Free tool First response time calculator with SLA hit rate
First response time is how long a customer waits for the first reply from a person. Paste the times from your chat or helpdesk export and see the average, the median, the slow tail and how many replies met your SLA — the numbers a single average hides.
- Average, median, P90
- Two SLA targets
- Runs locally
Measure your first response time
The sample is 32 chats from one afternoon, in mixed formats, with two that never got a reply. Paste your own list or switch to summary counts.
How to read values with no unit, such as 45.
One per line or comma-separated. Accepts 45, 1:30, 1:02:03, 2m, 1h 5m. Write - or unanswered for conversations with no reply.
All conversations in the period.
Count from your report.
Seconds or m:ss, e.g. 0:52.
e.g. 60 seconds or 1:00.
A looser second target, e.g. 5:00.
Average (m:ss)
1:19
Conversations with a reply.
Median (m:ss)
0:41
Half replied faster.
P90 (m:ss)
2:23
90% replied within this.
Within target 1
73.3%
Of replied conversations.
Within target 2
86.7%
Of replied conversations.
Within target 1, strict
68.8%
Unanswered counted as missed.
Replied
30
Unanswered
2
Distribution
Each range includes its upper limit, so "1–2 min" means over 1 and up to 2 minutes.
| Response time | Conversations | Share | Cumulative |
|---|---|---|---|
| ≤ 30 s | 9 | 30.0% | 30.0% |
| 30 s – 1 min | 13 | 43.3% | 73.3% |
| 1–2 min | 4 | 13.3% | 86.7% |
| 2–5 min | 2 | 6.7% | 93.3% |
| 5–15 min | 2 | 6.7% | 100.0% |
Share these inputs: the values go after the # in the link, so they never reach a server.
https://tapcub.com/tools/first-response-time-calculator
Version v1.0Updated Update history
Why the average is not enough
Response times have a long tail: most replies are quick, a few are very slow. Each number describes a different part of that shape.
Median
The typical wait. In the sample it is 40.5 seconds: half of the customers had a reply within that. It barely moves when one chat waits ten minutes.
Average
Pulled up by the slow tail. The sample's average is 1 min 19 s, almost twice the median, because of a 9:45 and a 6:10. A rising average with a flat median means the tail is getting worse.
P90
The wait that 90% of customers beat — the "bad day" for one customer in ten. It is the number that shows staffing gaps first.
Within target
The SLA hit rate. The normal version divides by conversations with a reply; the strict version counts unanswered conversations as missed, which is closer to what customers experienced.
Calendar time or business hours? A chat that arrives at 23:50 and is answered at 09:05 waited over nine hours of calendar time but five minutes of business hours. Decide which one your SLA uses, and convert before pasting if needed.
Three ways to use it
The calculator works with raw times or with totals from a report.
-
1
Paste an export
Copy the first response column from your chat or helpdesk export. Mixed formats are fine; unreadable lines are skipped and reported.
-
2
Set one or two targets
Use the target you promise customers, and optionally a looser one to see how far the misses run over.
-
3
Use report totals
If you only have counts — conversations, replied, replied within target — switch to summary counts for the hit rate.
-
4
Compare periods
Run this week and last week, or each hour of the day, and compare the median and P90 rather than the average.
Definitions used here
The same definition as TapCub's first response time: time from the start of a conversation to the first reply from a person.
- Average and percentiles use conversations with a reply; unanswered ones are counted separately
- Percentiles use linear interpolation (the same as Excel PERCENTILE.INC and NumPy's default)
- Automatic greetings and bot messages should not count as a first response
- Within target means less than or equal to the target
Average
average FRT = Σ first response times ÷ conversations with a reply
2,369 s ÷ 30 = 79.0 s = 1 min 19 s
Percentile
P = x₍ₖ₎ + (h − k)·(x₍ₖ₊₁₎ − x₍ₖ₎), h = (n − 1)·p
P90 of 30 values: h = 26.1 → 142.5 s
SLA hit rate
within target = replies ≤ target ÷ conversations with a reply
22 ÷ 30 = 73.3%
Strict hit rate
replies ≤ target ÷ (replied + unanswered)
22 ÷ 32 = 68.8%
Two weeks with the same average
Two weeks of chat on the same site, both averaging 58 seconds. Sample figures.
| Week | Average | Median | P90 | Within 1 min |
|---|---|---|---|---|
| Week 1 | 58 s | 49 s | 1 min 40 s | 64% |
| Week 2 | 58 s | 31 s | 2 min 55 s | 78% |
Week 2 looks better on the median and the SLA hit rate, but its P90 is much worse: most customers got faster replies while one in ten waited almost three minutes — typically a staffing gap at one time of day.
Where first response numbers mislead
Most disagreements about FRT come from what is counted, not how.
Counting automatic greetings
An auto-reply sent to every visitor makes FRT look instant. Measure the first reply from a person.
Reporting only the average
A few slow conversations dominate the average. Report the median and P90 next to it.
Dropping unanswered conversations
Leaving out chats nobody answered makes the SLA look better than customers experienced. Show the strict rate too.
Mixing calendar and business hours
Overnight waits inflate calendar-time FRT. Pick one convention and use it every time.
First response time, measured for you
TapCub's support report calculates this from every conversation, so you do not have to export and paste.
Average and median
First response from conversation start to the first agent reply, with the trend over time and the average per agent.
Learn moreResolution rate next to it
Resolved conversations ÷ all conversations, compared with the previous period.
Learn moreAI replies counted separately
Conversations the AI assistant handled alone are counted as AI-handled and AI-resolved, so they do not hide slow human replies.
Learn moreFirst response time questions
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
What is first response time?
The time between a customer starting a conversation and the first reply from a person. It is usually reported as an average across conversations with a reply; the median is a better picture of the typical wait.
How do I calculate average first response time?
Add up the first response times of all conversations that got a reply and divide by how many there were. 30 replies totalling 2,369 seconds give 79 seconds, or 1 min 19 s.
Should I use the average or the median?
Both. The median tells you what a typical customer experienced; the average and P90 show the slow tail. If the average rises while the median stays flat, a few conversations are waiting much longer.
What is a good first response time for live chat?
There is no single standard. Many teams aim for under a minute or two during office hours. Track your own median and P90 by hour of day and compare them with your SLA.
Do I count conversations nobody answered?
They have no response time, so they cannot be averaged, but they are the worst possible outcome. This calculator reports them separately and gives a strict SLA rate that counts them as missed.
Business hours or calendar time?
Decide based on your promise. If you say "we reply within an hour during office hours", subtract closed hours first. The time zone converter helps when customers and agents are in different zones.
How do I bring first response time down?
Staff the hours when conversations actually arrive (the chat staffing calculator helps), use canned replies for common questions, and let an AI assistant answer what your knowledge base already covers.
Update history
New features, fixes and data refreshes for this tool, newest first.
-
· v1.0
LaunchedFirst release.
Missing something or found an error? Tell us — requests that come up often go into the next update. All tool updates
Speed is half of it
A fast first reply that does not solve anything still leaves the customer waiting. Look at resolution and satisfaction next to response time.