The Clearance-Likelihood Read: Cleared, and to What Level

Deciding whether a sourced candidate probably holds an active US security clearance, and roughly at what level, is a daily inference problem in defense and govtech sourcing. The profile in front of you will almost never say "TS/SCI," and even when it does, the text is a claim, not a fact. This guide is the scoring model: the public signals that carry weight, how to read them into a level band and a status, and how to attach a confidence flag you can defend, all without asking a candidate to violate disclosure rules.

It is written for in-house recruiters, sourcers, and talent leaders working cleared roles. The ranking pages tell you to buy a cleared-talent database and post to niche boards. That is procurement, not sourcing. The actual skill is the read.

Why you infer a clearance instead of reading it

A public profile rarely states a clearance, so the job is inference from context, not lookup. You are reading employer type, contract vehicle, customer agency, facility location, and posting language into an estimate, then flagging how confident that estimate is.

Start with scale, because it tells you where the signal lives. Over 85% of clearance holders work with the Department of Defense, and DCSA oversees over 95% of federal background investigations. That concentration means the vocabulary you are hunting for is DoD and contractor vocabulary: program names, agency customers, and the language of a Performance Work Statement, which is the contract document that spells out what a contractor must deliver.

There are three formal levels under Executive Order 13526: Confidential, Secret, and Top Secret, each tied to the level of damage unauthorized disclosure could cause. SCI (Sensitive Compartmented Information) and SAPs (Special Access Programs) are not separate levels. They are additional read-ins layered on top of a Top Secret eligibility. So "TS/SCI" is one Top Secret clearance plus a compartment, not a level above Top Secret.

The signal you most want - an explicit self-label - is also the least reliable, because it is self-reported. That tension runs through everything below. You build the estimate from weak, corroborating signals and reserve certainty for a formal check that happens after the candidate is interested.

The signals, and what each one proves when it lies

Six public signals correlate with holding a clearance. Each proves something narrow, and each has a specific way of misleading you. Read them as a set, never one at a time.

The one signal that carries structural weight is a ClearanceJobs presence: candidates must have an active or current US federal clearance to be on the platform at all. That is a gate, not a self-report. But the level shown on a ClearanceJobs profile is still self-reported by the user and not verified by the platform. So presence is meaningful, level is a claim.

Employer is the signal most people over-trust. Named cleared employers - the likes of Leidos, GDIT, ManTech, Amentum, and Peraton - surface constantly in these searches. But a defense-prime badge only tells you the person works somewhere that does cleared work. It does not tell you they personally hold access. A govtech role adjacent to a prime, with no facility access, is the classic false positive.

Scoring the level band: default to Secret

Default every estimate to Secret and upgrade only on explicit evidence. Secret is the workhorse level, the majority of cleared positions sit there, and "clearance required" without specification usually means Secret.

The population data backs the default. In the FY2019 ODNI figures, most holders sat below Top Secret.

Confidential is rare in practice. Relatively few positions are cleared at Confidential alone; most roles that need access need Secret or above. So your practical band is a binary in most cases: Secret unless proven otherwise, or TS/SCI on explicit signals.

Match the level to the reinvestigation cadence too, because dates on a profile can hint at level. Top Secret is reinvestigated every 5 years, Secret every 10, and Confidential every 15 - though continuous vetting is replacing that fixed cadence for many positions.

Upgrade to TS/SCI only on explicit signals: SCIF work described in the profile, an intelligence-community customer, a stated polygraph, or the TS/SCI label itself. Everything else stays Secret.

The disclosure skew that distorts every list

Disclosure skews hard by level, so your raw search results over-represent TS/SCI and under-represent Secret. Correct for it or your pipeline will look more elite than the population is.

Refolk's index makes the skew visible. Compare the two most common phrasings on public US profiles.

Set that against the population. Top Secret was only 32.6% of holders in the ODNI figures, yet TS/SCI appears 20.9 times more often than the "active secret clearance" phrase in the index. TS/SCI holders disclose far more, partly for search visibility and partly because the rarer, higher-value clearance is worth advertising. Secret holders, who are the majority, often say nothing specific or write only "cleared."

There is a second lesson in the index: disclosure phrasing clusters by role. Of the 253 profiles that reference "active secret clearance," only 14 carry the title Intelligence Analyst, and 11 of those 14 sit in that single role. The exact phrase you search for shapes which roles you surface, so vary phrasing when you want breadth. If you only search for TS/SCI, you will miss the larger, quieter Secret population that never types the level at all.

This is precisely the friction a plain-English search removes. Rather than guessing which of a dozen phrasings a candidate used, you describe the person and let the index handle the variants across GitHub, LinkedIn, and the open web.

Scoring active versus lapsed: the 24-month rule

Status turns on one rule. If the periodic investigation is out of date, or there has been a break-in-service of two years or more, the person must be nominated for a new clearance and complete a fresh application like someone who never held one.

The exception is the person still in the seat: a clearance stays active as long as they are in a job requiring access to classified information. So the read is simple in shape and easy to get wrong in detail.

• Likely active. Currently in a cleared role, or separated less than 24 months ago with a strong self-label such as "Active."
• Likely current/inactive. Holds eligibility but not accessing classified daily; "Current" or "Eligibility on file" phrasing inside the window.
• Likely lapsed. Separated 24 months or more ago with no reinstatement signal.
• None. No cleared footprint, or a clean-transfer gap that broke reciprocity.

Two phrasings do heavy lifting. "Active" means holding and actively using it. "Current" means holding but not accessing classified daily. And "Eligibility on file" denotes someone within the 24-month reinstatement window but not currently in a cleared billet - a useful, reachable state worth its own search.

Count the window from separation date, not last investigation date. Continuous vetting - over four million positions had enrolled by September 2021, eliminating periodic reinvestigation for many - does not extend the lapse rule. That timeline still applies from the day the person left cleared work.

The scoring procedure

Run these seven steps in order. Steps 3 and 4 can swap - resume-focused writers lead with status, sourcing writers lead with level and vehicle - but do all of them before you spend outreach. Budget about 45 minutes per profile until the reads become automatic.

The output of the procedure is one line you can paste into a sourcing note. Keep it short and structured so the next person can audit it.

CLEARANCE ESTIMATE
Level band: Secret / TS/SCI / Unknown
Status: likely active / current-inactive / lapsed / none
Self-label (verbatim): "____________"
Basis: employer=____ | agency/vehicle=____ | facility=____ | ClearanceJobs=Y/N
Last cleared role end: MM/YYYY  (months since separation: __)
Confidence: High / Medium / Low
Verify at: pre-offer DISS check by FSO (not requested from candidate)

Presence on a cleared platform is a gate; the level typed beside it is only a claim.

How the read goes wrong

The failure modes below are where defensible estimates collapse into wishful ones. Treat this as the most important section: a scoring model that overclaims is worse than no model, because it sends outreach and offers into dead ends.

• Employer equals clearance. A defense-prime employee may sit on unclassified commercial work. The false positive is a FAANG-adjacent govtech role with no facility access. Check for agency and PWS keywords, not just the employer name.
• Self-label taken as verified. ClearanceJobs and LinkedIn levels are self-reported. Someone may type "TS/SCI" for search visibility on a lapsed clearance. Only a DISS check confirms it; treat text as a claim.
• Level inflation. Assuming SCIF or IC context means TS/SCI when the role is Secret. Default to Secret without explicit TS signals, because most roles that need access need Secret or above.
• "Active" versus "Current" conflation. A profile that just says "clearance" may be inactive. Listing a long-dormant clearance misleads recruiters about true availability. Check the last cleared-role end date against the 24-month window.
• Break-in-service miscount. Continuous vetting does not extend the lapse rule. Count from separation, not last investigation.
• Reciprocity over-read. A clean transfer does not carry SCI or a polygraph across agencies. Do not assume either moves cleanly.
• Compliance drift in outreach. Pushing a candidate to name programs to "prove" a level invites spillage. A detailed war story may be classified. Steer to functional scope only.

The most expensive of these is level inflation crossed with self-label trust: you flag a candidate as TS/SCI on their own typed label, build a slate around it, and discover at offer that the clearance is Secret and lapsed. The disclosure skew makes this easy to do, because TS/SCI is the loudest phrase in the data and the rarest state in the population.

Staying inside the disclosure and verification rules

You may estimate freely from public signals, but you may not push a candidate toward classified detail, and you may not verify through DISS for recruiting. These two limits are not optional; they protect the candidate and your program.

DCSA's DISS Account Management Policy governs verification: industry companies are authorized to verify a prospective employee's eligibility in DISS prior to an offer, but they may not use DISS for recruiting purposes. In practice, that means the DISS check is an FSO action at offer stage, downstream of your estimate. Nothing in your sourcing workflow touches it.

On the candidate side, unacceptable disclosures include compartmented programs, classified system names, specific polygraph results, and mission details. Ask candidates to clarify roles, levels of access, and security responsibilities without asking for classified details. If a candidate volunteers something specific, redirect to functional scope.

Get this right and the value is concrete. A cleared candidate saves an employer six to twelve months of processing and tens of thousands in investigation costs, and end-to-end processing across all tiers averaged about 243 days in DCSA 2025 data. The whole point of the estimate is to spend outreach where that saving is real, not where a self-label is bluffing.

Before you call the estimate done

Run this checklist against every profile before it enters a slate. Each item is a specific check, not a topic.

Keeping the model current

The rules in this model move on known mechanisms, so re-check the mechanism, not the number. Three things will drift: the reinvestigation-versus-continuous-vetting picture, the DISS verification policy, and the population totals. When you revisit, confirm the 24-month lapse window still governs status (continuous vetting has not changed it so far), confirm DISS is still off-limits for recruiting, and refresh the population split from the latest ODNI determinations report rather than trusting the FY2019 figures forever.

Re-run the disclosure skew inside your own tooling every quarter. The 20.9x TS/SCI-to-Secret ratio in Refolk's index is a snapshot of how people label themselves, and labelling habits shift. If Secret disclosure rises, your default-to-Secret rule stays correct but your search phrasing should widen. The model's spine - infer from context, default to Secret, count the 24 months from separation, verify only at offer - is durable. The dials on top of it are not, so treat them as things to re-measure, not facts to memorise.