Recruitment channels get added one stalled role at a time. A search drags, so someone posts the role on a specialist board or signs an agency for it, and when the role closes the new channel stays on. Channels rarely come off the list. A year later the team runs more of them than it has people to look after, and each extra one gets tended on whatever afternoon is spare.

I’d cap the list with one rule: each channel needs an owner, a named person with time in their week set aside for it. For a team of three or four recruiters, that comes to four to six channels. To choose which ones, I’d count how many of each channel’s candidates reach the hiring manager’s first interview, because one quarter’s hires are too few to rank channels on.

Count the recruitment channels you already run.

Start with an honest list of every way candidates reach your team today. Include the ones your team stopped thinking of as channels, such as a post still live on a board or an agency still on retainer for a role you filled months ago.

Next to each one, write who owns it and what owning it takes in an ordinary week. This is the version I’d start from:

Then add up the time in the middle column for every channel you run, and set it against the hours your recruiters have left once screens and debriefs are done. The channels you can run well are the ones that fit inside those hours with a named owner.

The arithmetic I’d use starts with time: a channel run properly takes a few hours of its owner’s week, and each one you keep also has to be counted and reviewed every quarter. Once screens and debriefs are done, I’d expect every recruiter to have room to own one channel at that standard, and one or two of them to have room for a second. On a team of three or four recruiters, that comes to four to six channels. A team of one or two should expect fewer, and a larger function should give each recruiting team or role family its own list.

Judge each channel by who reaches the hiring manager.

Once you know what each channel costs to run, you need one number for what it gives back. The usual answer is source of hire, the share of your hires that came through each channel. I’d still read it once a year. For choosing channels from one quarter to the next, though, it rests on too few people.

Spread a quarter’s hires across six or eight channels and most channels end up with a handful each, so a single hire more or less reorders the ranking. A channel that brought in two hires last quarter and one this quarter has told you very little about itself.

So I’d move the count one stage earlier: for each channel, count the candidates who get to the hiring manager’s first interview, and divide what the channel cost that quarter by that count. By then a recruiter has screened every one of them, so the count leaves out applicants who were never close to the role, and a quarter holds enough of these interviews for the gaps between channels to show.

Application counts point the wrong way. The channel that sends the most applicants can be the one eating the most screening hours, and a ranking by volume would tell you to give it more.

Put recruiter hours into each channel’s cost alongside invoices and fees. Referrals and outbound sourcing cost little in money and a lot in time, and a job board can cost the reverse. A cost in money alone flatters the channels your recruiters work hardest at.

Fix the source field before you read it.

Source of hire and the count of hiring manager interviews both come from the source field in your ATS, so they are only as good as the rules for filling it in. I’d write these four down and share them with everyone who touches a candidate record:

• One source per candidate: the first channel that put them in front of your team.
• The recruiter sets it at the first screen, instead of the candidate picking from a dropdown on the application form.
• A referral gets the credit when the employee named the candidate before they applied, and an agency only for candidates it submitted first.
• A past applicant who comes back keeps their old record, with the new channel noted beside it.

Without rules like these, the field fills with whatever was quickest to click, and the channel review ends up arguing about the data instead of the channels.

Keep four to six, and give the rest an end date.

Once every channel has a cost and a count of hiring manager interviews beside it, the choice fits in one meeting. I’d keep a channel when it gets candidates to the hiring manager at a cost its owner can carry, judged on the role families you hire most often.

I’d also keep one channel that reaches people no other channel on the list reaches, such as campus events for graduate roles or a specialist community for one hard engineering team, as long as someone writes down why it stays. The rest get end dates: the day each one’s spend stops and its posts come down.

A channel you cut can come back as a trial. Give the trial an owner and a fixed length of at least one full hiring cycle for the roles it targets. Agree before it starts how many candidates it has to get to a hiring manager to stay, so the decision at the end is already made.

If your roles split into families that fill from very different places, such as engineering and hourly support, the four to six can differ by family. Count owners across all of the lists, though, because every channel draws on the same recruiters’ weeks.

Where Metaview fits in recruitment channels.

Even a recruiter who sets the source field at the screen sets it from memory of what the candidate said a few minutes earlier, which makes it the weakest input to a channel review.

Once the candidate and the interviewer have both consented to recording, the Metaview Notetaker sits in as a visible participant and makes a recording and transcript of the recruiter screen. Afterward the call becomes a set of structured notes for the recruiter.

At the channel review this pays off in one field. Metaview captures every spoken word, so whoever checks a candidate’s source reads the candidate’s own answer about where they first heard of the role, word for word.

Job boards and your careers page show candidates the same post for a role, so a post that misdescribes the role drags down both channels’ counts at once. Metaview’s Job Posts creates a tailored job post for a role from its intake call, or from a blank page or a template. A post created from the intake call starts from what the hiring manager said in that conversation.

1

2

1. 1The tags under the role title, including the pay range.
2. 2The “What you’ll do” section, written as the work the hire will run.

I’d read a post like this to the hiring manager before it goes up anywhere, because a board can only send you people who match what the post says.

In the ownership table, outbound sourcing is the row that runs almost entirely on recruiter hours, and AI Sourcing belongs on that row. The sourcing agent takes one prompt in natural language, searches the open web and your connected ATS along with the conversations your team has recorded in Metaview, and gives back a set of candidates with the agent’s reasoning shown on every one. The agent’s work ends at that list, and your recruiters take it from there, using the reasoning to decide who fits and who hears from the team.

On your channel list it is one more owned channel, and I’d judge it the way I judge the others, by how many of its candidates the hiring manager goes on to meet.

Picking an AI Filter from the pre-set list in the manual filter flow prefills the criteria the agent applies, and the search stays inside whatever your manual filters left.

1

2

1. 1The breadcrumb, “Sourcing / AI Filters,” showing the filters sit inside a search in Sourcing.
2. 2One filter, “Location,” set to “United States.”

Name your four to six before the next role opens.

Write the list down where the whole team can see it, and give every channel you keep four lines:

• The channel and its owner.
• The role families it serves.
• How many candidates it should keep getting to a hiring manager each quarter.
• The date it is next reviewed.

Add each trial channel to the same page, with the count it has to reach beside it, so the whole team can see what it is being judged on. Reopen the list when an owner leaves or the team starts hiring a kind of role it has never hired before, and let each channel’s review date handle the rest.

If your list comes out at three, add outbound sourcing first, ahead of another job board, because it’s the channel you can point at a hard role this week. If it comes out at eight, cut the channels whose owners can’t say when their last candidate reached a hiring manager, and keep cutting until the channels left fit inside your recruiters’ weeks.