What the data shows

1
The strongest link measured: inside the same kind of work, roles open worldwide drew 23.0 times as many apply clicks per opening as roles limited to listed countries (across 4 kinds of work).
Figure 1
2
Roles restricted to US residents came out weakest: inside the same kind of work, they drew 98% fewer apply clicks for each opening than roles open worldwide.
Figure 1
3
With all kinds of work lumped together, roles whose title names a car, truck, drone or vessel drew 78% fewer apply clicks for each opening than roles whose title names none; inside each kind, they drew 66% more clicks for each open role. The pooled number reflects where they cluster: in kinds of work that draw fewer applicants per role.
Figure 1
4
Measured against their kinds' average clicks per open role, the best-paid fifth of roles in each kind of work came in at 3.03 times and the lowest-paid fifth at 3.35 times, with roles that give no rate at 0.61 times.
Figure 2
5
A role in its first week drew 5.70 times its kind of work's average clicks per open role, against 0.34 times for roles posted over three months earlier. Part of that gap comes from the site ranking recent roles first.
Figure 3
6
In 6 of 7 kinds of work compared, roles with a posted hourly rate drew more apply clicks per open role than roles without one. Among kinds with enough clicks and roles on both sides to stand on their own, the ratio ranged from 15.44x (Robot Teleoperation) down to 2.71x (AV & Drone Operators).
Figure 5
7
Against the average for their kind of work, worldwide roles drew 12.81 times the clicks per open role, and US-only roles 0.52 times.
Figure 4

Figure 1

Traits that draw applicants, measured inside each kind of work

Compared with similar roles that lack them, which traits bring a robot training role more apply clicks per open role?

Trait	Within kind (with / without)	All kinds pooled	Roles with the trait	Kinds of work compared	Compared with
Open worldwide	22.98x	130.71x	10	4	Roles limited to listed countries
Listing under 14 days old	7.23x	6.39x	175	7	Older listings
Title of 9 words or moreSmall sample	3.71x	1.59x	15	5	Roles with 4 words or fewer
States an hourly rate	3.24x	4.28x	429	7	Roles with no rate posted
Title names a vehicle, drone or vessel	1.66x	0.22x	632	5	Roles whose title names none
Title says part-timeSmall sample	1.20x	0.45x	32	3	Roles without it
Title names a night, swing or weekend shift	0.82x	0.49x	106	5	Roles whose title names none
Pay in its kind's top third	0.82x	0.67x	121	7	Roles in the bottom third
Open only to US residents	0.02x	0.01x	810	4	Roles open worldwide

Base: 1,348 roles that were open at some point from 1 Oct 2026 to 31 Oct 2026 drew 470 apply clicks across 4,118 open role-weeks.

A row is flagged, and left out of the rankings in the findings, unless both of its sides can stand on their own. Each side's weekly clicks per open role is pulled toward the two sides' combined rate by a weight fitted on their own clicks, and both sides need at least that many open role-weeks; when the two rates are no further apart than chance would leave them, no amount of open time is enough. The index figures below flag their rows against one another by the same rule.

Within kind: only kinds of work where both sides had open time take part. Add up the clicks the roles with the trait actually drew and divide by the clicks they would have drawn at their own kinds' rates (each kind's rate counted over both sides); do the same for the roles without the trait, then divide the first result by the second. 1.00x means no difference, and a trait is only rated when two kinds of work qualify. Pooled: the same ratio with all kinds of work lumped together. When the two numbers disagree, the trait is concentrated in kinds of work that draw more (or fewer) applicants per role.

Pay thirds are built inside each kind of work from the roles in the window that give an hourly rate, with tied rates sharing a third. Listing age is measured at the midpoint of the days a role was open in the window, and roles without a known listing date are dropped from that comparison. Vehicles are matched in titles using the robots and equipment report's vocabulary; shifts and part-time use the job-titles report's.

Wording in a title often comes from an employer's posting template: shift and part-time terms, for instance, turn up in the titles of a handful of employers. So inside a kind of work, a title trait partly compares the employers that use that wording against those that do not.

Every association here partly reflects how this site sorts roles. By default, new listings, roles with stated pay and roles at partner networks sit higher, and visitors see the roles open in their own country first. Read each ratio as a description of this site's users, not as a cause.

Not counted: 134 clicks in the window on employers' general sign-up links, which point to no specific role.

Credit: RobotTraining.jobs Research, What Makes Robot Training Job Ads Attract Applicants, October 2026 edition.

Figure 2

How pay level relates to applicants, inside each kind of work

Within one kind of work, do higher-paying roles attract more applicants than the rest?

Pay band in its kind of work	Within-kind index	Weekly clicks per open role	Median rate, per hour	Roles	Open role-weeks	Clicks	Clicks at its kinds' rates
Bottom fifth of rates in its kind	3.35x	0.52	$22	91	202	106	31.6
Second fifth	1.45x	0.19	$26	84	219	41	28.2
Middle fifth	1.56x	0.31	$30	63	148	46	29.4
Fourth fifth	0.53x	0.06	$27	111	233	14	26.5
Highest fifth	3.03x	0.38	$44	78	160	60	19.8
No hourly rate stated	0.61x	0.06	n/a	907	3,114	202	333.4

Base: 1,334 roles listed from 1 Oct 2026 to 31 Oct 2026, drawing 469 apply clicks, in the 6 kinds of work where 5 or more roles give an hourly rate (enough to put one in every fifth).

Within-kind index: a group's actual clicks divided by the clicks it would have earned at the average rate of its own kinds of work, so 1.00x equals that average. Summed over this figure's rows, observed and expected clicks come to the same total.

Each kind of work is split into fifths by the advertised rates of its roles in the window, counting each role once and putting tied rates in the same fifth. The median rate column pools all kinds and is there for scale.

A row with under 1.4 expected clicks carries a small-sample mark and is left unranked: under that count, the average across all rows says more than the row's own expected clicks do (a weight fitted on the counts themselves).

Credit: RobotTraining.jobs Research, What Makes Robot Training Job Ads Attract Applicants, October 2026 edition.

Figure 3

Fresh versus older listings, inside each kind of work

Within one kind of work, how many more applicants do fresh listings draw than older ones?

Listing age	Within-kind index	Weekly clicks per open role	Roles	Open role-weeks	Clicks	Clicks at its kinds' rates
First week	5.70x	0.64	120	121	78	13.7
Second week	1.95x	0.11	55	135	15	7.7
Weeks 3 and 4	2.24x	0.33	82	238	79	35.2
1 to 3 months	0.32x	0.05	312	1,043	50	154.4
More than 3 months	0.34x	0.03	534	2,102	63	184.1
Listing date unknown	2.47x	0.39	245	478	185	74.9

Base: 1,348 roles that were open at some point from 1 Oct 2026 to 31 Oct 2026 drew 470 apply clicks across 4,118 open role-weeks.

Age is measured at the midpoint of a role's open days in the window, counting from its listing date: the posting date on the job board, or its first day in the feed if the board gives none.

Within-kind index: actual clicks divided by the clicks expected at the average rate for the role's own kind of work. Across the rows, observed and expected clicks sum to the same total.

This site's default sort puts recent listings first, so this figure measures the site's own ordering as much as what applicants prefer. A newer role is also more likely to be one that people have not yet applied to.

None of the rows with expected clicks carries a small-sample mark: they vary so widely that each one stands on its own from the first.

Credit: RobotTraining.jobs Research, What Makes Robot Training Job Ads Attract Applicants, October 2026 edition.

Figure 4

Location rules and applicants, inside each kind of work

Within one kind of work, do worldwide roles pull more applicants than roles limited to certain countries?

Location rule	Within-kind index	Weekly clicks per open role	Roles	Open role-weeks	Clicks	Clicks at its kinds' rates
Open worldwide	12.81x	9.28	10	19	179	14.0
Only the United States	0.52x	0.05	810	2,254	121	233.6
Only other listed countries	0.76x	0.09	527	1,845	170	222.4
No rule statedSmall sample	0.00x	0.00	1	0	0	0.0

Base: 1,348 roles that were open at some point from 1 Oct 2026 to 31 Oct 2026 drew 470 apply clicks across 4,118 open role-weeks.

'Only the United States' means the rule lists the United States and nothing else. 'Only other listed countries' covers every other include-list, regardless of which countries it names.

Within-kind index: observed clicks as a multiple of what the role's kind of work would draw at its average rate.

Visitors see the roles open in their own country first. A role limited to certain countries therefore gets top placement only for visitors from those countries.

None of the rows with expected clicks carries a small-sample mark: they vary so widely that each one stands on its own from the first.

Credit: RobotTraining.jobs Research, What Makes Robot Training Job Ads Attract Applicants, October 2026 edition.

Figure 5

Posted pay and applicants in each kind of work

Which kinds of work see more applicants on roles that post an hourly rate than on roles that leave it out?

Kind of work	With / without	Weekly clicks per role, rate given	No rate stated	Roles with / without	Clicks with / without
Robot Teleoperation	15.44x	0.25	0.02	32 / 69	17 / 4
Embodied-AI Data Annotation	4.62x	2.42	0.52	15 / 14	61 / 12
Data & Demonstration Collection	3.45x	0.79	0.23	63 / 79	101 / 62
Egocentric & Wearable Capture	3.44x	0.61	0.18	36 / 115	48 / 82
Robot Fleet Operations	2.92x	0.08	0.03	132 / 174	22 / 13
AV & Drone Operators	2.71x	0.05	0.02	149 / 456	18 / 29
Robot Task Evaluation & QASmall sample	0.00x	0.00	0.03	2 / 12	0 / 1

Base: Roles listed from 1 Oct 2026 to 31 Oct 2026. Shown: the 7 of 7 kinds of work where roles with and without an hourly rate were both open during the window and those without one drew at least one click.

A row is flagged when its two sides cannot stand on their own, since a few popular roles can swing it: each side needs as many open role-weeks as the weight of the two sides' combined rate, fitted on their own clicks.

Posting a rate or not is mostly decided by each employer's listing template, so inside a kind of work this comparison is partly employer against employer.

Credit: RobotTraining.jobs Research, What Makes Robot Training Job Ads Attract Applicants, October 2026 edition.

Robotics labs · Fleet operators · Data vendors · Research teams

Buy the data behind this report

This report sums up records we gather ourselves, day after day:

1,432 robot training roles followed, history included
Every role on 100 employer job boards, read each day since 12 Sep 2026: its pay, its place, and the first and last day it was up.
536 apply clicks over 28 days
Which robot training roles people really apply to, by kind of work, pay and place. The clicks happen here, so no one else can count them.
25 sourced industry datasets
Funding, revenue, deals, fleets, deployments and prices across the robot training industry, each record linked to the publication it comes from.

On offer

The raw records
Every robot training role we track: title, employer, place, kind of work, pay as stated, and the days it was first and last open.
A regular feed
New and closed roles every week or month, as file drops or through the data endpoints.
A custom cut
Any figure here by country, city, kind of work, vehicle, employer or date, or your own hiring benchmarked against the market.

Need the people behind this data for your own robots? (you contract them yourself) or (we deliver the robot data).

Investing in robotics, AV or robot-data companies? The Robot hiring radar is these records role by role, every day, as files.

Robot hiring brief: free, once a month

The robot-training job market in figures, on the first weekday of each month: open roles, what opened and closed, who added and who cut, and posted pay by kind of work. To your work email.

How it was measured

Apply clicks on RobotTraining.jobs, linked to the clicked role and its open days, plus each role's pay, location rule, listing date and title as read daily from the public job boards. A single database query produces each figure for the window shown, and the live page and every monthly edition run that identical query. The note under each figure gives its base.

Apply click: A press of an apply button on RobotTraining.jobs that opens an employer's application page for a single role. Bot traffic is filtered out when logged. Clicks on an employer's general sign-up link belong to no role, so they are excluded.
Open role-weeks: Each role's listed days within the window, divided by 7, then added across roles. A role counts only if it was up at some point during the window.
Weekly clicks per open role: Apply clicks over open role-weeks.
Within-kind ratio: In every kind of work with open time on both sides of the trait (and at least one click without it), divide the weekly click rate with the trait by the rate without. Each side is first pulled toward its kind's pooled rate by a weight fitted for that trait across every kind's two sides, so a kind with little open time lands close to no difference. Average across those kinds, weighting by open role-weeks with the trait. Two qualifying kinds are required. 1.00x means no difference.
Pooled ratio: That ratio again, with all kinds of work lumped together. A gap between it and the within-kind ratio means the trait clusters in kinds of work that attract more, or fewer, applicants per role.
Within-kind index: Actual apply clicks for a group of roles, divided by what those roles would have drawn at their own kinds' average click rate. 1.00x equals that average.
Pay thirds and fifths: Cut separately for each kind of work from the hourly rates its roles advertised during the window, one entry per role, with tied rates kept in the same band. Fifths require 5 or more priced roles in the kind, enough to put one rate in every fifth.
Listing age: Days between a role's listing date (the board's posting date, or failing that its first day in the feed) and the midpoint of its open days in the window. Anything under 14 days old counts as new.
Location rule: Taken from the location text of each listing or, when that text names no country, from what the listing's own description says: worldwide, worldwide with named exceptions, named countries only (US-only when that one country is the United States), or no rule given.
Vehicle, shift, part-time: Must appear in the title as whole words, checked against the word lists used by the robots and equipment report and the job-titles report.
Kind of work: A role's title and description place it in one of the seven kinds of robot training work this site follows, ranging from robot teleoperation and demonstration collection through egocentric capture, embodied-AI annotation, robot evaluation and fleet operations to AV and drone operation. Each kind has a page of its own.
Window: Live page: the 28 most recent full days (UTC). Edition: a single calendar month (UTC).

Limits of this data

Some of every association comes from how this site orders roles. The default feed lifts fresh listings, roles that state pay and roles at partner programs, and shows each visitor the roles open in their country first, so the listing-age, pay and location results reflect that ordering as well as what applicants prefer.
Each ratio describes how RobotTraining.jobs visitors behave and does not prove cause. Traits come bundled (a fresh listing is often a night shift too), and only the kind of work is held constant.
Title wording often comes from an employer's house template, so inside a kind of work a title trait partly compares employers that use that wording with those that do not.
With only a few clicks per role, a few popular listings can swing a ratio between windows. Check a trait over several editions before leaning on it.
A click opens an application but says nothing about submission, acceptance or hiring, and a single visitor may click many roles.
Hiring that happens without a public job board is not captured.

Collection and classification of listings are explained on the methodology page. Freshness, completeness and consistency are tested daily in the data quality report.

Monthly editions

The live page spans the 28 most recent full days and updates daily. Every finished month also gets its own edition, locked two days after month end, so a number quoted from it stays fixed.

Live page28 most recent full days, rebuilt daily
October 2026 edition (you are here)Will lock when the next daily refresh runs

Quoting this report

You may quote, chart or republish these figures under CC BY 4.0, provided you link back to this page. This edition is locked: its numbers will not change.

Citation

RobotTraining.jobs (2026). What Makes Robot Training Job Ads Attract Applicants, October 2026 edition. https://robottraining.jobs/research/what-draws-applicants-to-robot-training-roles/2026-10

BibTeX

@misc{robottrainingjobs_what_draws_applicants_to_robot_training_roles_2026_10,
  author = {{RobotTraining.jobs}},
  title = {What Makes Robot Training Job Ads Attract Applicants, October 2026 edition},
  year = {2026},
  howpublished = {\url{https://robottraining.jobs/research/what-draws-applicants-to-robot-training-roles/2026-10}},
}

Files: every figure in one CSV (a row per value) or as JSON, emailed to your work address.Other open data: datasets and API

Keep reading

Browse all 91 reports