The cheapest candidates you will ever find are the ones who already applied, already got assessed, and already wanted to work for you. Most companies never contact them again.
Why your own database beats the open market
Someone who reached a late stage and did not get the offer is a candidate you have already spent money assessing. You know what they can do from evidence rather than from a CV. They have already decided they want the job. And they will almost always reply, because a message from a company they nearly joined is not cold contact.
Everyone below that level has value too. A near-miss for one job is often a strong fit for a different one, and near-misses build up much faster than hires.
How to do it
The process is the same as screening, pointed at a different pile.
- Run the new job's match profile against your existing candidate records instead of new applications.
- Filter by how recent they are. Records older than about two years describe someone who has moved on. Treat those as leads to check, not as facts.
- Add your own history: how far they got, what the notes said, and — most importantly — why they did not progress.
- Read the notes before contacting anyone. "Strong, lost to a better offer" and "withdrew after the technical stage" are opposite situations.
Two things make this harder than it sounds. Old records were read by whatever tool you had at the time, so the quality varies across your database. And reasons for rejection are usually recorded as stage codes rather than actual reasons — which is why the notes matter more than the status field.
Getting back in touch without annoying people
The difference between a flattering message and an irritating one is being specific.
- Name the previous process. Which job, roughly when. Vagueness reads as a mail merge, which is usually what it is.
- Say why you are back. Ideally something that has changed — a different job, a different level, a different team.
- Be honest about what happened. If they were close, say so. If your notes are thin and you do not know, do not invent warmth you cannot back up.
- Make it easy to say no. Someone who was rejected once and is now being chased should not have to work to decline.
What makes people angry is being contacted as though the previous conversation never happened — especially if it ended badly or with silence. This is where poor candidate experience from six months ago comes back to you.
The rules side
This only works if you were allowed to keep the data. That means a deletion period you actually apply, a reason for holding records beyond the original application, and having told candidates at the time that you would keep their details for future jobs. Module 11 covers it properly.
The practical failure to avoid: a database you cannot delete from. If someone asks to be removed and their details sit in three tools plus a recruiter's spreadsheet, you have a problem that gets worse the more you use this approach.
Measure it separately
People from your own database respond far better than cold contacts. Mixing the two makes your overall numbers look good while hiding which activity actually worked.
| Track separately | Why |
|---|---|
| Reply rate: database vs cold | The gap is usually large, and it justifies the time spent keeping records |
| Time to hire by source | Database hires are normally much faster |
| Offers accepted by source | Tells you whether "nearly joined" converts, or was a polite no |
| Records contacted per hire | Tells you how big a usable database you really have |
Do this before you search
Every new job: check your own database first, then go to the market for the gap. It takes twenty minutes and regularly produces two or three people who are further along than anyone you will find this week.