Job boards, applicant tracking, sourcing tools, AI screening, agencies, RPO. Each one takes a piece of hiring off your desk and leaves the rest. Once you see which piece each keeps, the whole market stops looking crowded.
You paid for a sourcing tool last month. The invoice cleared, a list of profiles landed in your inbox, and the actual hiring, the outreach and the screening and the scheduling, was still sitting on your desk. That gap is the whole story of the hiring market, and one question brings it into focus: how much of the work does a vendor take off your desk, and how much do they quietly hand back?
The market splits in two. On one side is software you buy and run yourself: job boards, applicant tracking systems, sourcing tools, AI screening. On the other are services that do the work for you: staffing agencies and RPO. Software is cheaper and leaves the labor with you. Services take the labor on and charge a multiple for it. Price is not what really separates them. What separates them is how much of the job is still yours once the tool is running or the contract is signed.
Start with the software you run yourself. A job board sells reach. You post a role, it goes in front of a large audience, and applications come in. That is the whole product: a faucet, not a filter. You still write the description, screen the flood, reply, schedule, interview, and close. Broad boards are built to maximize applications, which means high volume and low signal, and the cost of sorting through it lands on you.
An applicant tracking system is your system of record. It stores applications, moves candidates through stages, holds your compliance data, and keeps interviewers coordinated. It is essential, and on its own it adds nothing to the top of your funnel. Put noise in and you get well-organized noise out.
Sourcing tools find the people who are not applying. They search large databases of professional profiles, surface passive candidates, and pull contact details, and the good ones do it well. But what comes back is a list. You still qualify it, write the outreach, send it, chase the replies, and run every conversation yourself. The newer agentic tools speed up the finding, then hand the shortlist back to a person for the judgment, the interview, and the close.
AI screening automates the middle: chat screening, scheduling, assessments, scoring. It takes the logistics and leaves the decisions. A score can rank candidates, but it cannot pick the one you should hire, and the reasoning behind it is hard to audit. The criteria can quietly stand in for the wrong things, and when a call goes wrong it is the employer, not the vendor, who carries the outcome and the legal exposure. Someone on your side still has to confirm the top-ranked people are actually any good.
Now the services. An agency takes the role and runs the search end to end. It sources, screens, interviews, and hands you a shortlist of vetted, interested people. Of every category here, it removes the most per-hire work, and it charges the most for it: commonly 20 to 30 percent of first-year salary, which on a senior hire is tens of thousands of dollars for a single placement. Quality rides on the individual recruiter. Contingency firms work without exclusivity, so they lean toward the roles that are easy to fill and often send the same candidates to several clients at once.
RPO goes further and runs your recruiting function as an outsourced operation, priced as a monthly management fee or a rate per hire. It removes more than anything else on this list, the whole function rather than a single stage. It also embeds deep, carries real switching costs, and only pays off at sustained volume.
Line the market up by what it hands back to you, with Metix AI first.
| Metix AI | Self-run software | Agencies | RPO | |
|---|---|---|---|---|
| What you pay for | Interview-ready candidates on your calendar | Tools, reach, and access | A finished search and a shortlist | Your whole recruiting function |
| Pricing model | Per interview-ready candidate | Per seat or per company headcount | 20 to 30 percent of first-year salary | Monthly fee or per-hire rate |
| Who runs the search | The platform sources and screens; a human checks quality | You do | The agency's recruiters, by hand | An embedded outsourced team |
| What it leaves on your desk | The interview, the decision, the close | Everything after the tool | The decision, negotiation, onboarding | Hiring decisions, planning, governance |
| On a hard role | Point Metix AI at it; if no one clears your bar, no credit is spent | Same price, easy role or hard | Often deprioritized for easier roles | Absorbed into ongoing volume |
The column that decides everything is the one nobody markets: what still lands on your desk.
Read down that last column and the same gap shows up in every row, the one no vendor puts on a slide: each category hands back the finished candidate and all the work of getting that person ready to interview. The board leaves you the filtering, the ATS the sourcing, the sourcing tool the conversations, the screener the judgment call. They all stop at the same place.
Picture a founder hiring her first senior backend engineer. She posts on two boards and gets three hundred applications, most of them off-target. She buys a sourcing tool and pulls eighty promising profiles. Her ATS keeps all of it tidy. A screener scores the handful who reply. Three weeks in she has spreadsheets, scores, and a full pipeline, and still not one engineer she trusts enough to put in front of the team. Every tool did its slice. The hire is still hers to make.
One category does more than that. An agency actually assembles the shortlist: vetted, interested people ready to interview. So the honest version of the claim is sharper than saying nobody sells this. People do sell it. They just sell it the expensive, hand-built way, one recruiter at a time, at a percentage of salary, and they favor the roles that are easy to fill. Getting the right people to the table, interested and ready, is where the value actually sits. The market has priced it like a luxury and shipped it like a craft.
You can search a hundred million profiles in a second and still walk into Thursday without five people worth interviewing. Speed arrived everywhere in hiring except the part that decides the hire: getting the right few to the table, ready to talk.
If the finished hire is the thing almost nobody hands you, then how a vendor charges is the clearest signal of what they are really selling. You do not have to trust the pitch. Read the meter.
A per-seat price is a tool for your recruiters: you rent the software and bring the labor. A price tied to company headcount is infrastructure, so the bill climbs as you add people anywhere, whether or not you are hiring. Pay-per-click is raw traffic. Only a fee tied to a hire, a percentage of salary or a flat rate per placement, charges you for a result instead of for access.
Two of those quietly charge you for cost rather than value. An ATS billed on total headcount raises your bill every year even when hiring is flat, because you grew somewhere else. Enterprise screening and sourcing suites tend to add implementation fees and annual renewal bumps. You feel the price of the tool long before you feel the value of a hire.
Paying for outcomes fixes the deepest flaw in software pricing. Under a seat license a vendor gets paid whether or not you ever see value, and their most profitable customer is the one who pays and barely logs in. Tie that revenue to a delivered result instead, and the whole company has to reorganize around getting you the result rather than driving usage. A vendor willing to be paid on outcomes is also saying something it cannot fake: it believes it can deliver.
There is a catch most decks skip. Outcome pricing is only as honest as the metric it is built on, and a loose metric gets gamed. Pay a support vendor per resolved ticket and you will see tickets closed that reopen the next morning. Hiring has its own version: pay per qualified candidate and a careless vendor loosens the definition of qualified until your inbox fills with people who pass on paper and fail in practice. Then there is the oldest move in pay-for-results, creaming, where the provider pours effort into the easy cases and parks the hard ones. In hiring that means your junior reqs fly while your senior and niche roles, the ones you needed help with, sit untouched.
So there is one question worth putting to any outcome-priced vendor: what exactly is the metric, who checks it, and what stops you from loosening it or skimming the easy work?
When we started building Metix AI, we kept coming back to a question no tool on the market answered: why is hiring sold in pieces when what a company actually needs is the finished hire? The two problems above, the step nobody finishes and the metric everyone games, are not separate. Solve one carelessly and you make the other worse. So we designed for both at once.
Metix AI runs the search in software and puts a human on quality before a single candidate reaches you. You define the role, and the platform does the volume work: it searches across more than 860 million profiles, matches, reaches out, and runs the first-pass screen. That number is where the work starts, not the product. What matters is the step after it. A human delivery lead reviews quality before anyone reaches you, and nothing goes out until you approve the outreach yourself. Because the software carries the volume, the human layer stays thin and consistent: one quality gate held to the bar you set, not a different recruiter improvising each search. You do not run a tool, and you do not manage a recruiter. What arrives on your side is a calendar with interview-ready candidates on it.
Here, interview-ready is a specific bar, not a slogan. It means the candidate matches the role's must-haves, has confirmed they are interested, and is booked on your calendar. You own that bar, because you approve the outreach that goes out under it, and a person checks every candidate against it before delivery.
The pricing sits on that same bar. You pay per interview-ready candidate. A credit is a person, not a click: one credit is one interested, hire-ready candidate booked for an interview. If nobody clears the bar, the credit stays unspent. Plans run from $49 to $299 a month, ten percent off on annual plans, and a credit only does its work when a qualified person actually reaches you. No retainers, no placement fees. A single senior placement fee at an agency can cost more than a full year on a Metix AI plan.
You might be looking for the catch, since we just spent a whole section on how outcome pricing gets gamed. Fair. Worried a vendor could quietly lower the bar to burn through your credits? You approve every message that goes out, so the bar is yours, not ours. Worried about a flood of people who are technically qualified and obviously wrong? A person checks every candidate before you see them, and anyone who does not clear the bar does not cost a credit. Worried we would cream the easy roles? You point Metix AI at the one role you need filled, so there is no portfolio of easy wins to hide in. Bring your hardest req. If nobody clears your bar, you have paid nothing.
The numbers worth watching are about the finish, not the middle of the funnel. These are early, drawn from the roles Metix AI has run so far, and they move with the role and the market:
These come from finishing the job instead of selling a slice of it. Product Hunt named Metix AI its number one Product of the Day, and it took a 2025 HRTech INNO Award. Nice, but the number that actually matters isn't on the list at all: the hire got made.
Bring a role you actually need to fill, ideally a hard one. Metix AI runs it as a free trial: one real role, no card. You define it, you approve the outreach, and you watch interview-ready candidates land on your calendar. Then you decide.
Hiring software got powerful. Hiring outcomes did not. Metix AI is built to close that gap.