Candidates are ghosted because closing the loop is manual, unassigned, and has no deadline — so it loses to every task that does. Ending it requires making the outcome message a mandatory, automated, owned step in the process rather than a courtesy that a busy recruiter performs when there is time.
- Ghosting is a process defect, not a character defect. Nobody decides to ghost anyone.
- The four causes: no owner, no deadline, no template, and fear of saying something legally risky.
- A pipeline where candidates can sit in a stage indefinitely will ghost people by default.
- Automate the guaranteed message; reserve human time for the candidates who got furthest.
- Silence costs you the referral, the reapplication and the review — from the people who liked you enough to apply.
Nobody decides to ghost anyone
That is the thing worth understanding before trying to fix it. No recruiter concludes that a candidate does not deserve a reply. What happens is much more boring: the reply is a task with no owner, no deadline and a small amount of dread attached, sitting in a queue behind tasks that have all three.
Which means exhortation does not work. Nobody needs to be told that ghosting is bad. They need the task to stop being optional.
The four causes
1. No owner. Ask who is responsible for the outcome message. In a lot of processes the honest answer is "the recruiter, unless the hiring manager does it, or the coordinator, depending". Shared responsibility is no responsibility.
2. No deadline. "When the role is closed" is not a deadline — roles stay open, get reopened, get put on hold. Meanwhile the candidate sits in a stage with no expiry, and a candidate who can sit in a stage indefinitely will be forgotten by default.
3. No template. Writing from scratch each time makes it a writing task rather than an administrative one, and writing tasks get postponed. Especially unpleasant ones.
4. Fear of saying the wrong thing. Often the real blocker, and usually unspoken. Someone somewhere was told never to give reasons in case it creates evidence. So the safest available message is no message, which produces exactly the reputational damage the caution was meant to avoid.
Cause four deserves a direct answer, because it is the one that overrides all the process fixes.
The legal objection, addressed
The concern is that stating a reason creates a record that could be used against you. The response has three parts.
Silence protects nothing. It does not make an indefensible decision defensible. If your reason would be embarrassing to state, the problem is the reason, not the stating.
Structured scoring solves what silence was covering. If you assessed against a rubric written in advance, you already have a defensible, specific, non-inflammatory reason: the candidate scored below the bar on a defined dimension. That is a comfortable sentence to write. The reason improvised reasons feel dangerous is that improvised reasons are dangerous — which is an argument for structure, not for silence.
In some places the reasoning is already disclosable. Under GDPR-style access rights, interview notes and scores are generally the candidate's personal data and can be requested. Writing notes you would be comfortable sharing is the discipline either way — see GDPR and candidate data.
So the guidance is: be specific about performance against defined criteria; do not speculate about the person. "Below the bar on system design depth — specifically distinguishing mitigation from root cause" is safe and useful. "Not senior enough for us" is vague, unhelpful and more exposed, because it invites the question of what you actually meant.
The fix
Assign one owner per stage. By name, not by role.
Set a published service level. "An outcome within five working days of the stage closing." Publish it in the job ad. A public commitment is a different kind of commitment.
Make stages expire. A candidate sitting in a stage past the SLA should generate an alert. If a person can be forgotten silently, they will be.
Automate the guaranteed message. Every applicant gets an outcome automatically. This removes the willpower dependency entirely, which is the whole game.
Tier the human effort by investment. This is how you make it sustainable:
| Candidate got to | What they receive |
|---|---|
| Applied, not progressed | Automated outcome, promptly, with what was assessed |
| Completed an assessment | Outcome plus their score against the dimensions measured |
| Interviewed | Outcome plus specific written feedback against the rubric |
| Final stage | A call, if they want one |
Effort scales with what the candidate spent. Nobody gets nothing.
Measure it. Percentage of applicants who received an outcome; median time to outcome; number currently past SLA. Review weekly. An unmeasured commitment is an aspiration.
Kill "we'll keep you on file" unless it is real, with a real talent pool, real consent and a real retention period. Otherwise it is a softer ghost.
What silence actually costs
The candidate-experience framing understates this. The costs are concrete:
- Referrals you never get. Applicants are people who already believed in your company enough to spend their time on it. That is your warmest possible audience, and silence converts it into a cold one.
- Reapplications you lose. Today's below-the-bar candidate is next year's strong one. Ghosted people do not come back.
- Public reviews. Employer-review sites are largely populated by candidates describing process, not employees describing jobs.
- Your assessment data. If candidates learn your process wastes their time, your strongest applicants — the ones with options — stop finishing it. Your funnel quietly degrades, and it looks like a sourcing problem.
- Nothing to compare. A candidate who never heard anything cannot tell you your assessment was broken.
And the asymmetry is stark: a two-sentence message that takes seconds to send is the difference between a person feeling assessed and a person feeling disposed of.
How CalHire makes it structural
CalHire treats this as an invariant rather than a best practice: every applicant receives a growth-framed report card. It is enforced in code, not optional.
Specifically:
- The word "rejected" never appears. The report card shows where the candidate stood — strengths, the dimension to grow, and a suggested next step.
- It is generated from the same scoring that drove the decision, so it is specific without anyone having to compose it.
- No candidate can be auto-rejected. Below-threshold candidates are flagged for human review, so nobody exits the process without a person deciding — which also means there is always someone accountable for the outcome message.
- Because evaluation is anonymous, the feedback is about demonstrated skills rather than background, which is both more useful and much safer to put in writing.
The zero-ghosting mechanics are on the features page, and the candidate's view of the report card on for candidates.
If you take one thing: stop treating the outcome message as courtesy that happens when there is time. Automate the floor, tier the human effort above it, and measure the number. Ghosting ends when it is no longer possible, not when everyone agrees it is bad.
Frequently asked questions
- Why do companies ghost candidates?
- Almost never deliberately. In most processes, sending a rejection is manual, nobody specifically owns it, there is no deadline attached, and the writer worries about saying something that creates legal exposure. A task with no owner and no deadline loses to every task that has both, every week, forever.
- Is it legally safer to say nothing?
- Silence is not a legal strategy. It does not remove your obligation to have made a defensible decision, and it does not stop anyone bringing a claim — it merely means you have communicated nothing. The genuine legal risk is in unstructured, improvised reasoning, which is solved by scoring against a rubric, not by staying quiet.
- How quickly should you tell a candidate they were unsuccessful?
- Set a service level and publish it — for example, an outcome within five working days of the stage closing. The specific number matters less than it being defined, visible, and measured. An unmeasured intention is what produces a four-week silence nobody noticed.
- Do candidates actually want feedback, or just an answer?
- Both, and they are different products. Everyone wants to know the outcome — silence is the actual grievance. Candidates who invested hours in an assessment or interview also want to know where they stood. Automate the outcome for everyone; add specific feedback in proportion to how far someone got.