Hivenza

First response time and resolution time: how to measure and improve them

10 min readUpdated November 16, 2026

The short answer

First response time is how long a customer waits for the first real human reply. Resolution time is how long it takes from their first message until the problem is solved. Measure both as medians, in business hours if you have opening hours, and set targets per channel from your own history.

A customer writes to your garage on Monday at 4:50 p.m. to ask whether the car will be ready on Wednesday. You answer on Tuesday at 9:10. Was that fast? It depends on whether you count the night. And when does the story end: when you replied, or when the car was handed back?

Those two questions are the whole subject of this article. First response time and resolution time are the two oldest numbers in customer support, and they look simple until you try to put them in a report. The way you count changes the result more than the way you work does.

What the two numbers mean

First response time (FRT) is the time between the moment a customer reaches you and the moment a person on your team sends the first reply that actually deals with them. It is the same idea as response time in the glossary.

Resolution time is the time between that same first message and the moment the case is solved. Help desks report them as two separate clocks, because the two numbers answer different questions:

  • First response time answers: did someone notice me? It is about attention and reassurance.
  • Resolution time answers: did anyone fix it, and how long did I wait? It is about the outcome.

A team can be excellent at one and poor at the other. A bakery that answers every order question within minutes but takes a week to refund a wrong delivery has a fine FRT and a bad resolution time. A repair shop that says nothing for a day and then fixes everything in one visit has the opposite pattern.

Decide what counts as a first response

Before you measure, write down three rules. They are boring, and they decide whether the number means anything.

  1. Does an automatic reply count? Most teams say no. “Thanks, we got your message” tells the customer nothing. Some help desks do count it unless you exclude it, so check how your tool behaves.
  2. Does a reply from the AI agent or a chatbot count? If it resolved the question, it is a resolution. If it only greeted the person, it is not a first response from your team.
  3. Does a note to yourself count? Never. Only messages the customer can read.

In Hivenza, the response time for a conversation runs from the customer’s first message in the period to the first reply from your team after it. Conversations nobody has answered yet have no response time, so they stay out of the median instead of dragging it around. They show up in the backlog instead, which is where they belong.

Business hours or calendar hours

This is the choice that changes your numbers the most.

Calendar time counts every hour, including nights, weekends and holidays. It is what the customer lives through. A message sent on Friday evening and answered on Monday morning took about 60 calendar hours.

Business hours only count the time you are open. The same message took perhaps one hour, if you open at 9 on Monday and answer at 10. This is what your team can fairly be held to.

Both are legitimate. They serve different purposes:

Calendar hoursBusiness hours
Best forWhat the customer experiencesJudging the team’s own speed
Weekend messageCounts the whole weekendCounts only open time
Typical usePublic promises (“we reply within a day”)Internal targets and shift planning
RiskPunishes a team for being closedHides how long customers really wait

A sensible approach is to use business hours for the team’s targets and to keep an eye on calendar time as well, so you notice if a Friday-evening problem regularly sits for three days. If you have service hours that differ by day, or closed days, write them in your tool once so the clock pauses by itself. An example of how the pause works: with opening hours from 9:00 to 18:00 and a 15-minute target, a message that arrives at 17:50 has used 10 minutes by closing time, so it is due at 9:05 the next morning.

Hivenza works the same way. Its service targets (Pro plan) count first reply, next reply and resolution in your business hours and holidays, while the resolution figure in the support analytics is counted on the calendar, from creation to solve. Both views exist on purpose: one for what you promise, one for what customers feel.

Use the median, then look at the slow end

Say you answered five messages in 2, 3, 4, 5 and 300 minutes. The average is 62.8 minutes. The median (the middle value) is 4 minutes. Neither is wrong, but only one describes a typical customer. The average is pulled up by one bad case, which may be a message that arrived at 2 a.m. or a customer who forgot to reply to your question.

So report three things:

  • The median (the 50th percentile): half of customers wait less than this.
  • The 90th percentile: nine customers in ten wait less than this. It shows the slow tail your average hides.
  • The count of unanswered conversations right now, since they have no time yet.

The support analytics in Hivenza show the median and the 90th percentile for resolution, and the backlog split by how long tickets have been open. You read them for the whole desk from the Starter plan, which brings tickets, and for each team and each teammate from Pro; see support analytics.

Realistic targets: what the public numbers say

Everybody wants a benchmark, and the honest answer is that public ones are weak. Vendor blogs quote very different figures, often without saying what was counted, which channel, which industry, or whether the clock paused outside opening hours. Treat any single figure you read as a rough rule of thumb, never as a standard, and certainly not as a promise to your customers.

What the benchmarks do agree on is the order. Live chat and messaging are expected to be fast, in minutes. Social comments and direct messages come next. Email is slower, and phone calls are immediate by nature. A customer on WhatsApp who waits a day feels ignored; a customer who emails and waits a day often does not.

So set your own targets from your own history, in four moves:

  1. Export three or four weeks of conversations and read the median and the 90th percentile per channel.
  2. Pick a target slightly better than your median for the first reply, and a looser one for the 90th percentile. Aim for something you could meet nine days in ten.
  3. Write it in plain words that customers can read: “We answer messages within 4 working hours, Monday to Friday.”
  4. Review it after a month. If you meet it every day, tighten it. If you miss it every day, either add capacity or loosen it, and say so.

Priority matters too. A customer who cannot log in to the software they pay for should not queue behind a general question. A common layout gives the top priority a first reply within one business hour and the lowest within a working day, with matching resolution targets. The exact numbers are yours to choose; the layering is the point. More on that in support SLAs explained.

How to bring first response time down

Most delays are not about typing speed. They come from messages sitting unseen or unowned. Start with the causes before you buy anything.

  • One place to look. A shared inbox with every channel in it removes the “I thought you had it” delay.
  • One owner per conversation. Assign it on arrival, by hand or by rules, so the clock has a name next to it.
  • Saved replies for repeat questions. Opening hours, delivery costs, how to reset a password. See saved replies and canned responses.
  • An honest away message. It does not change the measured time, but it changes the customer’s patience. See after-hours support.
  • Cover the busy hours. Look at when messages arrive, not when your people are free. A school receives most parent messages before 9 and after 5; a clinic on Monday mornings.
  • Alerts before a breach. A target that only gets reported at month end changes nothing. One that warns the team an hour before the deadline does.

How to bring resolution time down

Resolution time is longer and has more moving parts, so it is more about process than about speed.

  • Solve at the first contact when you can. Give the agent the information and the permission to decide (a refund up to a limit, a replacement, a booking change) so the case does not need a second person.
  • Ask for everything at once. One message with the four details you need beats four exchanges over four days.
  • Route by topic. A billing question should land with the person who can touch billing, not travel through three people.
  • Track what waits on other people. Technical faults, suppliers and approvals have their own owners and need their own follow-up.
  • Count reopened tickets. A ticket closed in an hour and reopened twice has not been resolved. If many reopen, the speed is an illusion.
  • Read the reason when a customer is unhappy. A rating with a reason attached tells you which step to fix.

That last point links the two worlds. Speed means little without the customer’s view, so pair these numbers with ratings. Hivenza asks the customer to rate a solved ticket, on the channel they used, from the Starter plan; see customer ratings and the surveys explained in this blog.

Common mistakes

  • Rewarding speed alone. If agents are judged on FRT only, you get a quick “Hello, we are looking into it” and then silence. Judge it together with resolution and ratings.
  • Mixing channels in one average. Chat and email have different expectations. Report them separately, then together.
  • Changing the definition quietly. If you change what counts, say so in the report; otherwise the improvement is only on paper.
  • Forgetting the unanswered. A median over answered conversations looks lovely while twenty messages wait. Show the backlog beside it.
  • Setting a target nobody can meet. A target missed all the time teaches the team to ignore it.

For the bigger picture, including how to read these two numbers beside volume, backlog and satisfaction, read support team metrics for managers.

Frequently asked questions

What is a good first response time?

There is no single good figure, because it depends on the channel, the industry and what you promise. Chat and messaging are expected within minutes, email within hours. Start from your own median per channel and aim to beat it by a margin you can keep.

What is the difference between first response time and resolution time?

First response time is the wait until the first real reply from a person. Resolution time is the wait until the problem is solved. The first measures attention, the second measures the outcome.

Should I use the average or the median?

Use the median as the headline, because a few extreme cases distort the average. Add the 90th percentile to see how long your slowest customers wait.

Should I count business hours or calendar hours?

Use business hours for your team’s targets, since nobody can answer while closed, and keep calendar time visible because it is what the customer feels. Pause the clock while you wait for the customer.

Does an automatic reply count as a first response?

In most reports it should not. An acknowledgement tells the customer nothing about their problem. Count the first reply from a person, or from the AI agent only if it actually answers the question.

Keep reading

All articles

Put the guide to work on your next week of posts

A golden honeycomb holding small photographs of social media posts