How to calculate your job application response rate
Count applications that got a human reply, use the right total and see why a reply rate and an interview rate tell different stories.
Ghosted editorialPublished 12 min read
On this page
- Pick one question first
- What counts as a reply, and what doesn’t
- Work through one month
- Count it in five steps
- Keep the dates fair
- Five calls to make once, then keep
- What the two numbers say together
- Other people’s rates are context, not a target
- What to do with the result
- Record five things and the counting is quick
- Questions people ask about response rates
Start here
Pick one question first
“Did a person reply?” and “Did I get an interview?” measure different things. We suggest tracking both. For this worksheet, count a personal rejection as a human reply. Leave automatic receipts out.
Human reply rate
- Asks: did a person read it and answer?
- Counts personal rejections, questions and invitations
- Moves sooner, because replies come more often than interviews
Interview rate
- Asks: did what they read make them want to talk?
- Counts any first conversation about the role, a recruiter screen included
- Closer to what you want from the search, but slower to show a change
Both divide by the same number: every application you sent in the period. That’s what lets you put them side by side, and it’s why an interview rate is never worked out from replies alone.
Why not count interviews only?
Interviews are rare enough that a month may bring none. Replies come more often, so they show a change sooner, and they tell you whether anyone is reading at all.
Definitions
What counts as a reply, and what doesn’t
Sort by who wrote the message, not by whether the news is good. A rejection from a person is a reply. A cheerful automatic email isn’t.
| The message | Human reply? | Why |
|---|---|---|
| “We’ve received your application”, a minute after you applied | No | A receipt. Nobody has read anything yet. |
| “Thanks, Alex. We’ve gone with someone closer to the role.” from Jo, the recruiter | Yes | A person read it and answered. Bad news counts. |
| “Could you do a call on Thursday?” from Jo | Yes, and an interview | It counts once in each rate. |
| An online test sent the moment you applied | No | Everyone gets it. A test sent after someone reviewed you counts. |
| A LinkedIn message from the hiring manager about your application | Yes | Where it arrives doesn’t matter. Who wrote it does. |
| “We will not be moving forward” from a no-reply address, weeks later | Your rule | An answer, but not a person engaging. Pick once and keep it. |
An interview is a first conversation about the role. A recruiter screen counts. So does a call with the hiring manager, or a video interview someone booked with you. Count the application once, however many rounds follow.
Worked example
Work through one month
Say you sent 20 applications in April. Four employers sent a personal reply, and two of those invited you to interview.
- Interview (2)
- Personal reply, no interview (2)
- No human reply (16)
Human reply rate20%4 ÷ 20 · “4 of 20”
Interview rate10%2 ÷ 20 · “2 of 20”
Five messages from one recruiter still count as one application with a reply.
Each rate is one division. Human reply rate: 4 ÷ 20 = 20%. Interview rate: 2 ÷ 20 = 10%. Both divide by all 20 applications, including the 16 that heard nothing. Leave those out and every month looks perfect.
There’s a third number hiding here. Of the four people who replied, two invited you: 2 of 4. That answers a different question, what happens once someone reads your application, and it rests on four applications. Treat it as a hint, and keep it apart from the two rates.
One reply is five points
With 20 applications, each reply moves the rate by five percentage points. With 50, by two. Small months swing: that’s arithmetic, not a verdict on your search.
Step by step
Count it in five steps
- Pick the period Applications you sent between two dates, by the day you sent them: 1 to 30 April. Not the day you saved the job, and not the day a reply came.
- Give each the same wait Every application in the period gets the same chance to be answered. Two weeks after the last one is a practical default. Write down the day you count.
- Count the total Every application sent in the period, including the ones that heard nothing. Saved jobs you never applied to stay out.
- Mark replies and interviews One mark per application for a human reply, and one for an interview, however many messages each one brought.
- Divide, and write both forms Replies ÷ total, then interviews ÷ total. Write “20% (4 of 20)”, so next month you can see how much each number rests on.
Timing
Keep the dates fair
Pick applications sent in the same period, then give them the same amount of time to receive a reply. Write down when you checked. Otherwise the applications you sent yesterday make this week look worse than last month. Keep saved jobs that you never applied to out of the total.
Here’s the same April, counted too early. On 3 May, three of the four replies had arrived. Six applications went out in the last days of April and hadn’t had two weeks yet.
- Personal reply (3)
- Sent in the last two weeks (6)
- No human reply (11)
Human reply rate, counted early15%3 ÷ 20 · “3 of 20”
15% against 20%, and nothing changed but the day you counted. There are two fair fixes. Wait until 14 May and count all 20. Or count now, but only the 14 applications that have had two weeks: 3 of 14 is 21%, and the six newer ones wait for the next check.
One check date
- Count everything sent in the period on one day, at least two weeks after the last application
- Easy in a sheet: one date to write down
- Early applications get longer than late ones, so keep the gap the same each month
The same window for each
- Count a reply only if it came within two weeks of that application
- Fairer: every application gets exactly the same chance
- Needs the date of each reply, not only a tick
58%
Roughly 58% of candidates who never got an interview were archived, or closed out, within nine days, in Ashby’s analysis of over 54 million applications from 2021 to March 2026. Interviewed candidates took 21 days on average. Archived doesn’t always mean you were told.
Before you count
Five calls to make once, then keep
Most arguments about response rates are really about definitions. Settle these before your first count, write them at the top of the sheet, and don’t change them between months.
Whether a template rejection counts
“We will not be moving forward”, from a no-reply address. Counting it measures whether you got an answer; leaving it out measures whether a person engaged. Either works. Switching between months doesn’t.
Replies that came after a follow-up
They count: a person answered your application. Note the follow-up on the same line, so you can see later how many of your replies a nudge brought.
How long you wait
Two weeks suits most searches. If your field moves slowly, use three or four. Whatever you choose, use it for every period you compare.
Applications you withdrew
Keep them if they had the full wait before you pulled out. Take them out if you withdrew within days: they never had a fair chance at a reply.
Replies that come after you count
Leave the old count as it was and note the late reply on the application. Changing past months makes every comparison shaky, and a run of late replies tells you to wait longer next time.
Reading the result
What the two numbers say together
Each rate on its own says little. Side by side, they point at where things stall: before anyone reads your application, or after.
| Reply rate | Interview rate | Where it stalls | Try first |
|---|---|---|---|
| 10% (2 of 20) | 0% (0 of 20) | Before anyone reads it | Check the top of your CV against each posting, and that you meet the must-haves |
| 30% (6 of 20) | 5% (1 of 20) | After someone reads it | Look at fit: level, location, salary. People are reading and passing |
| 10% (2 of 20) | 10% (2 of 20) | Reaching a person | When someone reads it, it works. Find more ways to reach people |
| 25% (5 of 20) | 15% (3 of 20) | Nowhere obvious | Keep going. If you need more interviews, send more of the same |
These are leads, not verdicts. Look for a pattern that holds for two or three periods, then change one thing and count again. If replies are the problem, tailoring your CV to the posting is the usual first fix. To split the same count by where you found each job, see which job boards get you interviews.
Outside numbers
Other people’s rates are context, not a target
It’s natural to want to know whether 20% is good. Published numbers can tell you roughly what’s common. They can’t tell you what’s good for you: they measure other people’s searches, and rates differ by field, level, country and year.
Most published figures also come from the employer side. They count every application a company received, including the ones sent without reading the posting. Your rate counts applications you chose, so it should look different.
What outside numbers are good for is calibration. If most of your applications hear nothing, that’s ordinary, not a sign that something is broken. Whether your search is getting better is a question only your own count can answer.
3%
The share of inbound applications, from people applying on their own, that went on to an interview at companies hiring with Ashby, across over 38 million applications from 2021 to 2024. For referred candidates it was 40%.
Source: Ashby, “Are referred candidates more likely to get hired?” (2025)
Over time
What to do with the result
If you have two replies from five applications, write “2 of 5” beside the percentage. Next month, compare the same definition and waiting period. This is a worksheet for your own search. Label the measure clearly if you compare it with an app that counts replies differently.
| Month | Sent | Human replies | Interviews |
|---|---|---|---|
| April | 20 | 4 of 20 (20%) | 2 of 20 (10%) |
| May | 12 | 3 of 12 (25%) | 2 of 12 (17%) |
| June | 30 | 6 of 30 (20%) | 1 of 30 (3%) |
May looks like the best month, but it rests on 12 applications: one reply fewer and it would drop to 17%, below April. June is the one to look at. Replies held at 20% while interviews fell to 1 in 30. People were still answering; fewer of the answers were invitations. Something about the roles you picked in June, or what they saw when they opened your CV, changed.
Change one thing at a time, such as the kind of role or the top of your CV, and count again after the same wait. Change three things at once and you won’t know which one worked.
Keep the count
Record five things and the counting is quick
You don’t need a dashboard. Five fields per application are enough, filled in as things happen rather than pieced together at the end of the month.
- Company and role
- Northline, Support Engineer
- Sent
- 3 April
- First human reply
- 11 April, from Jo (recruiter)
- Interview
- Invited 11 April, recruiter screen on 15 April
- Counted on
- 14 May, with the rest of April
Our job application tracker already has the date you applied and a status; add a column for the date of the first reply. How to organize job applications covers keeping it up. If an application goes quiet, a short follow-up email is normal: note the day you sent it.
Ghosted’s Stats page does this count from your board: it gives every application the same 14 days before judging it, lists newer ones as too new to judge, and shows the count beside the percentage.
Questions
Questions people ask about response rates
What’s a good job application response rate?
There isn’t one number that holds across fields, levels and countries, and published figures count other people’s applications in other ways. A good rate is one that’s higher than your own last period, counted the same way. If most of your applications hear nothing, you’re in common company.
Do automatic rejections count as responses?
Decide once and keep it. Counting them measures whether you got an answer; leaving them out measures whether a person engaged. Automatic receipts (“we’ve received your application”) never count: nobody has read anything yet.
How long should I wait before counting an application as unanswered?
Two weeks after you sent it suits most searches. Wait longer if your field moves slowly. The exact number matters less than using the same one every time.
Does a recruiter screen count as an interview?
Yes. Count any first conversation about the role with someone at the company or their recruiter, and count each application once, however many rounds follow.
How many applications do I need before the rate means anything?
More than you’d think. At 20 applications, one reply moves the rate by five points; at 10, by ten. Under about ten, treat the rate as a hint and compare over two or three periods.
Should I count jobs I saved but never applied to?
No. Only applications you actually sent go in the total. Saved jobs are a to-do list, not a result.
My rate dropped this month. Is it my CV?
Check the dates first: a month counted before its last applications had the full wait can only look worse. If the drop holds on a fair count, look at what changed: the roles, where you found them, or the CV you sent.
Your next move, on the board
See who replied, on one board.
Each application sits in the column for where it got to, and Stats works out your reply rate from it.