Active sourcing: how to proactively find top-quality outbound hires.

An active search for a hard-to-fill role tends to begin with the job title. The list that comes back is full of people whose profiles say the right things, and the hiring manager passes on most of the first batch, because nothing on those profiles shows that anyone has done the work the team needs done next.

An agent built on artificial intelligence (AI) searches from whatever it is given, so in AI active sourcing the starting point decides even more of what comes back. I’d start from the work: what the new hire has to deliver in their first year, and where someone has already delivered it. A finished project leaves a trail that a profile doesn’t, in a product launch or a conference talk, and the people at the end of it are the top-quality outbound hires I’d go looking for.

Name the work before you name the person.

Active sourcing means going out to find people for a role that is open now, instead of waiting for applications. Passive sourcing engages people who fit a specific kind of role before they start looking for one. Most of the people an active search finds aren’t looking either: what makes it active is the open role.

A job description lists what the hire should be. For an active search, I’d ask the hiring manager a different question when the role opens: what does this person have to get done in their first year for you to be glad you hired them? Ask for one or two pieces of work, and record each one as a result someone already delivered at another company.

You can check the right-hand column in a way you can’t check the left. If a team has done that work, someone on it did, and that person is who you’re looking for. One or two pieces of work is enough to search from. When the hiring manager adds a third, it is often a trait dressed as a task, such as “owns stakeholder relationships,” and it narrows the search without telling you where to look.

Find the teams where that work has already been done.

Work done at another company rarely stays private for long. It turns up in public, usually with a date attached, and each kind of trace tells you something different about who might have done it.

I’d read old job posts first. They point at a team while the capability was still being built, which is often a year or more before anyone gives a talk about it, and the people who did that work may now want to do the next version of it somewhere new. I think most searches skip them, because a job post from two years ago reads like history rather than a lead.

Keep the list of teams short and specific. Write each lead as a team and its work, with the date beside it, because that is something you can follow. A company name on its own sends you back to searching titles inside it.

Work back from the team to the people who did it.

A trace usually names a team, or one person speaking for it, and the person you want may be neither. I’d work back to names in this order:

1. Start with whoever the trace names, such as the speaker or the author, and read who they thank. A launch post or a talk often credits the people who built the thing.
2. Check who was on that team while the project was under way, using the dates on their profiles. Someone who joined after the launch didn’t build it, however closely their profile echoes it.
3. Ask people who have since left that company who built it. Treat the answer as a lead to check against what was published, and keep it out of your first message.

Write the work and its date beside each name, because those are what the first call checks. On that call I’d ask the person to walk through one decision they made on the project. The person who made the calls can say what they’d change if they did it again, while a colleague who watched from nearby tends to retell the launch post.

Some of the people at the end of a trail may already have a record in your team’s applicant tracking system (ATS) from an earlier role. Candidate rediscovery, approaching a past candidate again, starts from how their last process ended.

Send the hiring manager the work itself.

When the first names are ready, I’d send the hiring manager the traces before the profiles: the talk or the launch post itself, with the person’s name and dates beside it. Reading a profile, a hiring manager weighs titles and employers. Reading the work, they tell you which work they’d want repeated, and that answer steers the next round of the search better than a yes or no on a name.

Where they say a piece of work is exactly what the role needs, look for the rest of that team, and for other teams that did the same work. Where they say it’s close but at the wrong scale, or for the wrong kind of customer, your description is missing that detail, so add it before you look again.

Keep the descriptions the hiring manager approves. A trail you followed once is a head start when the same kind of work comes up again, because the teams that did it are still the first places to look.

Where Metaview fits in AI active sourcing.

The first call is where this method finds out whose work it was. Ask the candidate beforehand whether it can be recorded, in the way your recording policy sets out. On a video call they agree to, the Metaview Notetaker joins as a separate guest in plain view of the candidate. Its recording comes back as a transcript and structured notes. Metaview captures every spoken word. Later, the hiring manager can read the candidate’s account of their one decision next to the launch post you sent them.

1

2

1. 1The template this call’s notes follow, named above the notes.
2. 2A template matched by the meeting’s name, “Sales Culture,” rather than by its interview stage.

Siadhal Magos, Metaview’s chief executive, traced disagreement about good hires to missing data when the company raised its Series B:

For one role, the missing data is often a plain record of what the hiring manager wants done and what each candidate says they did.

With the description agreed at the intake, the recruiter writes it as one natural-language prompt for Metaview AI Sourcing, such as people who built a design system that more than one product team adopted. The agent looks in your team’s Metaview recordings and your connected ATS, and it searches the web too. Beside every candidate, it explains why that person came back.

The candidates it returns are where the recruiter’s date check begins. The recruiter reads each reason for the project behind it, then decides who fits the role and writes the first message to each person they pick.

A few of the people worth checking may already have talked to your team. Reports answers questions about what was said in your team’s recorded interviews, so a recruiter can look for past candidates who described work like this, with the date where they gave one, and start the date check with people the team has met.

1

2

1. 1The plus sign after the last column heading, where the report has room for another column.
2. 2Timothy S., whose answer came from a conversation called “Sales Jam,” not a screen or an interview.

Test the next shortlist against the work.

Before the next shortlist goes to the hiring manager, I’d run one test on it. Beside each name, write the work and its date: the piece of work that put the person there, and when it was done. A name you can’t fill in came from a title, and a shortlist that leans on titles has drifted back to the search this guide set out to replace.

When that happens, return to the hiring manager’s account of the work before you search again. A vague description is usually how titles creep back in, and tightening it with the hiring manager costs less than another batch they pass on.

A top-quality outbound candidate found this way is a person you approached because of something they had already built, with the dates checked and the hiring manager’s read of it on record. A shortlist where every name carries its work and its date is one the hiring manager can judge on the work.