The colleague who knows the company you want to join may be someone you have not spoken to in years. The useful connection is easy to overlook: they are no longer part of your daily work, but they remember a version of it that a new contact has never seen.
Fresh US data puts that continuity in perspective. The Bureau of Labor Statistics reported on 24 September 2026 that median tenure with the current employer was 4.1 years in January 2026, compared with 3.9 in January 2024. For workers aged 25–34 it was 3.0 years; for those aged 55–64, 9.6. These are elapsed durations in current jobs, not predictions of when anyone will leave. [1]
The figures do not prove that people are moving faster than ever. They show why a working life can contain several distinct sets of colleagues—and why a current organization chart represents only part of a professional network.
A large experiment supports the value of less-close ties
A study published in Science in 2022 tested variations in LinkedIn’s connection recommendations. The research covered more than 20 million people over five years, during which participants formed about two billion new ties and experienced roughly 600,000 job changes. Those totals describe the study’s scale, not jobs all caused by the experiment. [2]
The relationship between tie strength and job transmission was nonlinear. The result depended on how strength was measured and varied across industries; some moderately weak connections were especially useful, while stronger ties mattered more in less-digital industries. That is more specific than “strangers are always the best people to ask.” [2]
The study supports a causal role for changing access to parts of a network in that setting. It does not turn every unanswered acquaintance into a job opportunity, or establish that adding the largest number of people is the best personal strategy.
An old colleague can combine familiarity with different information
The useful feature of a former colleague is easy to describe without assigning them a network score. They may know your work and now spend their time somewhere else. That combination can make a conversation informative even before anyone discusses a referral.
A strong first question is therefore about the situation they know: what has changed in their field, what a role actually involves, or what they wish they had understood before moving. It gives them something concrete to answer and gives you information for deciding whether the opportunity fits.
A vague request to “pick your brain” shifts the work of defining the conversation onto the recipient. A specific question, grounded in a real connection, makes the exchange easier to assess.
For example, this editorial illustration supplies context and a bounded request:
“We worked together on the onboarding redesign. I’m looking at a similar role in your sector and wondered which part of that work matters most now. A short reply would be helpful; no need for an introduction.”
Adjust the wording to the relationship and the recipient’s preferences.
Keep the evidence at the right level
| Finding | Reasonable implication | Unsupported leap |
|---|---|---|
| Median elapsed tenure was 4.1 years in January 2026 | Your past workplaces may remain relevant sources of context. | Predicting that the average person will leave after 4.1 years |
| Recommendation changes affected job transmission in a large platform experiment | Access to different connections can matter. | Guaranteeing an outcome from a particular message |
| The tie-strength relationship was nonlinear | Seek relevant connections rather than assuming the weakest is always best. | Treating contact count as a performance target |
The evidence notes keep the population, period and unit of each finding separate.
Recover the context before making the request
Before reaching out, find the last exchange you actually had. Check the person’s current role from a source they maintain or from your recent conversation, rather than assuming the old email signature is still correct.
Then identify what connects you. A shared project, a question they helped answer or a previous offer can make the reason for contact clear. An old offer to introduce you is context, not standing permission to use their name indefinitely. Ask again when circumstances have changed.
A private note can make this easier: how you worked together, what you last discussed and what you plan to ask. Keep it factual and limited to what is useful. There is no need to turn a job search into a database of private assessments about everyone you know.
Kluro can help retrieve the original exchange from connected history. The find-people-who-can-help page demonstrates that task. The decision to contact someone and the wording of the request remain yours.
Leave room for a relationship after the application
A useful conversation does not have to end in a referral. You may learn that the role is a poor fit, that the team needs different experience, or that the timing is wrong. That can still save effort and support a more informed decision.
If someone helps, thank them for the specific contribution. Where appropriate, close the loop later. Do not treat silence as an invitation to escalate across multiple channels. A person’s ability to help can be limited by workload, policy or circumstances you cannot see.
The enduring advantage of past colleagues is often the shared work itself. Recovering that context is a better starting point than trying to imitate a large platform’s network statistics.
Method and dates
We reviewed the September 24, 2026 BLS release, whose tenure measurement refers to January 2026, and the 2022 randomized LinkedIn research. We did not collect hiring outcomes, run an outreach experiment or estimate job-search success. The two datasets concern different populations and cannot be pooled. All practical examples are editorial recommendations rather than experimentally validated instructions.
Sources
- Employee Tenure in 2026 — US Bureau of Labor Statistics. 2026-09-24. Checked against the 28 September 2026 research cutoff.
- A causal test of the strength of weak ties — Rajkumar and colleagues / Stanford Digital Economy Lab. 2022-09-15. Checked against the 28 September 2026 research cutoff.
