for a human reply then jumps from 2.2 hours during the day to around 15 hours.
This is exactly the gap one metric makes visible: first response time. But only if you measure it honestly. Many shops look at a value in their helpdesk without knowing whether it includes acknowledgement emails, whether weekends are excluded or whether a few extremely slow tickets distort the average.
In this article you will learn what first response time measures exactly, how to calculate it so that it shows how long your customers wait, and which values per channel are realistic.
What first response time measures and what it doesn't
First response time (FRT) is the time from the arrival of a customer request to the first real reply. It answers a single question: How long do your customers wait until someone takes care of their concern?
The key word is “real”. An automatic acknowledgement such as “Thanks for your message, we'll get back to you as soon as possible” is not a reply. It only tells the customer that her email has arrived. That is why many helpdesk systems don't count such autoresponders at all. Meta also does not count instant replies and away messages for Facebook pages when it calculates response time. A real first reply, on the other hand, addresses the concern: it gives the shipping status, answers the sizing question or says specifically what happens next.
Besides FRT there are a few related metrics that are often confused:
| Metric | What it measures | Example |
|---|---|---|
| First response time | Arrival to first real reply | Email at 10:00, first reply at 11:30: 1.5 hours |
| Average reply time | Waiting time for all replies in the conversation, not just the first | Customer writes back and waits again |
| Resolution time | Arrival until the concern is resolved | Replacement delivery confirmed after two days |
| Handle time | How long your team actively works on a ticket | eight minutes of research and writing |
FRT is not the most important of these numbers, but it is the first one your customers feel. While waiting for the first reply, they decide whether to stay patient, write a second email, follow up on Instagram or claim their money back through the payment provider. Follow-ups through another channel often create another ticket. So a slow first reply doesn't just make customers unhappy, it also makes your inbox fuller.
Whether your FRT is accurate, however, depends heavily on how you measure it.
How to measure FRT properly: four decisions
Depending on how they are measured, the same tickets can result in an FRT of 40 minutes or of more than 13 hours. Four decisions make the difference.
1. When does the clock start?
The clock starts when the message arrives, not when someone on the team opens it. That sounds obvious, but it isn't: if you measure from the moment a ticket is assigned or opened, you remove exactly the waiting time that bothers customers most.
2. Business hours or calendar time?
Many tools can do both. Business hours show you how fast your team works when it is there. Calendar time shows how long your customers actually wait. The customer from Friday afternoon has an FRT of 35 minutes in business hours and of around 64 hours in calendar time. Look at both values. If you only report one, report calendar time.
3. Median instead of average
A few extremely slow tickets distort the average so much that it no longer says anything about the normal case. An example with five tickets:
| Ticket | Wait until first reply |
|---|---|
| A | 10 minutes |
| B | 25 minutes |
| C | 40 minutes |
| D | 2 hours |
| E | 64 hours (weekend) |
The average is around 13.5 hours, even though four out of five customers had a reply within two hours at most. The median, meaning the middle value when you sort all tickets by size, is 40 minutes. It describes the normal case much better.
You still shouldn't ignore the outlier, because ticket E is a real customer. So add a second number to the median: the share of tickets that wait longer than your target. In this example, with a target of four hours: 20 %.
4. Separate by channel
Chat, WhatsApp and email have completely different rhythms. If you lump everything together, fast chat replies pull the overall value down and your email waiting time disappears in it. Measure each channel on its own.
How to calculate FRT: For each ticket: time of the first real reply minus arrival of the message. For the period: median across all tickets, separated by channel. In addition: share of tickets above your target.
What is a good first response time?
There is no single correct value. There are expectations per channel and benchmarks that show you where you stand. For all benchmarks: they measure differently, and most come from helpdesk providers analysing data from their own customers.
What benchmarks show:
The helpdesk provider Freshworks analysed around 19 million tickets for its 2024 benchmark report. For the median Freshworks customer (median across customer accounts, not across tickets), the first response time across all industries was 1 hour 50 minutes, in Europe 2 hours 34 minutes and in retail 3 hours 6 minutes, each calculated in business hours. In chat and other messaging channels this value was 36 seconds, but calculated in calendar time. The data covers the period January 2023 to April 2024.
For e-commerce brands with around 10 million US dollars in revenue, Gorgias arrives at a median of 6.3 hours in a 2026 analysis, and 8.8 hours for fashion retailers of that size.
What customers expect and platforms require:
In a survey by the social media tool provider Sprout Social (September 2026), 68 % of respondents expect a reply on social media within 24 hours.
Facebook only shows the badge “Very responsive to messages” if a page has reached a response rate of 90 % and a response time of 15 minutes over the last seven days (as of October 2026). Why such signals carry more weight at Meta than many think is explained in our article on the Facebook feedback score.
Via the WhatsApp Business Platform, which helpdesk tools connect to, you can reply freely for 24 hours after a customer message. After that, only pre-approved templates are possible. So a late reply there is not just impolite but technically restricted.
From this we derive the following guideline values. They are our recommendation for shops, not a measured benchmark:
| Channel | Guideline for FRT (median) |
|---|---|
| Chat | a few minutes |
| WhatsApp, Instagram, Messenger | a few hours during your service hours, under 24 hours in calendar time |
| reply on the same business day, under 24 hours in calendar time |
If you are above that, it's no disaster. But it is worth knowing why.
Why speed matters, but isn't everything
There is evidence that a fast first reply has an effect. A study on complaints via social media shows: the faster the first and the final reply come, the more satisfied customers are with how their complaint was handled, regardless of what they actually wanted to achieve.
Speed alone isn't enough, though. Gorgias compared chat and email: in chat the first reply comes around 38 times faster, yet the share of satisfied ratings is 2.5 points lower than for email. A fast reply that doesn't solve the concern is just a fast reply.
That is why you should never look at FRT on its own. Read it together with resolution time and your customers' satisfaction. How strongly poor service affects a business overall is summarised in Why poor customer service measurably harms companies.
Five measurement errors that make your FRT look better than it is
Most shops don't measure their FRT wrongly because they want to cheat. They simply take over their tool's default setting. These are the five errors we come across most often.
Autoresponders are counted. If your tool counts the automatic acknowledgement as the first reply, your FRT is close to zero and says nothing anymore.
Only business hours are reported. Then evenings and weekends disappear. Yet according to Gorgias, for the typical shop more than half of pre-purchase questions arrive exactly then.
All channels are in one number. Fast chat replies hide slow email handling.
Replies outside the tool are missing. Anyone who replies from a private inbox or directly in the WhatsApp app often doesn't show up in the statistics. The tickets then look unanswered, or the reply time is calculated too long.
The average hides the normal case. See the calculation example above: one weekend ticket can multiply the value.
If your FRT suddenly looks better, first check whether the measurement has changed and not your service. Clear responsibilities, by the way, often help more than any tool, see Why structured customer service is crucial in e-commerce.
Lowering FRT: where AI helps and where the second clock is ticking
AI can significantly shorten the work before the first reply: reading the email, finding the order, checking the shipping status, writing the reply. But for your FRT, what counts is not when the draft is ready but when the reply reaches the customer. That is why two clocks are running.
The first clock is the draft time. With ZYBR, a reply draft is ready in under a minute, after 54 seconds at the median. If it is about an order, the draft already includes the order and shipping status.
The second clock is the sending time. With ZYBR, automatic sending is off by default. Every draft then waits in the ticket until someone on your team approves it. In this mode your FRT is as fast as your approval. For topics you have approved for automatic sending, the reply goes out with the delay you set. The default is 30 minutes to two hours, and nothing is sent between 10 pm and 6 am. Exception: if ZYBR detects warning signs such as threats or lawyers' letters, it holds the reply back as a draft. And on WhatsApp the reply always remains a draft that your team approves.
In practice we see the whole spectrum. In some shops drafts sit for longer, in others support is quite chaotic. Others have clearly defined who handles which topics when, and the replies go out quickly. Some teams only read the summary of the conversation and approve on that basis. In our observation, most of these shops reply faster than before. Those who approve topics for automatic sending reply significantly faster still.
Three levers that help with the second clock:
Approve simple topics. Questions like “Where is my order?” can often be answered completely with live data and are well suited for automatic sending.
Fixed approval times. Better three short rounds a day than “whenever someone has time”.
Plan evenings and weekends deliberately. Decide which topics may go out even then and which wait until Monday.
The 54 seconds are the median time from arrival to reply draft, measured across 3,733 drafts from late August to late September 2026. They are not a first response time.
Conclusion: measure the waiting time, not your working time
First response time is only as honest as its measurement. Start the clock on arrival, calculate in calendar time, use the median, separate the channels and leave autoresponders out. Then you see how long your customers really wait and can work on it in a targeted way.
If you want to see how fast drafts are created with your own requests: you can test ZYBR free for 14 days. The trial ends on its own after 14 days or 500 AI replies, whichever comes first, and a subscription only starts when you choose a plan.
Sources
All sources accessed in October 2026. Studies by software providers are marked as vendor studies.
Freshworks: Customer Service Benchmark Report 2024 (vendor study) https://freshworks.com/assets/resources/Customer_Service_Benchmark_Report_2024.pdf
Gorgias Ecom Lab: Stop benchmarking against the average (1 April 2026, vendor study) https://www.gorgias.com/research/stop-benchmarking-against-the-average
Gorgias Ecom Lab: Your team is sleeping on 53% of shopper questions (30 July 2026, vendor study) https://www.gorgias.com/research/your-team-is-sleeping-on-53-of-shoppers
Gorgias Ecom Lab: Email isn't dying, chat is doubling (19 May 2026, vendor study) https://www.gorgias.com/research/email-isnt-dying-chat-is-doubling-but-its-not-that-simple
Sprout Social: The state of social media in 2026 (23 September 2026, vendor study) https://sproutsocial.com/insights/the-state-of-social-media/
Istanbulluoglu, D. (2017): Complaint handling on social media: The impact of multiple response times on consumer satisfaction, Computers in Human Behavior 74, pp. 72-82, DOI 10.1016/j.chb.2017.04.016
Facebook Help Center: badge “Very responsive to messages” https://www.facebook.com/help/994476827272050
Facebook Help Center: Response rate and response time https://www.facebook.com/help/1625715647696553
Meta for Developers: WhatsApp Cloud API, Service messages https://developers.facebook.com/documentation/business-messaging/whatsapp/messages/send-messages
Zendesk: Understanding ticket reply time https://support.zendesk.com/hc/en-us/articles/4408821871642-Understanding-ticket-reply-time
Paul Lory, Founding Partner and Technical Advisor
Paul is responsible for stability, security and efficiency in day-to-day operations. With an eye for risk and performance, he makes sure the platform stays lean, secure and fast over the long term.
