Candidate Database Reactivation: Turn a Dormant ATS Into Your Best Source of Hires
Key takeaways
- Candidate database reactivation is working the people already in your ATS instead of sourcing new ones. Every record cost you money to acquire, and you have already paid it.
- Segment before you dial. Five tiers, ordered by how warm the relationship is and how good the data is. Work tiers 1 to 3 first, treat tier 4 as the prize, and never call tier 5.
- Reactivation is the highest-TCPA-exposure activity in staffing, because it is by definition the act of calling the records whose consent is oldest. Almost nobody writing about reactivation mentions this.
- Consent does not expire on a timer. It ends when it is revoked, revocation can be made in almost any reasonable way, it must be honoured within ten business days, and it carries across calls and texts.
- The tooling has shipped in every major agency ATS for years and your database is still dormant. The constraint was never software. It was that nobody had the hours to make two thousand calls.
The asset you already paid for
Every record in your ATS cost your staffing agency money. A job board slot, a sourcing licence, a recruiter’s afternoon, an advert that ran for eleven days.
You paid all of it, once, and most of those records have not been touched since.
Here is the part that should bother you. Bullhorn has shipped candidate re-engagement automation for years. So has almost every other agency ATS. The button has been sitting there the whole time.
So the reason your database is dormant was never the software. It was that reactivation means making a few thousand phone calls with no live requisition attached to most of them, and no recruiter is ever going to choose that over a job order that pays this month. Nor should they.
What changed is that the calls no longer have to come out of a recruiter’s day. Which makes the question a different one: not whether to work the database, but which records, in what order, and what you have to check before you dial.
That last part is where this guide spends most of its time, because it is the part nobody else writes about and it is the part that can cost you real money.
This is a companion piece to our guide to AI voice agents for staffing agencies, which covers the category end to end.
What candidate database reactivation actually is
Candidate database reactivation is the practice of systematically re-engaging people already in your ATS, past applicants, silver medallists and former contractors, so they become live pipeline again, instead of sourcing new candidates for the same roles.
It sounds obvious written down. It is worth separating from two things it gets confused with.
Reactivation, sourcing and nurture are three different things
Sourcing brings new people into the database. It costs money every time.
Nurture keeps warm people warm, on a drip, usually by email, usually without anybody deciding who should be on it.
Reactivation is a campaign with a start and an end, run against a defined segment of records that have gone quiet, with the explicit goal of finding the small number who are looking right now. It is closer to a sales motion than a marketing one.
Why the unit economics beat a job board
A placement out of your database carries no sourcing cost. You are not paying for the advert, the board licence or the sponsored slot, because you already paid all of it eighteen months ago.
You are also skipping most of the qualification. For a decent chunk of these records you already know the rate they wanted, whether they were W2 or C2C, and what a client thought of them. That is real information a new applicant does not come with.
The catch is that the response rate is lower, because most of them are not looking. That is the trade: cheaper per contact, worse per contact, and it only works if you can make enough contacts that the arithmetic lands.
Why the database goes dark
Three things decay, and they decay at different speeds, which matters because they need different responses.
Contact data decays slowly and permanently. Numbers get reassigned, people change carriers, work emails die with the job. Nothing you do brings a wrong number back. This is the one with legal consequences, and it comes up again later.
Availability decays fast and reverses. Somebody who was unavailable in March is on the bench in September. This is the whole reason reactivation works at all. A record being cold tells you almost nothing about whether that person is looking today, which is exactly why the only way to find out is to ask.
Relationship decays gradually and asymmetrically. A contractor you placed remembers you. Somebody who applied and never heard back remembers you differently, and that shapes how the call goes.
The share of a typical database that is genuinely unusable is a real question, and it deserves a properly sourced answer rather than a number asserted in passing. Most of the figures circulating on this topic trace back to a vendor describing its own book.
Segment before you dial
Every guide on this subject says segment. Almost none of them say what the segments are, and the ones that do use buckets like “recently inactive” and “silent for a long time”, which sort by date and tell you nothing about quality.
Sort by pipeline stage instead. Your ATS already holds it, so this is a filter, not a data project.
| Tier | Who they are | Why they rank here | Share |
|---|---|---|---|
| 1 | Placed, assignment ended cleanly | Best data, warmest relationship, proven placeable. They took your money once | 5% |
| 2 | Submitted, not placed. Silver medallists | A client saw them and you thought they were good enough to put forward. Highest quality per record | 10% |
| 3 | Screened, never submitted | You hold structured answers already: rate, work authorisation, availability. Cheapest to re-qualify | 18% |
| 4 | Applied, never screened | The biggest tier, and the one nobody has ever spoken to | 52% |
| 5 | Sourced, never engaged | Scraped or imported. No relationship, and usually no consent to call | 15% |
Those shares are an illustrative shape for a 5,000-record database, not a measurement of yours. Run the filter on your own ATS and the proportions will be different. The order almost never is.
Tier 4 is the prize and the risk
Over half the database, and not one of those people has ever had a conversation with your agency.
That tier exists for a specific, arithmetic reason. A three-recruiter desk gets through around 54 first-touch calls in two working days, against requisitions that routinely pull 200 applicants. Everyone past that never gets called. Tier 4 is that gap, accumulated over two years.
Which means tier 4 is not a pile of rejects. It is a pile of people nobody assessed. There is no reason to think it is worse than tier 3 on average, only that it is unknown, and unknown is a different problem from bad.
The same gap is what drives your time-to-submit, and closing it is most of what an AI screening call is actually for.
Tier 5 does not get called
Scraped and imported records are not the same asset. Nobody in tier 5 gave you anything, including permission.
Email or a platform message if you want to work them. Not the phone. The next section explains why that distinction is worth more than the extra volume.
The compliance floor: what changes when the record is old
This section is general guidance, not legal advice. Have your own counsel review your setup before you run a campaign.
Here is the thing that makes reactivation different from every other calling workflow in your agency. Screening calls go to people who applied this week. Reactivation calls go, by definition, to the records whose consent is oldest and whose contact details are least reliable.
It is the highest-exposure calling activity in staffing, and across two full pages of search results on this topic, almost nothing addresses it.
Recruiting calls are covered, and that is settled
There used to be an argument that job recruitment sat outside the Telephone Consumer Protection Act because it is not selling anything.
In Loyhayem v. Fraser Financial, decided by the US Court of Appeals for the Ninth Circuit in August 2021, that argument lost. As the employment law firm Littler Mendelson set out in its analysis of the ruling, a pre-recorded recruiting call to a mobile number requires prior express consent, because the statute reaches any call to a cell phone unless it is an emergency or the recipient authorised it. The court read the prohibition as applying without regard to the content of the call, which is what removes the recruiting exemption people used to assume.
Statutory damages under the TCPA run at $500 per negligent violation and $1,500 per willful or knowing violation. Per call.
Consent does not expire, but it does get revoked
There is no timer. A consent given in 2023 does not lapse in 2025 on its own.
What ends it is revocation, and the rules on that moved recently in ways that matter for anybody working an old list.
In a Report and Order in CG Docket No. 02-278, adopted in February 2024 with the main provisions effective 11 April 2025, the Federal Communications Commission set out that a person may revoke consent in any reasonable manner. That includes replying STOP to a text, using a phone menu, filling in a form on your site, or simply telling somebody. The Commission listed stop, quit, revoke, opt out, cancel, unsubscribe and end as words that must always be honoured, and applies a rebuttable presumption that whatever method the person used was reasonable.
Two consequences follow immediately, and both bite on a reactivation campaign.
Ten business days. That is the outer limit for honouring a revocation once you have received it. Not “next time we refresh the list”.
It crosses channels. A revocation applies to calls and texts regardless of which medium it arrived through. Somebody who replied STOP to an SMS two years ago has revoked consent for the phone call you are about to place.
If your opt-outs live in your texting tool and your calling list is built from the ATS, you have a problem, and it is a data problem rather than a legal one, which means it is fixable this week.
The rule that keeps being postponed, and why you should build for it anyway
There is a third provision, and it is the one most commentary gets wrong.
Section 64.1200(a)(10) would treat a revocation given to one part of your business as applying to unrelated future robocalls and robotexts from you. One opt-out, everything stops, even on a different subject.
That provision is not currently in force, and it has now been postponed twice. The FCC first waived it in order DA 25-312 until 11 April 2026. Then, in order DA 26-12, released 6 January 2026, the Consumer and Governmental Affairs Bureau extended it again to 31 January 2027, to allow time to review the record compiled in response to a Further Notice of Proposed Rulemaking on whether the requirement should be modified at all.
Three things follow from that, and the third is the one worth acting on.
First, if you read an article telling you this rule is live today, that article is out of date. Plenty are.
Second, the waiver is narrow. It covers only the unrelated-matters part. Everything else in the revocation rules, including the ten business days and the cross-channel scope, took effect in April 2025 and applies to your campaign now.
Third, build your suppression list as though the rule were already in force. A single agency-wide opt-out list is the correct design regardless of what the FCC decides, it is not meaningfully more work than a fragmented one, and the alternative is discovering in 2027 that your opt-outs are scattered across four tools.
Reassigned numbers, and the only safe harbour there is
The candidate consented. Then they gave up the number, the carrier gave it to somebody else, and that somebody else has consented to nothing.
The FCC answer is the Reassigned Numbers Database, administered by Somos as the FCC-designated administrator. You submit a number and a date, and get one of three answers.
- Yes, the number was reassigned. No safe harbour. Do not call it.
- No. The safe harbour may apply, provided the number has not been permanently disconnected since consent was given.
- No data. No safe harbour. The database does not know, and not knowing protects you from nothing.
A “no data” response generally means the number and a permanent disconnect date are not in the database and the date you queried falls before 27 January 2021, which is when all service providers became obliged to report permanently disconnected numbers.
Read that last sentence against what a reactivation list is. Most reactivation lists are full of numbers whose consent dates predate the database coverage, which is precisely the population that returns “no data”. This is not a theoretical concern. It is the default outcome on an old list.
The arithmetic nobody does
Take the illustrative 5,000-record database from the last section. Tier 4 is 2,600 records.
Run that as a weekend campaign without checking consent, and the theoretical exposure is 2,600 multiplied by $500, which is $1,300,000. At the willful rate it is 2,600 multiplied by $1,500, which is $3,900,000.
Nobody is suggesting a full-book judgment is the likely outcome, and any article implying otherwise is selling fear. The point is the order of magnitude. A reactivation campaign is not a marketing decision that happens to involve phones. At two thousand records and up, it is a balance sheet decision, and it should be signed off like one.
Do the same multiplication on your own list size before you run anything.
The four-step floor
Nothing exotic. Four things, in this order, every campaign.
- Pull the consent record before the campaign, not after. If you cannot show when consent was given and how, the record does not get dialled. That is a filter, and it will shrink your list.
- Scrub against your own suppression list and the national Do Not Call registry. Treat every prior opt-out as permanent and cross-channel, including the ones sitting in a different tool.
- Run the list through the Reassigned Numbers Database. Drop the “yes” and the “no data”. Only the “no” carries protection.
- Log the disclosure and the opt-out offer in every transcript. Build the evidence while the calls are happening, because reconstructing it afterwards is not a thing you can do.
Designing the first campaign
Five decisions, made before anybody dials. The script and the cadence are their own subject.
Pick one tier and one role family. Not the whole database. Tier 2 against a single skill set is the right first campaign for almost every agency, because the records are warm, the count is small enough to review by hand, and the quality is the highest in the building.
Have a live requisition. This is the one people skip. A reactivation call with no actual job behind it is a survey, and it burns the record for the campaign you run next quarter when you do have something. If there is nothing live, wait.
Choose voice deliberately. The common advice is to start with SMS because it is cheap and fast. On a dormant contract database that is backwards. A two-year-old mobile number is exactly the case where you do not know who is on the other end, and a text gives you no signal about that at all. A conversation tells you in ten seconds whether you are speaking to the person you meant to reach, and it produces a transcript, which is the evidence the previous section was about. Use SMS for confirmations and follow-ups, once a human voice has established that the number is still theirs.
Cap the volume. A few hundred records, not the whole tier. You are testing the list quality and the script before committing the rest.
Decide what good looks like first. Write the number down before the campaign starts. Otherwise you will find a way to be pleased with whatever happens.
What to measure
Four numbers. Three of them are diagnostics and one is the answer.
Reachable rate. Of the records you dialled, how many reached a live human who was the person you meant to reach. This measures your data, not your campaign, and it is the number that tells you how much of the database is genuinely recoverable.
Re-qualified rate. Of the people reached, how many completed the conversation and gave you current availability, rate and authorisation. This measures the script.
Submissions from reactivated records. Of those re-qualified, how many went to a client. This measures whether the segment matched the demand you actually have.
Placements sourced from the database, as a share of all placements. This is the only one that survives contact with a profit and loss statement, and it is the one to put in front of whoever signs off the spend. If it moves from near zero to something visible over two quarters, the programme works. If the first three numbers all look healthy and this one does not move, you are reactivating the wrong segment.
No independent research body publishes a benchmark for any of these, which is worth knowing before somebody quotes you one. Measure your own first campaign and let that be the baseline.
What goes wrong
Starting with tier 4 because it is biggest. The largest tier is the coldest, the least documented and the highest risk. It is also where the volume is, so it feels like the obvious place to begin. Start warm, learn the script on people who remember you, then go wide.
Treating an old opt-out as expired. It is not. It is permanent and it crosses channels. The opt-out from 2023 sitting in a texting tool nobody has opened since is still binding on the call you place tomorrow.
Reactivating into nothing. No live requisition means no reason for the call, and the candidate can hear that. You get one honest reason to interrupt somebody’s Tuesday and it should be a real job.
Going again too soon. A second campaign into the same records three weeks later reads as a machine, because it is one. Whatever cadence you settle on, the interval between campaigns to the same segment should be measured in months.
Getting started
Run the filter first. Five tiers, sorted by pipeline stage, on your own ATS. It takes an afternoon and it will probably surprise you, because most agencies guess their tier 2 is bigger than it is and their tier 4 smaller.
Then do the compliance pass on the tier you intend to call, before you write a word of script. If that shrinks the list, the list was never really what you thought it was.
We built Voicegun for staffing agencies around agency workflows rather than corporate talent acquisition, which means consent and disclosure logged in every transcript, native write-back to Bullhorn, Avionté, JobDiva and Ceipal, and campaigns you can cap and stop.
Book a demo and bring a segment of your own database. We will run the tiering with you and show you what a compliant first campaign against it would look like.