// Recruitment Technology, Recruitment Strategies

How Fast a Candidate Database Goes Stale

The bottom line

A candidate database loses accuracy far faster than most agency refresh cycles assume. Agencies that audit their records commonly find only 15 to 20 percent are still current after two years. Job titles, employers, contact routes and consent all move while the record sits still.

The answer is continuous enrichment and job-change signals running in the background. A cleanup project scheduled for a quiet quarter will not hold.

Why every database refresh project stalls halfway

Ask an operations director how many records sit in the CRM. You get a confident number in the hundreds of thousands. Ask how many of those they would trust for a live role tomorrow, and the answer gets vague. That gap between volume and usable data is the true condition of most agency databases. It widens every quarter without anyone deciding to let it.

A candidate database is a snapshot of a labor market that started moving the second you saved it. Titles change, employers get acquired, direct dials get reassigned, and whole teams get restructured. The consent captured three years ago quietly stops covering the outreach you want to send now. None of that generates an alert. The record looks exactly as authoritative on screen as it did the day a recruiter typed it in.

Most agencies respond with a cleanup project. Every agency also has one that reached about 40 percent completion. It now lives in a shared folder nobody opens.

How fast does a candidate database actually go stale?

Faster than an annual refresh can absorb. Across recorded conversations with more than forty agencies, the same figure keeps surfacing. Only 15 to 20 percent of records stay current after two years, once title, employer, contact route and consent are checked. The rest are not worthless. They are simply not placeable without work.

The labor market math explains the pace. In the US, median employee tenure has fallen to 3.9 years, the lowest reading since 2002. It runs far shorter across younger and more mobile segments. Half your database changes employer inside four years by definition. On technology and GTM desks, the practical half-life sits closer to two.

The commercial effect shows up long before anyone diagnoses the cause. Emails bounce, a gatekeeper mentions the person left months ago, and a one-day shortlist takes three. Across B2B more broadly, 44% of companies report annual revenue loss of over 10% attributed to CRM data decay. For an agency, that lands as missed re-engagement windows and slower time-to-send. It also lands as BD calls placed to someone who moved on two roles ago.

Which fields in your candidate information rot first?

Job title and employer go first, and they take everything downstream with them. When someone moves, the work email dies and the direct dial usually goes with it. The seniority your team filtered on last year is now wrong by a level. The rest of the candidate information degrades more quietly, which makes it more dangerous.

Salary expectation, notice period, location preference and availability have a shelf life measured in months. A candidate marked available in March is under contract by June, and nothing in the record says so. This is why grading and tagging every applicant at the point of capture beats a periodic tidy. Accuracy has to be set when the record is created, then maintained from there.

Consent is the field agencies forget. GDPR and comparable regimes tie your lawful basis to a stated purpose and a time window. A record untouched for three years raises a compliance question as well as a data quality one. A dormant database carries risk alongside its cost. It sits there unsearchable, still holding personal data your team can no longer clearly justify keeping.

The fields carrying the most commercial value are the ones legacy systems never captured at all. Think of what the candidate said on the call, or why they turned down the last offer. That context is the difference between a name and a warm candidate. Holding onto it is the whole purpose of recruitment AI memory. It fades the same way a phone number does.

Why do one-off cleanups fail a recruiting candidate database?

Because a cleanup measures a moving target once. The project finishes and the data is accurate for roughly a week. Then the decay curve resumes exactly where it left off. The work also tends to fall to whoever has capacity, which means it competes with billing and loses.

The structural problem sits underneath the effort. 56.16% of agency recruiters describe their technology setup as functional but fragmented. In practice, candidate information lives in a CRM, a sourcing tool, an inbox and a spreadsheet at once. Cleaning one of those four achieves very little. The same research found manual work is the biggest operational challenge agency leaders name.

There is a second failure mode that most agency leaders have already lived through. A cleanup depends on recruiters updating records by hand, and recruiters are measured on billings. CRM adoption fails for the same reason every time. The admin is immediate, the payoff is deferred, and the desk wins that argument.

Agencies that treat data decay as an event rather than a condition run the same project every two years. Same cost, same temporary result.

What should continuous enrichment look like in your recruitment software?

It should run whether or not a recruiter remembers to trigger it. Continuous enrichment means the system watches for role and company changes across your network. It updates the affected record when someone moves, and flags the change while it is still news. That last part carries the commercial weight. A job change is a live BD signal and a warm candidate at the same moment.

Keeping records current without adding admin is what agentic AI is genuinely good at. It is the design principle behind Atlas, an AI-powered CRM and recruitment platform for agencies. Its real-time tracking of role and company changes updates candidate records automatically. You are notified the moment a contact moves, and Atlas can draft a personalized message so you reconnect before a competitor does.

Capture matters as much as monitoring. An AI-powered CRM that parses every resume and syncs every contact on arrival means the record enters clean. Work history, skills, seniority and company context arrive already structured. Enrichment then has accurate information to maintain. The alternative is a half-filled record built from a CV attached to an email in 2022.

Current records only pay off when a recruiter can find them under pressure. Plain-language search across your recruitment database software turns a maintained database into a working talent pool. Pace Global handles 30% more candidates and roles and saves over an hour a day. Its consultant named search as the feature with the biggest impact on results.

There is a candidate experience argument running alongside the commercial one. Message a senior candidate about a role at the company they left eighteen months ago. It tells them precisely how much attention their record has had. Multi-touch outreach built on current data reads as a relationship. Built on stale data, it reads as a mail merge.

Frequently asked questions (FAQs) on candidate database decay

What your week looks like when the database keeps itself current

A candidate database that maintains itself changes the shape of a recruiter’s week. The hours spent verifying titles, chasing dead numbers, rebuilding longlists and re-checking consent go back into calls and briefs. Shortlists land sooner. Job changes arrive as opportunities on the day they happen, rather than as bounce reports found three weeks late.

That shift is what agentic AI inside Atlas is built to deliver. The record keeping happens in the background, so the desk stays on revenue. If a database cleanup is already penciled into your next quarter, that is the quarter worth reclaiming.