The Openness-Signal Read: Reach Now, Watch, or Skip a Passive Candidate

You have a list of hundreds of passive candidates and one Tuesday. Messaging all of them the same day is the default the top search results still assume, and it wastes your InMail budget on people who are not moving. This guide gives in-house recruiters, sourcers, and founders doing their own hiring a framework for reading dated, individual openness signals so you can sort any list into reach-now, watch, or skip with a cutoff you can defend.

This is a different read from two you may already know. Scoring a candidate's barriers to moving tells you whether a move is feasible. Scoring a company's talent exposure tells you where to hunt. Neither reads the individual, dated signals that a specific person is warming up right now. That is the prioritization judgement this framework makes.

What "openness" actually is, and why a snapshot misses it

Openness is a change over time, not a state you can read from one profile view. The strongest dated evidence comes from a Stanford and NBER working paper using monthly profile vintages: among workers who change employers, the probability that a worker edits their profile begins to rise about six months before the transition, climbs sharply in the final few months, peaks close to the employer change, and then falls back.

The load-bearing number: workers are 66% more likely to change employers during the period in which they update their profiles. The edits observed are headline and summary rewrites, title tweaks, and rewritten descriptions of past roles, often alongside a new headshot. Practitioner accounts line up with this - when a passive candidate updates their headline, rewrites their summary, or adds new skills, they are polishing their profile for visibility, and this happens weeks before they apply anywhere.

The consequence for your workflow is structural. If you only look at a profile once, you see the badge and the current title, and you miss the diff that carries the real information. A snapshot-only process defaults to the lagging indicator (someone already moved) and never sees the leading one (someone is preparing to).

The signals that matter, and what each one proves

Six public-footprint signals carry weight. For each, know what it proves and what it looks like when it lies.

• Headline or summary rewrite. Proves active profile polishing, the earliest openness indicator. Lies when it is a one-off cosmetic tweak with no other change around it.
• New skills added. Proves preparation for a search when paired with other activity. Lies as keyword-chasing: retroactive AI and LLM term additions surged, so a fresh "LLM" skill alone may be noise.
• New photo or headshot. Proves the candidate is grooming for visibility. Weak alone; it earns weight only when it clusters with a rewrite.
• Following or engaging with target companies. Proves directional interest. People who follow your company are 81% more likely to respond, which makes this both a signal and a reply-rate lever.
• A new certification or credential. Proves investment in a next step. Lies when it is unrelated to any move (compliance renewal, hobby course).
• Open to Work badge. Proves self-declared availability, but with heavy selection bias covered below.

The single most important discipline: score these separately. Track the actual scoring model you use and keep each component visible rather than blending everything into one number that hides which input actually drove the reply. Stacked signals compound. A 20-month-tenured engineer who just refreshed their skills section and started commenting on hiring posts is a different prospect from someone with only one of those traits, and only visible component scores let you later prove which signal earned the response.

Why the Open to Work badge is a trap, not a shortcut

The badge lifts reply rates because it selects people who are already active, which is exactly why it down-ranks your highest-value cohort. LinkedIn's own figures show recruiters who reached out to members with the signal saw a 14.5% positive response rate compared to 4.6% for members without it. That lift is real. So is the bias.

The self-reporting skews toward candidates more comfortable signaling availability, often because they are between roles or actively looking, while the senior performers quietly open to a move typically do not set Open to Work. A practitioner source cites candidates with the recruiters-only setting replying at roughly 37% higher rates, but that figure comes from a single blog and is not verified against a primary source; treat it as directional only.

Adoption confirms how partial the signal is. About 200 million professionals have turned on Open to Work privately, while just 40 million people signal availability publicly. And the public frame can cost the candidate: a 2024 Resume Builder survey of 1,000 hiring managers found 22% less likely to consider a candidate with the public badge, 18% positive, and 60% reporting no effect. Use the badge as one input, never the score, and never let its absence rule anyone out.

Where the false positives hide

Four common signals look like openness and are not. This is the most valuable part of the read, because a wrong reach-now wastes a throttled resource and burns a contact.

Old-role edits read as live intent. A candidate fixes a title from years ago and your tool flags "profile updated." But 19.7% of established U.S. LinkedIn users retroactively edit the title or description of a job they have already left, and the median edit happens more than four years after leaving. Tech and information workers time-travel far more than most, nearly one in three, compared with roughly one in seven in construction or real estate, and MBA holders topped every group at 29.3%. Workers under 30 were nearly four times more likely to edit an old job than those over 60. Rule-out: only recent, clustered edits count. A single fix to an ancient role is noise.

Recent starters read as open. Someone six weeks into a new job with a polished profile polished it for the move that just happened. Professionals in new roles are 3-5 times more likely to evaluate new vendors in their first 90 days, the same honeymoon that makes a fresh start date bearish, not bullish. The same "recent change" label points opposite directions depending on whether it is an edit or a start date, which is why start-date is the single highest-yield rule-out.

"Just changed jobs" treated as reach-now. That record is lagging. Job-change vendors track changes on refresh cycles - one tracks 4M+ contacts weekly, another warns that changes within the last one to two weeks may not be captured, and industry consensus is that prospect databases should be refreshed every three to six months at minimum. A provider refreshing every six weeks may be selling contacts who already moved. The person in a "just changed" record has already made the decision you wanted to influence.

Same-day blasts on a public event. When a company posts layoffs, everyone with the same alert messages the same week and your note becomes a commodity. For account-level signals, waiting two to four weeks until the noise clears can beat same-day outreach. This is a genuine disagreement in the literature: some sales sources argue act same-day on a fresh event. My read is that a live individual edit you caught yourself is worth acting on quickly, while a broadcast public event is worth waiting out.

The procedure: from a frozen list to a sorted list

Run this end to end the first time, then keep steps two through seven on a weekly loop.

The hardest part is step two: detecting dated changes at all. Reading a diff across hundreds of profiles by hand is not realistic, and buying a "changed jobs" feed only gives you the lagging signal. This is where asking for the segment directly beats scrolling. With Refolk you can request the exact cohort - tenure band, recent edit, geography - in plain English and get it back as a list, so you spend your Tuesday on the read, not the retrieval.

Scoring and the three-way cutoff

Convert stacked signals into one composite score per candidate, keep the components visible, and cut against reply rate. No recruiting-specific published model exists for this, so the framework below is mine, built on transferable principles from the dossier.

Start with a simple weighting. Give the most weight to the leading, hard-to-fake signals and the least to the noisy or biased ones.

Then apply the three bands. Reach-now is a candidate with a stacked, recent, genuine signal set and a clean start date. Watch is a single weak signal, or a strong signal you cannot yet corroborate. Skip is a false positive: under 90 days in role, an old-role edit, or a lagging already-moved record.

The cutoff is not arbitrary. LinkedIn Recruiter's reported average response rate is around 13%, and that figure doubles as a performance floor: recruiters must keep their rate at or above 13% across 100+ InMails per 14-day assessment period, or LinkedIn triggers an InMail Improvement Period. Set your reach-now cutoff where the cohort's measured reply rate clears about 13% comfortably. That makes calibration a compliance issue, not just an efficiency one.

Use these bands as reference, not as your target. Your own reply rate by band is the only number that should move your cutoff.

Sizing the pool and seeding a watchlist

Before you score, know how big the field is and where the people sit, because that shapes both your list length and your target-company watchlist. Refolk's index gives you counts and current employers you can act on.

The gap between markets is not just scale; it changes strategy. A U.S. list this large forces hard prioritization, while a German list of 22,622 is closer to something a small team can monitor more completely. One caveat from the index: a seniority-band filter on the same query returned an anomaly, so do not trust seniority splits from this source. Filter on tenure and edit signals instead.

Where those engineers work now seeds your watchlist. Follow and engagement signals only mean something if you know which companies to treat as targets or competitors.

The U.S. concentrates in big-tech incumbents while Germany's top names skew to scale-ups, which affects how you read a "started following" signal: following a scale-up carries different weight from following an incumbent everyone follows anyway.

How this goes wrong, gathered in one place

The failure modes are worth restating as a single checklist, because most of them share a root cause: reading a signal without its date or its context.

The eight traps, in short:

• Old-role edit read as live intent. Median edits happen over four years after leaving.
• Recent starter read as open. Under 90 days is settling in, not leaving.
• Open to Work over-weighted. The badge inflates the pool with active seekers and misses quiet movers.
• "Just changed jobs" treated as reach-now. That is lagging by the refresh cadence.
• Blended score hides the driver. One number cannot tell you which signal earned the reply.
• Same-day blast on a public event. Waiting two to four weeks can beat the commodity note.
• Cutoff set by gut. A reach-now band drifting under 13% triggers throttling.
• AI-added skills mistaken for upskilling. Pair skill adds with corroborating activity.

Writing the reach-now note

Once a candidate clears the read, the message should exploit what you learned. Short and specific beats generic: InMails under 400 characters have 22% higher response rates than average, yet only 10% of senders keep messages this short. And LinkedIn message steps outperform email in recruiter sequences, 16.4% candidate reply versus 4.9%. LinkedIn also refunds one InMail credit for every message answered within 90 days, so a tight, well-targeted note protects your budget twice.

Hi [first name] - noticed you have been building in [specific area from their rewrite] at [current company]. I am hiring for a [role] that leans hard on exactly that, and I think the scope would stretch you past where you are. Worth a 15-minute call this week? No pressure if the timing is off.

Do not mention that you have been monitoring their profile edits. Reference the substance of what changed, not the fact that you watched it change.

Keeping the read current

This is a loop, not a one-off. The baseline you froze goes stale the moment someone edits again, so re-run change detection weekly and rebuild the baseline each cycle. Two things drift and need scheduled attention.

First, your cutoff. Track reply rate by band every month and move the reach-now threshold so that band keeps clearing the 13% floor. If reach-now drifts below it, tighten the cutoff; if watch is out-replying reach-now, your weighting is wrong and the fix is to look at which stored component scores the repliers shared.

Second, the signal set itself. Retroactive AI-term additions surged, and if a new skill becomes a mass keyword-chasing behavior, its predictive value collapses; that is why corroboration is a permanent rule, not a one-time check. Re-check any vendor claim against your own data before you trust it. One vendor states the best candidates stay on the market for about 10 days once they start actively searching, but that is a single vendor claim and unverified independently. Do not build your response-time policy on it; build it on the reply rates you measure yourself.

The precise per-vendor lag for profile-change events is not established publicly, so treat any purchased change feed as lagging by default and prize the live edits you detect yourself. That is the whole shape of this framework: catch the diff early, rule out the fakes, stack what survives, and cut where your own replies say the floor is.