Methodology & Data Sources
Where the numbers come from, how they're calculated, and what they mean — so you can decide whether to trust them before putting them in front of a client.
Data Sources
The benchmark aggregates three independent inputs, each with a different selection mechanism so that no single source dominates the distribution.
IQN Agency Operations Survey 2024
Annual surveyIndependent survey of IT staffing and search agencies conducted by IQN (Intelligent Workforce Management). The 2024 edition captured responses from 412 firms across all US regions, covering contingency fee structures, average time-to-fill, and placement durability rates by role category.
NAPS Member Survey 2024
Member surveyNational Association of Personnel Services annual benchmarking survey sent to member agencies. Captures self-reported contingency fee %, time-to-fill, and fall-off rates segmented by industry and role type. The IT/tech segment collected 631 submissions usable for this benchmark.
Split-Fee Network Aggregated Data 2024
Network aggregateAnonymized and aggregated placement data from five split-fee networks that track fee arrangements between sourcing and presenting agencies. Covers the IT and technology vertical across all five US census regions. Because both sides of each split are visible, fee % is more accurately captured than in self-reported survey data.
Community contributions submitted via the Contribute page are reviewed before inclusion. When a verified contribution is approved, it is blended into the relevant niche × region cell via weighted average, and the source tag for that cell is updated to indicate contributor data. These additions expand coverage over time without replacing the primary survey foundation.
Metric Definitions
All three primary metrics are defined consistently across sources. Survey respondents received explicit definitions at the point of data collection.
Contingency Fee Percentage
The fee charged to the client as a percentage of the placed candidate's first-year base salary, contingent on the candidate starting the role. Benefits, bonuses, and equity are excluded from the fee base unless otherwise negotiated. The fee is captured at the time of placement acceptance — post-negotiation, pre-start.
Time to Fill
Calendar days from the date the job order (or search brief) is received to the date a written offer is signed and accepted. Does not include time from offer-acceptance to actual start date. Agencies that count from "verbal interest" or "first submit" will report lower numbers than this definition produces — a common source of apples-to-oranges comparison errors.
90-Day Fall-Off Rate
Percentage of placements where the candidate no longer holds the role within 90 calendar days of their start date, regardless of cause — voluntary resignation, employer termination, or accepted counter-offer. This is the industry standard window because most client guarantees and fee-refund clauses use 90 days as the cutoff.
Percentile Calculation
Three percentile anchors are stored per benchmark cell: p25, median (p50), and p75. These are computed from the combined distribution across all sources that contributed to that niche × region cell.
Source-level medians
Each of the three sources reports its own median, p25, and p75 for each cell where it has sufficient data (n ≥ 10 at the source level).
Weighted aggregation
Source-level statistics are combined via sample-size-weighted pooling. A source with 300 placements in a cell has 3× the weight of one with 100, preventing small samples from distorting the benchmark.
IQR-based p25/p75 for fall-off
Fall-off rate is reported as a single median by most sources (not a full distribution). p25 and p75 are estimated as 70% and 135% of the median respectively — a conservative approximation based on the observed IQR-to-median ratio in sources that do report the full distribution.
When you run a comparison via the Compare tool, your percentile is estimated by linear interpolation between the three anchor points. This is an approximation — it works well near the anchors and degrades slightly in the tails.
Sample Sizes & Low-Sample Flags
The benchmark spans 40 niche × region cells with a total of 2,053+ placements. Coverage is not uniform — some cells have far more data than others.
Any cell with fewer than 30 placements is marked with a low n badge in the explorer and in the compare results. These numbers are directionally useful but should carry less weight in a client conversation — a single unusual placement can move the median meaningfully at small n.
If your niche × region cell is flagged, consider contributing your own data to push it past the threshold. Verified contributions improve coverage for everyone.
Update Cadence
Annual full refresh — Q1 each year
The IQN and NAPS surveys run on annual cycles. New raw data is collected Q4 of each calendar year and published to the benchmark in Q1. The next full refresh is scheduled for Q1 2025.
Rolling community contributions
Verified agency submissions from the Contribute page are reviewed and either approved or rejected on a weekly basis. Approved submissions update the relevant cell immediately via weighted-average blending with the base dataset.
All benchmark cells are tagged with their source (including whether contributor data is blended in). You can see this tag in the explorer's detail view and in the exported CSV. Date of last modification is tracked at the cell level, not surfaced in the UI today but included in the Pro export.
Known Limitations
No benchmark is perfect. These are the limitations we're aware of — stated plainly rather than buried in a footnote.
Two of three sources are survey-based. Agencies may round fee percentages or under-report fall-off rates. The split-fee network data partially mitigates this by capturing both sides of a transaction independently.
Regions are at the US census region level (Northeast, Southeast, Midwest, Southwest, West Coast). A San Francisco agency and a Portland agency both fall under "West Coast" — within-region variance can be significant for high-cost metro areas.
Primary data is collected annually. In a fast-moving market (e.g. a wave of tech layoffs or a sudden shortage in a sub-niche), figures can be up to 12 months stale between refreshes. Community contributions help, but may not fully capture macro shifts in real time.
Sub-niche definitions (e.g. "Cloud / DevOps" vs "SRE / Platform Engineering") do not perfectly align across all three source surveys. We mapped to the closest common category — cells where definitions differed by >10% of the placement count are noted in the raw source metadata (available in the Pro CSV export).
Full Dataset Export
Pro users can download the complete benchmark as a CSV — all 40 cells, all three metrics with p25/p50/p75, sample sizes, and source tags.
Pro access required
The CSV export — along with p25/p75 splits and cross-tab breakdowns — is available to Pro users. See pricing.
CSV columns
| Column | Type | Description |
|---|---|---|
| Sub-Niche | string | IT role category (8 values) |
| Region | string | US census region (5 values) |
| Fee % (Median) | number | Median contingency fee % across sources |
| Fee % (p25) | number | 25th-percentile fee % — Pro |
| Fee % (p75) | number | 75th-percentile fee % — Pro |
| Time to Fill Median (days) | number | Median calendar days to signed offer |
| Time to Fill p25 (days) | number | 25th-percentile time to fill — Pro |
| Time to Fill p75 (days) | number | 75th-percentile time to fill — Pro |
| Fall-Off % (Median) | number | Median 90-day fall-off rate |
| Sample Size | integer | Number of placements underlying the cell |
| Source | string | Data source tags (e.g. "IQN + NAPS + Split-Fee") |
Ready to benchmark your numbers?
Free access to medians. Pro unlocks the full distribution.