Your no-show rate is probably near 5%. You are being sold against as if it were 45%.
Temporary shift workers miss roughly one booked shift in twenty. The 50% to 60% show rate used to sell against staffing agencies implies they miss nine in twenty. One of those numbers comes from two national statistics agencies. The other comes from a footnote that says partner feedback.
Temporary shift workers miss roughly one booked shift in twenty. Salaried employees miss roughly one working day in thirty. That is the entire gap. US and UK government data puts blue-collar shift work at 3.4% absence, part-time workers run about 1.5 times the full-time rate, and the two together put temp shift work near 5%, in a range of 3% to 8%.
The figure used when selling against agencies, a 50% to 60% show rate, means nine in every twenty booked workers never arriving. It is about nine times the rate either government measures for comparable work, and the platforms that publish it footnote it as partner feedback rather than measurement.
What this costs you. Two things. If your rate is above about 8%, you are losing gross profit and coordinator hours you could recover, and the calculator below prices it. If your rate is already under 8%, you are losing something harder to see: buyers who believe the 50% figure discount your reliability before you quote. On both counts the fix is the same, which is to measure your own number and be able to show it.
- Scope
- US and UK, temporary and shift labour
- Primary sources
- 11, of which 4 government
- Derived estimate
- 3% to 8%, centre near 5%
- Next update
- March 2027
Key findings
- Two governments put blue-collar shift work within a tenth of a point of each other. US production, transportation and material moving occupations sit at 3.4%; UK process, plant and machine operatives at 3.3%, the highest UK occupational group. The two are computed differently, so read this as corroboration rather than confirmation.12
Medium evidence · high agreement · medium confidence - Across all 65 occupation and industry categories the US government measures, the absence rate runs from 2.2% to 5.0%. Not one category exceeds 5.0%. The figure implied by gig-platform marketing is 40% to 50%.1
Robust evidence · high agreement · high confidence - The US industry sector that contains temporary help services shows a 2.8% absence rate, below the 3.2% all-worker average. Administrative and support and waste services is where the Current Population Survey places agency temps.1
Robust evidence · medium agreement · medium confidence - Part-time workers are absent about 1.5 times as often as full-time workers. UK data puts part-time at 2.8% against full-time 1.9%, which matters because both government datasets otherwise under-represent shift labour.2
Medium evidence · medium agreement · medium confidence - Our derived band for temp shift-work no-shows is 3% to 8%, with a central estimate near 5%. This is an inference from adjacent measurement, not an observation. Treat it as a starting hypothesis to test against your own data.
Limited evidence · medium agreement · low confidence - Do not benchmark yourself against a published attendance figure. Use a stated definition instead. A no-show rate is a division, and figures in circulation do not say whether a worker who cancels the night before belongs on top, or whether the bottom is shifts booked, orders received or workers on the roster. That is why this report publishes its own: Table 5 gives the definition and denominator, Table 3 gives every step behind the band above. Both can be checked, and both can be rejected.8
Robust evidence · high agreement · high confidence
Why you cannot simply look this up
Gig platforms sell against staffing agencies with a number: agency workers turn up 50% to 60% of the time. If that were true, four in every ten booked shifts would go unstaffed and no agency would still be trading. Most founders sense the figure is wrong. Almost none of them have anything to check it against.
That is the real problem, and it is not a shortage of data. Healthcare settled its equivalent question decades ago. A review of 105 studies puts medical appointment no-shows at 23%, and any clinic administrator can compare their own rate against that distribution.3 A second study prices one missed appointment at $196.4
Staffing has pieces of the same thing. Platforms publish show rates for their own workers, agencies publish fill-rate targets, and buyer-side guides list the metrics a client should ask for. What is missing is the connective tissue: a shared definition of what counts, a stated denominator, and a distribution rather than a single headline. Whether a late cancellation counts, whether an unfilled order counts, and what sits on the bottom of the fraction are all answered differently by whoever is publishing.
So this report does three things. It gathers every figure that can actually be measured. It grades how far each one can be trusted. And it derives a working band you can hold your own number against, with the arithmetic shown so you can reject it.
What the perception costs, even when your number is fine
The gap between 5% and 45% is not an academic dispute about statistics. It is a sales argument, and it is running in rooms where agencies are being compared with platforms. A buyer who believes that agency workers turn up half the time will discount an agency's reliability before a rate is even quoted, ask for a lower rate to compensate for risk that is not there, or move volume to a platform on the strength of a figure that no measurement supports.
Exhibit 1 is the same claim drawn twice, out of a hundred booked shifts.
Out of 100 booked shifts, how many go unstaffed
Left: the rate this report derives for temporary shift work. Right: the rate implied by a 50% to 60% show rate. Each square is one booked shift.
5 in 100
A 5% no-show rate. One shift in twenty needs covering, and most of those get covered.
45 in 100
A 55% show rate. Nearly half of every roster failing to arrive, every day, indefinitely.
Source: left panel, this report's derived central estimate (Table 3, low confidence); right panel, the midpoint of the 50% to 60% show rate in Instawork product marketing.5
Drawn this way the second panel answers itself. An agency losing nearly half of every roster would not keep a client for a quarter, and no staffing firm operates that way for long. The useful response is not to argue about the figure, which is unfalsifiable, but to be able to produce your own with a stated definition behind it.
What gets quoted, and what each figure measures
Four figures do most of the work in conversations about staffing attendance. They come from different places: two from US government statistics, one from medical research, one from platform marketing. Exhibit 2 puts all four on a single axis, which is the quickest way to see that they are not alternative estimates of the same quantity. They are measurements of four different populations.
Four figures, four different populations, one axis
None of the four measures a temporary worker failing to attend a booked shift. The spread between them is a difference of subject, not a disagreement about staffing.
Source: BLS Table 47, 2025 annual averages;1 Dantas et al., Health Policy, 2018;3 Instawork product marketing, accessed 10 September 2026.5 Axis 0 to 50%.
The closest available measurement: US absence data
Of those four, the US absence data is the strongest: a government statistical series, published annually, broken down finely enough to isolate the work temporary staffing actually supplies. Exhibits 3 to 5 take it apart, because if a defensible starting point exists anywhere it is here. The Current Population Survey has measured workplace absence for decades. Table 47 reports it by occupation and industry, splitting illness or injury from other personal reasons, and publishing both an absence rate and a lost worktime rate.1
US absence rate by industry, and where temporary help sits
Share of full-time wage and salary workers with an absence during the reference week, 2025 annual averages. The Current Population Survey places agency temps in administrative and support and waste services.
Source: BLS Current Population Survey, Table 47, 2025 annual averages.1 Axis 0 to 6%. Absences exclude vacation, personal days and holidays.
Occupation tells the same story with a different cut. The group that light-industrial staffing overwhelmingly supplies, production, transportation and material moving, sits at 3.4%. The highest occupational rate anywhere in the table is personal care and service at 4.3%.
US absence rate by occupation group, 2025
Highlighted row is the occupational group that light-industrial and warehouse staffing supplies.
Source: BLS Current Population Survey, Table 47, 2025 annual averages.1 Axis 0 to 6%.
Exhibit 5 collapses all of that detail into a single band and sets it beside the other two figures from Exhibit 2, on the same axis.
The full measured range, beside the figures it is often compared with
The measured band spans all 65 occupation and industry categories in Table 47. The third band is the absence rate implied by a 50 to 60% show rate, shown for scale rather than as a competing estimate.
Measured, 65 US categories
Medical appointments, 105-study mean
Asserted for staffing agencies
Source: BLS Table 47, 2025;1 Dantas et al., 2018;3 Instawork product marketing.5 Single linear axis, 0 to 50%.
An independent cross-check: UK sickness absence
The Office for National Statistics measures a different thing in a different way, which makes it a genuine independent check rather than a second opinion on the same data. ONS reports the share of working hours lost to sickness or injury. In 2025 that was 2.0%, or 148.8 million working days, an average of 4.4 days per worker.2
UK sickness absence by worker characteristic, and the part-time premium
Percentage of working hours lost to sickness or injury, 2025. The highlighted pair is the reason both national datasets under-state shift labour.
Source: ONS, Sickness absence in the UK labour market: 2025, released 1 May 2026.2 Axis 0 to 4.5%. Official statistics in development.
| Measure | Definition | US (BLS) | UK (ONS) |
|---|---|---|---|
| Worker-based rate | Share of workers absent in the reference period | 3.2% | not published |
| Hours-based rate | Hours lost as a share of hours usually worked | 1.7% | 2.0% |
| Blue-collar shift occupations | Production and machine operative groups | 3.4% | 3.3% |
| Public sector | Government employment | 4.2% | 2.9% |
| Private sector | Non-government employment | 3.0% | 1.7% |
Note: only the hours-based row is a like-for-like comparison. The 3.4% and 3.3% figures in row three are close but are computed differently, one worker-based and one hours-based. Their proximity is suggestive, not confirmatory. Sources: BLS Table 47, 2025;1 ONS, 2025.2
How platform figures sit alongside measured data
Gig-staffing platforms publish attendance figures more prominently than anyone else, so theirs are the numbers agency owners hear most often. They are worth reading closely, footnotes included, because the footnotes are more careful than the headlines.
| Claim | Metric | Disclosed basis | Grade |
|---|---|---|---|
| 98% worker show rate | Attendance, own platform | Own operational data, method not published | Self-reported |
| 50 to 60% agency show rate | Attendance, competitors | “Based on Instawork partner feedback” | Anecdotal |
| 90% of jobs filled in 24 hours | Fill rate | US markets, all positions, shifts posted 2 to 7 days ahead | Self-reported, scoped |
| 7 to 10 day agency time-to-fill | Speed, competitors | Same partner-feedback footnote | Anecdotal |
| 94% nationwide fill rate | Fill rate, second platform | Self-reported, no method published | Self-reported |
Sources: Instawork product marketing, accessed 10 September 2026;5 Bacon platform marketing, accessed 10 September 2026.6
We have no reason to doubt the platforms about their own operations. A marketplace with ratings, standby cover and workers choosing their own shifts may well achieve high attendance. The comparator is the part worth pausing on. A 50 to 60% show rate implies four or five workers in every ten booked never arriving, which sits well above what either government measures for similar work. The platforms themselves footnote it as partner feedback rather than measurement, and on the evidence available that footnote is doing the right thing.
A 50 to 60% show rate would put temp staffing at twelve to fifteen times the absence rate two governments measure for the same kind of work.
One further point of comparison belongs on the record. Bullhorn's 2026 survey of nearly 2,300 recruitment professionals found that 56% of the highest-growth staffing firms, meaning those growing revenue by more than 25%, report average placement times under ten days.7 An industry average and the operators a buyer is actually choosing between are not the same reference point.
A derived band for temp shift work
A benchmark report that only says the number is unknown is not much use. So here is an explicit derivation, with each step exposed so you can reject any of them.
| Step | Input | Value | Source or basis |
|---|---|---|---|
| 1 | Anchor: US absence, production, transportation and material moving occupations | 3.4% | BLS Table 47, 20251 |
| 2 | Cross-check: UK machine operatives | 3.3% | ONS, 20252 |
| 3 | Part-time premium, applied because both anchors are full-time | ×1.47 | ONS part-time 2.8% ÷ full-time 1.9%2 |
| 4 | Adjusted central estimate | 5.0% | 3.4% × 1.47 |
| 5 | Unmeasured upward pressure: short tenure, no accrued leave, multi-apping, travel distance | not quantified | Directional only |
| 6 | Reported band | 3% to 8% | ShiftMargin derivation |
This band is an inference from adjacent populations, not an observation of temp shift attendance. Evidence: limited. Agreement: medium. Confidence: low. It is offered as a hypothesis to test against your own data, and we expect operators with real distributions to contradict it.
Two honest observations about that band. It brackets the under 4% to 6% operating target we run against in live deployments, which is either mild corroboration or evidence that we anchored on our own priors. And it excludes recoverability entirely: a no-show you cover before shift start has different economics from one you do not, and no public dataset distinguishes them.
What does a 5% no-show rate actually cost you?
Put your weekly shifts and your own rate into the calculator. It prices the loss on the gross profit spread rather than the bill rate, and treats the share you cover before shift start as its own input.
What improvement is worth, on your numbers
Two levers move the cost of no-shows, and they work differently. Confirmations timed to when workers actually read reduce how many no-shows happen. A pre-warmed backup pool does not reduce them at all, it raises how many get covered before shift start. The first protects margin, the second protects the client relationship.
Put your own volumes in and the model below applies both. The 30% relative reduction it uses for confirmations comes from meta-analysis of appointment reminders, not from staffing, so treat it as directional and test it against your own before-and-after.8
| Scenario | No-show rate | Cover rate | Annual gross profit at stake |
|---|---|---|---|
| TodayYour numbers as entered | 8.0% | 65% | $64,038 |
| Timed confirmationsMessage and call timed to risk, with a real reply path | 5.6% | 65% | $44,827 |
| Confirmations and a warmed backup poolStandby workers matched and asked the evening before | 5.6% | 85% | $38,214 |
Costed on the gross profit spread at a $24 bill rate and a 25.9% margin, 8-hour shifts, coordinator recovery at 25 minutes an incident. Assumptions editable in the calculator.
On the default inputs, doing both is worth about $25,825 a year in recovered gross profit, and 74% of that comes from the confirmations rather than the backup pool.
Evidence grading, every figure in this report
The grading scheme follows the approach used in IPCC assessment reports, where a finding carries a stated evidence level and a stated degree of agreement, which together produce a confidence level.9 We use it here because the brief for this report was to be explicit about what is verified and what is directional.
| Figure | What it measures | Population | Confidence |
|---|---|---|---|
| 3.2% | US absence rate | Full-time wage and salary workers, US, 2025 | High |
| 2.2–5.0% | Full range across 65 US categories | Full-time wage and salary workers, US, 2025 | High |
| 2.8% | Absence, sector containing temp help | Administrative and waste services, full-time, US | Medium |
| 3.4% | Absence, blue-collar shift occupations | Production and material moving, full-time, US | High |
| 2.0% | UK sickness absence, hours lost | All UK workers, 2025 | Medium |
| 3.3% | UK sickness absence, machine operatives | Process, plant and machine operatives, UK | Medium |
| ×1.47 | Part-time to full-time absence ratio | UK workers, 2025 | Medium |
| 23.0% | Medical appointment no-show rate | Patients, 105 studies, 1980 to 2016 | High for patients |
| $196 | Cost of one missed appointment | Ten VA clinics, 2008 dollars | Medium |
| 98% | Platform show rate | One platform's own workers | Self-reported |
| 50–60% | Claimed agency show rate | Unspecified agencies, partner feedback | Very low |
| 3–8% | Derived temp shift-work band | Inference, not observation | Low |
Confidence combines evidence strength and agreement between independent sources. “High for patients” marks a well-evidenced figure describing a population other than the one this report is about, which is the most common way this number is misused.
Proposed standard definition
The benchmark cannot exist until the industry agrees what counts. Healthcare resolved this by building non-recoverability into the definition: a no-show is a patient who did not attend, or who cancelled too late for the slot to be refilled.4 Below is the definition we use and propose.
| Event | Counts as | Denominator | Why separate |
|---|---|---|---|
| Worker assigned, did not attend, no notice before shift start | No-show | Shifts assigned | The attendance failure itself |
| Worker cancelled with time to cover | Scheduling event | Shifts assigned | Costs coordinator time, not margin |
| Order received, never assigned a worker | Fill-rate failure | Shifts requested | Different denominator entirely |
| No-show covered before shift start | Recovered no-show | No-shows | Report cover rate alongside no-show rate |
| Worker attended late | Lateness | Shifts attended | Client impact without a fill failure |
Minimum reporting set: no-show rate, cover rate, and both segmented by client, site, shift start hour and booking lead time. An aggregate figure without segmentation is a reporting number, not an operating one.
Methodology
How this report was built
Inclusion. A figure was included only if we could reach a primary document stating it: a government statistical table, a peer-reviewed paper, or a vendor's own published page. Secondary aggregators were used to locate primaries and then discarded.
Verification. Each figure was checked for its exact value, its year, and the population it describes. Three figures were removed during drafting because the primary source did not support the claim: a widely repeated sample size for the SIA and Bullhorn Staffing Indicator that Bullhorn does not in fact disclose, a time-to-fill sub-figure absent from Bullhorn's own release, and a weekly headcount figure we could not trace.
Government data. BLS Table 47 was read from the published PDF rather than a summary. ONS figures were read from the 2025 release. Where the two agencies measure differently, Table 1 states the mismatch rather than blending them.
Grading. Evidence level and agreement were assigned per figure following the IPCC uncertainty framework,9 then combined into a confidence level. Self-reported vendor metrics are graded separately from measured data regardless of how plausible they look.
Limitations
Where this report is weak
- No dataset measures the thing this report is about. Every anchor is an adjacent population. Absence among full-time employees is not no-shows among booked temp workers, and the report should not be read as if it were.
- BLS counts notified sick leave as absence and excludes vacation. A no-show is defined by the absence of notice. The two concepts overlap without matching.
- Both government series carry their own caveats. BLS 2025 annual figures are eleven-month averages excluding October, because collection stopped during the federal shutdown, and a 2022 industry reclassification breaks comparability with earlier years.1 ONS labels its sickness absence series official statistics in development and advises caution.2
- Part-time is a proxy for shift work, not a substitute. The ×1.47 premium is the best adjustment available and remains an assumption.
- The derived band is not an observation. Low confidence, stated as such, and the first real distribution published by an operator or an ATS vendor should replace it.
- Geographic scope is US and UK. Nothing here speaks to EU, GCC, African or Asian markets, where both labour patterns and measurement differ.
Recommendations
For staffing operators
- Adopt the Table 5 taxonomy, or any fixed one, and publish your no-show rate and cover rate as two numbers.
- Pull ninety days of assignments and timesheets and segment by client, site, shift start hour and booking lead time before drawing any conclusion.
- Price it on gross profit rather than bill rate. An unfilled shift costs the margin spread, roughly a quarter of the revenue figure, and only uncovered shifts cost the full spread.10
- Treat any vendor comparator without a disclosed method as sales material, including ours.
For ATS vendors and trade bodies
- The underlying data already sits in timesheet and assignment records. A banded, unattributed distribution of no-show rates would give the industry a shared reference point, and would extend the benchmarking already published rather than replace it.
- State the denominator alongside it. Fill-rate figures are hard to compare today because the base is sometimes orders, sometimes shifts and sometimes placements.
Test your own number, and what it is costing
On 400 shifts a week at 8% with 65% cover, the calculator returns about $64,000 a year, of which $12,000 is coordinator recovery time paid whether or not the shift is covered. At 30% cover it moves to $80,600. Every assumption is editable and the arithmetic is printed on the page.
Corrections made before publication
Three figures were removed during drafting because the primary source did not support the claim. A sample size widely attributed to the SIA and Bullhorn Staffing Indicator turns out not to be disclosed by Bullhorn, which describes its panel only as a sizable sample. A time-to-fill sub-figure repeated across several summaries does not appear in Bullhorn's own release. A weekly headcount figure could not be traced to a primary at all.
One of the three had been carrying an argument in an earlier draft. They are recorded here because a benchmark is only worth the verification behind it, and because those same figures are still in circulation elsewhere.
Until a shared distribution exists, the most useful benchmark is the one you build from your own roster.Where to start
Want your own distribution instead of an industry estimate?
The Ops Audit runs on your call logs, ATS events and timesheet trails and produces your real no-show rate by client, site and shift start hour, your cover rate, and what recovery costs. $7,500 fixed, and free if it does not identify ten times its fee in annual savings. Anonymised results from these audits are what will eventually replace the derived band in this report.
Request a free 30-minute teardown →Sources
- US Bureau of Labor Statistics, Current Population Survey, Table 47: Absences from work of employed full-time wage and salary workers by occupation and industry, 2025 annual averages. Total absence rate 3.2%, lost worktime rate 1.7%. Range across all occupation and industry categories 2.2% (utilities) to 5.0% (federal government). Production, transportation and material moving occupations 3.4%; administrative and waste services 2.8%; transportation and warehousing 3.1%; retail trade 3.4%; accommodation and food services 3.4%; health care and social assistance 3.8%; construction 2.3%; manufacturing 2.9%; personal care and service 4.3%; management 2.5%. Absences cover people who usually work 35 or more hours per week and worked fewer than 35 in the reference week for illness, injury, child care, family or personal obligations, civic or military duty, or parental leave; vacation, personal days and holidays are excluded. 2025 estimates are eleven-month averages excluding October, when collection stopped during the federal shutdown, and are not strictly comparable with earlier years following the 2022 NAICS reclassification.
- UK Office for National Statistics, Sickness absence in the UK labour market: 2025, released 1 May 2026. Sickness absence rate 2.0%, defined as hours lost to sickness or injury as a share of total hours; 148.8 million working days lost; 4.4 days per worker. Process, plant and machine operatives 3.3%, the highest occupational group; managers, directors and senior officials 1.0%. Part-time 2.8% against full-time 1.9%. Men 1.7%, women 2.4%. Aged 16 to 24 1.3%, aged 65 and over 3.3%. With a long-term health condition 4.0%, without 1.0%. Public sector 2.9%, private sector 1.7%. Minor illnesses account for 30.4% of absence occurrences. Labelled official statistics in development, with caution advised.
- Dantas, Fleck, Cyrino Oliveira and Hamacher, No-shows in appointment scheduling: a systematic literature review, Health Policy 122(4):412–421, 2018. 105 studies published 1980 to 2016; mean medical appointment no-show rate 23.0%; highest in Africa at 43.0%, second highest South America at 27.8%, lowest Oceania at 13.2%; long lead time and prior no-show history the strongest predictors.
- Kheirkhah, Feng, Travis, Tavakoli-Tabasi and Sharafkhaneh, Prevalence, predictors and economic consequences of no-shows, BMC Health Services Research, 2016. Michael E. DeBakey VA Medical Center, ten clinics, FY1997 to FY2008; mean no-show rate 18.8%; average cost of one no-show $196 in 2008 dollars; ten clinics absorbed $14.58m in one fiscal year; a no-show defined to include cancellations too late to refill the slot.
- Instawork, product marketing homepage, accessed 10 September 2026. Claims a 98% worker show rate against traditional staffing agencies at 50 to 60%, time to fill under 24 hours against 7 to 10 days, 90% of jobs filled within 24 hours, 4.8 average worker rating, and 7 million or more workers. Footnote: “Staffing agencies stats above are based on Instawork partner feedback.” Second footnote: fill rate is “based on shifts filled in US markets across all positions and when shifts are posted 2-7 days in advance”.
- Bacon, on-demand staffing platform marketing, accessed 10 September 2026. Self-reported nationwide fill rate of 94%, individual case studies at 93% and 89%, and a backup-worker programme sending standby workers to every shift. No measurement method published.
- Bullhorn, 2026 GRID Industry Trends Report, published 25 February 2026, surveying nearly 2,300 recruitment professionals. “56 percent of the highest-growth firms report average placement times under 10 days”, where highest-growth means firms growing revenue by more than 25%. The report measures placement speed and growth and does not report attendance.
- Bullhorn and Staffing Industry Analysts, SIA | Bullhorn Staffing Industry Indicator. Tracks weekly hours worked by temporary workers at staffing companies using Bullhorn, weighted and benchmarked against US Bureau of Labor Statistics data. Bullhorn discloses no worker or firm count, describing the panel only as “a sizable sample of the staffing industry, reflecting a wide range of occupations, client industry verticals, and geographic footprint”. Reports hours worked and contains no attendance, no-show or absence measure.
- Mastrandrea et al., Guidance Note for Lead Authors of the IPCC Fifth Assessment Report on Consistent Treatment of Uncertainties, IPCC, 2010. Evidence summarised as limited, medium or robust; agreement as low, medium or high; confidence expressed on a five-point scale from very low to very high. Used here as the grading scheme for Table 4.
- Level CFO, Staffing industry gross margin and markup benchmarks. Light-industrial gross margin 25.9%, from TrueBlue's FY2024 10-K; employment burden 10 to 14% of payroll. Basis for the margin arithmetic in the calculator and in recommendation three.