What this has actually done
Live, from the database, every time you load this page. It goes down as well as up, and it says so when the answer is none.
No reply rate yet. With fewer than 3 people using it, a percentage is one person's diary rather than a statistic, and quoting it as though it were evidence would be the exact thing this page exists to prevent.
What normal looks like
So you can judge the numbers above rather than take a word for them. These are other people's figures, not ours.
| Measure | Reported | Source |
|---|---|---|
| Applications that reach interview | 2–3% | 2026 industry average |
| Sent 11–20 applications | 9.25% | Huntr, Q1 2026 |
| Sent 100+ applications | 2.58% | Huntr, Q1 2026 |
| A bulk auto-apply tool, tested | 0.5% | 200 applications, one callback |
Read the third row again. Applying to more jobs made it worse — people who sent over a hundred applications interviewed at less than a third of the rate of people who sent eleven to twenty. Volume is not a strategy that is merely inefficient. It is one that backfires.
Where the founder's own numbers come from
The 171 applications and 32 replies quoted elsewhere on this site are Harry's, sent by the machine this product was built out of, from his own mailbox, before there was a product. They are counted from his sent folder and his replies — not from this database, which is why they are not in the panel at the top.
They were never typed in by hand. That machine wrote them out at the end of every run until 2026-09-23, when his own search moved onto this site, so they are every letter it sent rather than a good fortnight chosen out of it. From then on his letters are counted in the panel at the top with everybody else's.
They also measure a different thing from the table above. Those are interview rates. His is a reply rate: an answer from a human being at the company. Putting the two side by side as though they were the same number would be dishonest, so they are kept apart.
Take the numbers
All of this is at /numbers.json,
machine-readable, CC BY 4.0. Three sets kept separate, because they measure
different things: the fixed 86-email study, Harry's live job hunt, and what
this product has sent for its users. Every rate carries the definition of
"reply" it was measured with, and every thin row carries its caveat.