How does Via know who actually knows whom?

Via AI determines who actually knows whom by scoring relationships on real work history — not LinkedIn connections. Via uses tenure overlap, role proximity, recency, and communication signals to distinguish genuine relationships from loose social connections.

Most tools assume that if two people are connected online, they "know" each other. Sales teams know that's not true. The difference between seeing that two people are connected on LinkedIn and knowing they spent three years on the same team is the difference between a dead-end intro request and one that actually converts.

The signals Via uses to score relationship strength

Via uses public and partner data to establish who overlapped where, for how long, and in what roles — without depending on LinkedIn connection data. Each signal tells a different part of the story:

• Tenure overlap. Three years on the same 20-person engineering team is a real relationship. Three years in the same 10,000-person company might mean they never met. Via distinguishes between close collaboration and distant org-chart proximity.
• Role proximity. Two people who shared a function — both in product, both in sales — are far more likely to have worked together directly than two people who happened to be in the same building but in unrelated departments.
• Recency. A relationship from last year carries more weight than one from 2016. People change roles, cities, and priorities. Stale history counts for less so you're not asking someone to make an intro based on a relationship that faded years ago.
• Frequency of overlap. If two people worked together at two different companies over the course of a decade, that's a pattern — not a coincidence. Multiple shared stints signal a relationship that's been maintained through career changes.
• Company size at the time. Overlapping at a 30-person startup is fundamentally different from overlapping at a Fortune 500. At a small company, everyone knows everyone. Via adjusts for this so it doesn't treat a 50,000-employee overlap the same as a Series A team.

How private signals strengthen the picture

Public work history gives Via a strong baseline, but when users opt in, private signals fill in what public data can't see — and recency and frequency are where they matter most.

• Email activity. If two people exchanged emails in the last 90 days, that relationship is alive right now — not a guess based on a job that ended two years ago. Frequent email threads signal an active, ongoing connection.
• Calendar overlap. Shared meetings — especially recurring ones — are one of the strongest indicators of a working relationship. A weekly 1:1 or a standing project sync tells Via that these two people are actively collaborating, not just names in each other's contact lists.
• CRM data. Logged interactions, deal history, and relationship notes add another layer of context that public data alone can't provide.

Via never exposes your private content without your consent — it uses these signals only to strengthen confidence scores. Your emails and calendar details stay private; only the relationship signal is surfaced.

Compounding signals: why multiple layers matter

No single signal is enough to guarantee a good intro. What makes Via's scoring reliable is that signals compound. Two people who worked together at a startup for three years, still email each other monthly, had a meeting last week, and are connected on LinkedIn — that's a high-confidence relationship. The more signals that stack, the higher Via's confidence that this person can actually make an introduction — and that the introduction will be well-received. This is why Via doesn't show you every possible path. It shows you the paths where the evidence is strong enough that you can ask with confidence.

Why this matters for warm intros

Weak intros waste time, damage trust, and make reps hesitant to ask for help again. Knowing who actually knows whom — not just who's adjacent on a graph — is the difference between a warm intro that converts and a cold email in disguise.

Consider two scenarios. In the first, a rep sees that a colleague is connected to a VP at a target account on LinkedIn. They ask for an intro. The colleague barely remembers the person — they accepted a connection request three years ago after a conference panel. The intro goes out, gets ignored, and the colleague feels awkward. The rep burns a favor and gets nothing.

In the second scenario, Via shows the rep that a colleague overlapped with that same VP for two years at a previous company, they've exchanged emails recently, and they share a LinkedIn connection. The rep asks for the intro with context: "You two worked together at Acme — would you be open to connecting us?" The colleague is happy to help because the relationship is real. The VP takes the meeting because the intro came from someone they actually trust.

Same target, same rep, completely different outcome — because the second path was grounded in evidence, not guesswork.