Recruiting for hard-to-fill roles: what actually works
A role stays open for one of four reasons, and only two are sourcing problems. How to tell which one you have before you spend another month.
A role stays open for one of four reasons, and only two of them are sourcing problems.
Most teams respond to a stuck role by sourcing more. If the cause is elsewhere, that adds a month and changes nothing.
So the first move is not a better search. It is finding out which of the four you have.
The four causes: nobody matches, they match but do not reply, they reply but drop out, or the brief is wrong.
Each has a different fix, and three of them are not fixed by a sourcing tool.
This page gives the diagnosis, then what to do for each case, and where AI sourcing genuinely helps.
TL;DR:
- Count before you act. How many did you find, contact, and hear back from?
- Few found: the brief is too narrow, or written in your company's vocabulary.
- Found but silent: one channel, or a first line that could go to anyone.
- Replies then drops: the problem is the offer or the process, not sourcing.
- A rare role is a small pool by nature. Widening beats searching harder.
Key facts:
- Kalent database: 850 million profiles, 250 million enriched
- Share of returned profiles directly reachable: around 80%
- Automated channels: email, LinkedIn, SMS, WhatsApp, on Copilot
- Price: $149 and $179 a month, per licence
The diagnosis: three numbers
Before anything else, write down three figures for the last month on this role.
How many profiles you found that you would genuinely have sent to the hiring manager. How many you contacted. How many replied.
The shape of those three numbers names your problem.
| What you see | What it is |
|---|---|
| Found very few | a brief problem, or a vocabulary problem |
| Found many, contacted few | a reachability problem |
| Contacted many, few replies | a channel or a message problem |
| Replies, then everyone drops | the offer or the process, not sourcing |
If you cannot produce the three numbers, that is the finding. You are working the role on impression.
Case 1: you find almost nobody
Two causes, and both are in your own words.
The brief is narrower than the market. Your hiring manager listed nine requirements. Three are real. Each extra "and" cut the pool.
The vocabulary is internal. You wrote "Growth Engineering". Nobody puts that on a profile. Describe the work, not your org chart.
The fix is to widen, not to search harder. Describe the job in a sentence, add the two or three things a career would have to show, and let the search go past your words: adjacent titles, equivalent paths, companies where the same job carries another name.
Case 2: you find people but cannot contact them
This is the case teams misread most often.
You have forty good profiles and you are reaching them on one channel, because that is the only contact detail you have.
A profile with no verified number is a research result, not a candidate.
At Kalent, 250 million of 850 million profiles carry a verified number and email, and around 80% of what the agent returns is directly reachable.
On a rare role this matters more than anywhere else, because the pool is small by nature. Losing half of forty people to a missing phone number is losing the search.
Case 3: you contact them and nobody replies
Two things to check, in this order.
How many channels. At FAB Group, 50 consultants, 98% of approaches left as InMails and the reply rate settled at 15%. Moving to four automated channels took it to 30%, across more than 18,000 qualified profiles in 12 months.
The first line. If it could be sent to anyone else on your list, it is not personalisation. The reason for the match is the first line: what in their career made them a fit, written when the profile was ranked.
"I came across your profile" is the most sent and least personal sentence in recruiting.
Case 4: they reply, then they drop
Then it is not a sourcing problem, and no tool on this page will help.
Look at where they leave. After the salary conversation, after meeting the manager, after the technical round, or at the offer.
Each of those points at something different: the range, the manager, the process length, or a competing offer.
Sourcing more people into a funnel that leaks at stage four is the most expensive mistake in this list. You will spend another month proving the same thing.
What is actually true about rare roles
The pool is small, and that is the whole difficulty.
There are not thousands of Rust engineers with payments experience in your city. There are perhaps forty, and every serious tool will find roughly the same ones.
We do not publish scarcity figures by trade, because we have no source for one that would hold.
So the competitive question is not who finds them. It is who reaches them, on how many channels, with a message that shows the career was read. On a pool of forty, that is the entire contest.
A working order for a stuck role
- Produce the three numbers. Found, contacted, replied.
- If found is low: rewrite the brief around the work, cut to three real requirements.
- If contacted is low: check reachability before writing anything.
- If replies are low: add a channel before you add candidates.
- If they drop later: stop sourcing and go look at the funnel.
- Re-run weekly, with the same three numbers, so you can see which move worked.
Related reading
The AI sourcing comparison cluster, all read on the same date against the same criteria:
On the same subject: how to write a sourcing prompt, and boolean search vs AI sourcing.
On reachability: why 80% phone coverage is the outreach standard.
And the product pages: pricing, ATS integrations, trust center.
Frequently asked questions
Why is my role hard to fill?
One of four reasons: nobody matches, they match but do not reply, they reply then drop out, or the brief is wrong. Count how many you found, contacted and heard back from, and the shape of those numbers names the cause.
Does AI sourcing help on rare roles?
On two of the four causes: it widens a search past the brief's vocabulary, and it returns profiles with verified contact details so you can reach more of a small pool. It does not fix an offer or a broken process.
Should I loosen the requirements?
Usually yes. Nine requirements is a wish list. Three real ones plus what a career should show will find more of the right people, not fewer.
How many candidates should I expect on a niche role?
Fewer than you want, and roughly the same number on any serious tool. The pool is small by nature, so the difference is made in outreach, not in search.
What if candidates reply and then drop out?
Stop sourcing. The problem is downstream: the range, the manager, the process length or a competing offer. Sourcing more people into a leaking funnel repeats the month.
