In November 2017 Plamen Ivanoff published a LinkedIn piece called "Contingency Recruitment is Dying. Will You Die with It?" He still offers an ebook on the subject. Its landing page announces that "The death of contingency recruitment is upon us!" and promises to explain why "Contingency Recruitment WILL Die".
Nine years is a long time to be dying.
The argument is always built on one pair of numbers: retained search completes around 90 percent of its assignments, contingency completes a fraction of that, so move your desk upmarket. If those numbers hold, the debate is over. They are worth checking, because a surprising amount of the recruiting industry's self-knowledge turns out to be copy-paste.
Page one cannot agree with itself
Search for fill rates by model and count what comes back. Google's first page in July 2026 returned nine results. Five carried numbers.
| Source | What it claims | What it is actually measuring |
|---|---|---|
| Retrained Search | 25% contingency, 97% fill rate for its own approach | A recruiter's fee-conversion rate, not a client-side fill rate |
| Louise Archer, LinkedIn, December 2025 | Contingent fill rate 15 to 25%, retained 90 to 100%, described as "an 80% failure rate" | Share of a contingent recruiter's work that earns nothing. The 80% is 100 minus her own 15 to 25, in pounds |
| JRG Partners, July 2025 | Contingency 20 to 35%, retained 85 to 95% | Calls both a "completion rate" and a "success rate," never defines either |
| Fusion Recruiters, November 2025 | Contingent searches 25 to 35%, and a 95% fill rate for itself | The only one of the five that says "fill rate" and means the client-side measure |
| An unattributed template | 10 to 25% contingency, 90 to 100% retained | Unknown. This text appears verbatim under at least nine different LinkedIn accounts |
Three things about that table matter more than any single number in it.
The five sources are four businesses. Louise Archer founded Retrained Search and writes its content, so rows one and two are one commercial interest, and they contradict each other: the company page says 25 and 97, her post says 15 to 25 and 90 to 100.
They are not measuring the same thing. Two say completion rate without defining it, one means the client-side fill rate, one means how much of a recruiter's own effort converts to a fee, and one measures unpaid hours in a different currency. Stacking them into a single comparison, which is what every article in this genre does, is the error before the sourcing error.
And not one of them cites a study, a survey, a sample size or a year. JRG Partners carries a single outbound link, to an SHRM topic hub, unattached to any of its numbers.
Every source here has a commercial interest in the retained answer. Two sell retained-search training to recruiters. One sells retained search. Fusion, to its credit, pitches contingency in the same article it publishes those numbers in.
Where the 90 percent comes from
The trail ends at one name: ExecuNet.
Be careful here, because the easy dismissal is wrong. ExecuNet is not a content farm. It published the Executive Job Market Intelligence Report annually for more than two decades, surveying executives, hiring companies and recruiters, and the national press quoted it regularly. It ran at least twenty-one consecutive annual surveys.
The problem is narrower and harder to wave away. No ExecuNet publication containing a 90 percent retained completion rate can be located. There is no report title, no year, no page, no document. The citation names a real research organization and points at nothing, and it has since been copied into AI-generated HR glossary pages, which is where most recruiters now meet it, sitting beside four or five other bare organization names with no documents attached either.
Nobody who holds the data publishes a completion rate
Two organizations are positioned to settle this and neither has.
The Association of Executive Search and Leadership Consultants surveys its members at scale, and it does not put the results in public. Its last public State of the Profession report discloses a healthy sample, more than 500 member firms plus 300 clients and 1,300 candidates, and contains charts for average searches per consultant and average fee per search. Both are rendered as images with no values printed, on pages stamped members-only. That report carries 2018 data, and as of July 2026 nothing has replaced it in the public insights feed, which now runs press releases, member news and thought leadership. The association's current benchmark research, a three-year global benchmark survey fielded in 2025, is a members-only report.
Be precise about what that does and does not prove, though. Those two withheld charts are workload and pricing. Neither is a completion rate, and no combination of them produces one. What AESC keeps behind the login is revenue data. As far as I can establish it has never published a completion rate at all, which means the 90 percent figure is not being withheld by the association. It simply does not appear to exist anywhere.
Bullhorn runs one of the largest agency applicant-tracking datasets anywhere. Its glossary defines fill rate simply as job orders filled divided by job orders received, with no time bound. Its Hiring Outlook series uses a stricter measure, the share of job orders placed within 90 days of opening, on a three-month trailing lag, and reports the change rather than the level: permanent fill rates in the month ending June 2026 ran 11 percent above the same month in 2025. The absolute level does not appear in that series. Its GRID report does publish one threshold, noting that firms with AI embedded across the workflow are more than twice as likely to report fill rates above 75 percent, which is self-reported and a band rather than a measurement.
So the two best-positioned sources in the industry offer a members-only image and a percentage change.
The buyer-side number, and the awkward thing about it
There is one quantified figure that comes from the people paying for searches. The Executive Search Information Exchange, a benchmarking community of in-house executive recruiting leaders running since 1996, reports that about 40 percent of retained searches fail to complete, with a companion finding that clients cause roughly 70 percent of those failures.
It has a stated basis, which the 90 percent does not. Simon Mullins, who leads ESIX, published it in 2018 describing the source as fourteen years of benchmark surveys and more than 200 membership meetings, and restated it in 2020 as almost twenty years of surveys and more than 400 meetings. There is still no sample size, and it is a six-to-eight-year-old figure, so date it whenever you quote it.
Now the awkward part, which most people repeating this number do not mention. Mullins published it through ExecuNet. The same organization is the cited origin of the 90 percent completion rate and the distribution vehicle for the 40 percent failure rate, and the two point in opposite directions. That is not an argument for either number. It is the clearest possible illustration that this industry recycles statistics without checking them.
One more caution, because it applies to us as much as anyone. The 40 percent reached me through a retained search firm's blog, where it functions as the setup for a pitch: searches fail, the argument goes, because other firms do not do real research, so hire us. It is not published against commercial interest. The reason to weight it above the 90 percent is narrower and duller: ESIX surveys buyers rather than sellers, and it says what its basis is.
Decide on arithmetic instead
The model question has an answer no ebook will sell you, because it is arithmetic and it is specific to your desk.
Call it the expected fee per hour. Take the fee, multiply by your honest probability of collecting it, and divide by the hours the assignment will consume whether you win or not.
Work it on an $80,000 role, which at Staffing Industry Analysts' surveyed median direct-hire fee of 20 percent is a $16,000 contingency fee. At a one-in-four hit rate that is $4,000 expected. Spend 40 hours on it and you have earned $100 per expected hour.
Now convert the same client to a retained or engaged arrangement at 28 percent, or $22,400. Apply the industry's own 40 percent failure rate and assume you keep roughly two of three installments on a failed search, and the expected value is about $19,400. Against the same 40 hours that is $485 per expected hour, which looks like the argument for going upmarket, and it is exactly where the sales pitch stops.
Do not stop there, because the hours are not the same. The commonly quoted effort figures, themselves unsourced, put retained search at 200 to 400 hours per assignment against 20 to 60 for contingency, and that gap is the entire justification for charging a retainer. Divide $19,400 by 200 hours and you get $97, which is a wash with the contingency desk. The break-even sits near 194 hours.
That is the honest shape of it. The model does not decide the outcome. The hours do, and so does your own hit rate, and neither of those numbers is in anybody's ebook.
Two things follow. Your own hit rate over your last twenty assignments is the input that matters, and an industry average tells you very little about your desk even when it is real. And hours committed is the variable with the most give in it, which is the one worth attacking.
What the desk actually earns
The stakes are worth stating, because they are lower than the ebooks assume.
Top Echelon's 2026 State of the Recruiting Industry Report found that 48.2 percent of responding recruiters brought in under $200,000 of desk cash-in during 2025, roughly flat year over year, with fewer clearing $500,000 than the year before. The largest single group made one to five placements in the year, and 14.6 percent expected to make 26 or more in 2026. Recruiterflow's data from more than 2,100 firms shows how wide the spread inside the profession runs: 5.21 placements per recruiter in the top quartile against 1.38 for everyone else, though those are its own customers rather than a market sample.
Read it with two caveats, the second of which is not in the report. It is self-selected from a split-fee network and 62.8 percent of respondents have more than 25 years in the business, so it skews experienced. And the response count, which is not published, appears to be about 137: every percentage on the page is an exact multiple of one 137th. One respondent moves any figure by roughly three quarters of a point, so the decimal places are decoration.
Directional, then. But a profession where roughly half of experienced recruiters run a sub-$200,000 desk is one where the hours term in that arithmetic decides whether the business survives, which is why the model question deserves a calculation rather than a narrative.
Worth reading against the other side of the ledger. Our US recruiting market report puts the median posted salary for a staffing or agency recruiter at $50,000, the lowest of any recruiting specialization, precisely because base pay understates a job where commission carries most of the earnings. The cash-in distribution above is what that commission looks like in practice, and it is far wider in both directions than a salary figure suggests. If you are weighing an agency seat against an in-house one, recruiter pay by route covers that comparison. For what each model charges in dollars by role, guarantee terms included, see our recruiting fees breakdown. This piece is about what each model returns.
The variable nobody prices: cash flow
Expected value per hour is necessary and not sufficient, because the two models fail differently.
Contingency revenue arrives after the placement starts, sometimes 30 to 60 days after that, and never if the client hires elsewhere. A retained assignment pays part of the fee before the outcome is known. Two desks with identical annual revenue can have very different survival odds depending on which one funds six months of work from savings.
Concentration is the quieter risk. In the Top Echelon data, 13.9 percent of recruiters placed with only one to three clients across the year. On a contingency desk that is obviously fragile. On a retained desk it is fragile more slowly, because losing the relationship removes a scheduled income stream rather than a lottery ticket. Either way the fix is not a fee model: it is winning clients when hiring is slow, which 60.7 percent of those recruiters named as their top priority, alongside winning new search assignments.
Engaged and container search
One honest paragraph. The hybrid models, where a smaller upfront deposit is credited against a success fee, are real, and for a priority search they are usually the sensible middle. What does not exist is any data on how common they have become. Every claim that engaged search is growing fastest or being increasingly adopted that I checked came from the website of a firm selling engaged search, with no survey behind it. Use the model if your own math supports it. Do not repeat adoption statistics that nobody has collected.
The 70/30 rule is not a rule
Google will ask you about this, because it surfaces in People Also Ask across this whole topic. The question is an artifact rather than a convention.
Four unrelated meanings circulate: hire someone who meets 70 percent of the requirements and develop the rest; spend 70 percent of your time sourcing and 30 percent on admin; weight a decision 70 percent on skills and 30 percent on attitude; and split a placement fee 70/30. The last is the weakest. NPAworldwide describes a 50/50 commission split as the most common arrangement in split-fee work while explicitly noting it is not the only one, and Top Echelon lists both 50/50 and 60/40 as options negotiated per deal. Nobody publishes a distribution.
The question ranks because it is cheap for a job board to auto-answer, not because recruiters use it.
Where Glozo fits
The variable with the most give in the arithmetic above is hours per assignment, and specifically the wasted ones: hours spent on a role that was never priced to be fillable, and submissions the client rejects or has already seen.
That is what Glozo is built to compress. Smart Search reads the intent of a search rather than keywords and returns profiles from more than 30 sources, ordered by how closely they match what the role actually needs. A market compensation estimate tells you whether an assignment is priced to be fillable before you commit the hours, which on a contingency desk is the difference between an expected fee and an unpaid month. The Open to Offers signal narrows the list toward people who look receptive, which is where a good deal of the wasted outreach goes.
None of that changes your fee model. It changes the denominator, which is the part of the calculation you actually control.
Several claims above are negative ones: that a figure has no locatable source, or that a named source published a number without citing anything. Negative claims are the easiest kind to get wrong, and the people best placed to correct them are the ones named here. So if you can produce the ExecuNet publication behind the 90 percent completion rate, or a primary source for any contingency fill rate, send it to us and we will update this page and say plainly what changed. The same offer stands for anyone whose figures we have represented inaccurately. We would rather be corrected than be one more page in this genre repeating something it never checked.