A role opens, and someone on the team remembers last spring’s runner-up. The recruiter writes to her, she says she’s interested, and then the process breaks down in one of two directions. She’s sent back to the first screen and asked the questions she answered a year ago. Or she’s walked straight to the hiring manager as “already interviewed,” and the offer conversation leans on notes written for a different job.

Both come from the same skipped step. Candidate rediscovery starts from records your team already holds, the past interviews and the applicant tracking system (ATS) entry for each person, and I think those records usually get opened too late, after the candidate has replied.

I’d read them first and use them for two decisions: where the candidate rejoins the process, and what the new interviews need to test. Let the old interviews set the start, and let the new panel make the call on evidence gathered for the role that’s open now.

Read how the last process ended before anyone writes.

Start at the end of the old process, because that ending tells you what the candidate thinks of your company now. Someone who turned down your offer remembers a negotiation. Someone your panel turned down remembers a rejection email. The first message and the stage they rejoin at both depend on which it was:

The reason usually sits in two places. The ATS holds a close reason picked from a dropdown, and it’s often too coarse to act on, because a reason like “not selected” covers a finalist the panel loved and a candidate it didn’t. The scorecards and interview notes hold the reason in the panel’s own words, so read the scorecards first.

If your records carry no close reason at all, fix that before the next role closes. A separate guide covers the fields a candidate database needs, the close reason among them.

Sort the old evidence by how fast it goes stale.

Last year’s interviews are accurate about some things and out of date on others. Before you read the notes as a picture of the person today, sort what they hold by how long you’d trust it:

Then lay the new role’s scorecard beside the old interviews. For each competency the new role needs, mark whether the old loop asked about it, and when. Plan the new interviews around what the old loop never tested and any gap the last panel named, and list the untested competencies before you book a single call.

Watch the level, too. A candidate who interviewed for a senior engineer role and is now in the running for a lead role was measured against a lower bar, so every rating in the old loop tells you less about this one.

Write a first message that names the history.

A returning candidate already knows your company, and a message that reads like first contact tells them their last process left no trace. Send it from whoever they spoke to most, usually the recruiter who ran the process, and build it from the record:

• The earlier role, and roughly when you spoke.
• One specific thing from their interviews that your team remembered, taken from the notes instead of from memory.
• What has changed on your side since then, whether that’s the role or the thing they turned you down over.
• A question about where things stand for them, asked before you describe the new process.

If your side ended the last process, say so in a sentence, and say why this role is different. If they turned down an offer over pay, find out whether the range has moved before you write, and if it hasn’t, say that in the first message. A candidate who gives you a second call and hears the same number won’t give you a third.

Here is how that could sound to a finalist who withdrew to stay with her employer: “We spoke last spring about the platform engineering lead role, and the panel still remembers how you walked them through the payments migration. That role is filled, and we’ve since opened a second team on the same systems, with more of the on-call design work you said you wanted. Before I tell you more, how are things where you are now? Would twenty minutes next week suit you?”

Pick the stage where the candidate rejoins.

Get this decision wrong and the old interviews go to waste. I’d keep any stage whose evidence still holds for the new role and repeat any stage whose evidence has gone stale, which usually means the screen runs again, because pay and timing are the answers I’d expect to have moved most.

Then add the interviews for whatever the old loop never tested or left open. For a finalist who lost out to another candidate, that might be one conversation with the new hiring manager. For someone the panel turned down over a gap, it’s an interview built around that gap, scored on the same scorecard as everyone else in the role.

Tell the candidate which stages they’ll do and why. Someone asked to repeat an interview they already passed will assume you lost their file, unless you explain how the role has changed.

Give the hiring manager the same plan in writing before the first call: each piece of old evidence with its date and the interviewer behind it, the questions still open, and the interviews that will answer them. A hiring manager handed “interviewed last year” and nothing else tends to do one of the two things the opening of this guide describes, wave the candidate through or start again from the beginning.

Sometimes the answer is a later date: the finalist would move, just not this quarter. A separate guide on keeping a hiring pool covers holding people like that against a role on your hiring plan until their date comes round.

Give the new panel questions, and hold back the old ratings.

The new interviewers need to know what to test. They don’t need last year’s ratings before they meet the candidate, and I’d keep the ratings back until their own scorecards are in. An interviewer who has read “weak on system design” goes into the call listening for that weakness, and one who has read “strong hire” listens for reasons to agree.

So brief each new interviewer with the questions their interview has to answer, taken from the untested competencies and the named gap you listed earlier, and the competency each question belongs to. Once every scorecard is submitted, set the old evidence beside the new in the debrief with its dates showing, then let the new loop decide. Where the two disagree, talk it through: a gap that closed over the year is news the hiring manager should get from the people who watched it close.

Score the returning candidate on the same scorecard as everyone else interviewing for the role. A former finalist and a new applicant should be compared on one set of competencies, or the comparison says more about whose record was easier to read than about either person.

If the role has no agreed scorecard yet, a guide on building interview scorecards covers setting one up before the first interview.

Where Metaview fits in candidate rediscovery.

Before any of that reading, someone has to decide which past candidates are worth opening, and the reading itself depends on the old interviews having been kept in full.

When a role opens, describe it to AI Sourcing in plain language. The sourcing agent hunts on the web, in your team’s past Metaview conversations, and through the candidates your connected ATS has synced into Metaview, then lists who it found with the reasoning for each. Each name arrives with that reasoning and nothing more. Judging fit and making contact sit with the recruiter, who opens the person’s close reason and scorecards, the reading in the first two sections, before deciding whether to write.

For the interviews your team records, Metaview’s Notetaker joins the interview as a visible participant. Where your own consent process allows the recording, it records and transcribes that conversation, then builds structured notes from the transcript.

With a recording that captures every spoken word, a recruiter getting ready to write to last year’s finalist can read her answer to each question as she gave it, with the date she gave it, and quote her own answer in the first message.

1

1. 1The question says “currently managing,” so the answers below describe her team on the day of the call, a fact to ask about again.

To read a whole loop in one place, multi-source notes can pull several conversations and documents together into a single notes document. For a returning candidate I’d add last year’s interviews and the new role’s job description to one of those, so the old answers sit next to what the role asks for now.

Those are the lines I’d hunt for first in an old loop: what an interviewer wanted to know and ran out of time to ask. Each one hands the rejoining interview a question it can open on.

Pay is one of the first answers to go stale, and it’s the one to check before you write to someone who turned down an offer. In Reports, a question about pay can be put to your team’s own interview data, and an earlier design of its web page showed one answered across a set of calls:

1

1. 1The single round figure Gemma B. gave on a recruiter screen, which is the answer to ask about again before any talk of an offer.

A figure like that is what the candidate said on that call, so treat it as the place your first conversation about pay starts, and ask again before you put a number in an offer.

Follow one returning candidate from record to decision.

I’d trust this method on my own team only after seeing it hold up on one real file. Pull up the last past candidate your team brought back and read their record from the old process to wherever they are now:

• Was the first message written for them? If a stranger could have received it unchanged, no one read the history, so make reading the old interviews part of drafting the message.
• Which stages did they repeat? If they sat an interview whose evidence still held, decide the stages first next time, and give that plan to the hiring manager in writing.
• What did the new interviewers see first? If last year’s ratings reached them before the call, send the open questions next time and keep the ratings for the debrief.
• What carried the decision? If the debrief leaned on the old scorecards more than the new ones, the new loop was too thin, so add the interview for the open question.

A file that holds up shows the method doing its job. The past interviews and the ATS record pointed the new loop at what was still unknown about someone your team already rated highly, and the new loop answered it before anyone decided whether to make the hire.