Within the same company, remote wording shows no detectable discipline gap
In 4,996 listings from 11 applicant-tracking systems, remote wording differs by discipline when pooled, but no gap is detectable within the same company.
In Hired Creative's listings, AI and machine-learning engineering roles carry remote wording more often than brand and graphic roles: 41.2% of 582 listings against 19.8% of 524. The base is 4,996 of 20,581 listings (export of 2026-10-22), those from 11 applicant-tracking-system (ATS) sources whose location text can show remote status. Within the same company the gap is not detectable. Across 95 companies posting both UX and brand roles, the average remote share is 33.3% for UX and 35.6% for brand and graphic, and 70 of the companies tie. The test has low power and cannot show that discipline never matters.
Key points
- Among 4,996 listings from 11 applicant-tracking systems, AI and machine-learning engineering roles use remote wording more often (41.2%) than brand and graphic roles (19.8%), before adjusting for company.
- Within the same company, no discipline gap is detectable: in 95 companies posting both UX and brand roles, remote wording averaged 33.3% for UX and 35.6% for brand.
- Companies differ sharply: of 134 with 10 or more listings, 63 had no listing with remote wording and 15 had only listings with it.
- It covers 4,996 of 20,581 listings, reads remote wording in location text (a lower bound), and cannot show that discipline never matters.
What the data covers
Remote status can be measured for 4,996 of 20,581 listings (24.3%). The 9,750 LinkedIn listings (47.4%) never contain a remote word in their location text. The 2,926 listings from remote-only boards (14.2%) have no onsite roles to compare, and 2,909 more come from sources with no work-arrangement text. The measured set is a few thousand listings from 472 companies, 56.4% of them (2,816) in the region value "americas". It is not the whole creative market and supports no regional comparison.
The source moves the number. Remote wording appears in 63.3% of Ashby listings (n=768), 28.0% of Lever (n=243), 16.4% of Greenhouse (n=2,781) and 2.7% of Personio (n=110). These gaps reflect which employers use which system and how each system words location, not board quality or market behaviour. Any pooled figure is partly a source-mix figure.
Show the numbers
| Source | % | Listings (n) |
|---|---|---|
| Ashby | 63.3% | 768 |
| Lever | 28% | 243 |
| Greenhouse | 16.4% | 2,781 |
| Personio | 2.7% | 110 |
Source: Hired Creative listings export, 2026-10-22 (n=4,996 ATS listings; four of 11 sources shown). Ashby and Lever state workplace type in a structured prefix or suffix; the other sources are read from free text.
Pooled by discipline
Pooled across the 11 sources, the share with remote wording differs by discipline.
Show the numbers
| Discipline | % | Listings (n) |
|---|---|---|
| AI and machine-learning engineering | 41.2% | 582 |
| UX and product design | 34.4% | 761 |
| Copywriting | 26% | 192 |
| Communications and PR | 23.6% | 242 |
| Social and content | 19.9% | 296 |
| Brand and graphic | 19.8% | 524 |
| Motion, video and 3D | 18.2% | 457 |
| Art and creative direction | 18.2% | 308 |
Source: Hired Creative listings export, 2026-10-22 (n=4,996 ATS listings). Not a ranking of all eight: most 95% intervals overlap. Shares are not adjusted for source or company.
Company-cluster 95% intervals are wide. AI and machine-learning engineering (41.2%, interval 30.0 to 52.6) sits above brand and graphic (19.8%, 13.7 to 27.2), motion, video and 3D (18.2%, 11.8 to 27.2) and art and creative direction (18.2%, 11.3 to 26.7). UX and product design (34.4%, 27.4 to 41.6) clears brand and graphic only just. The other buckets overlap and are not ranked.
AI and machine-learning engineering is not a design discipline, and the brand and graphic bucket was only about 80% precise in hand reads. With exact repeats removed, no pooled share moved by more than about 1.4 points.
Same company, no detectable gap
If discipline set remote status, the spread between disciplines would be larger than the spread left after shuffling labels within companies. The test covers 3,495 classified listings from 429 companies, 260 of which post in more than one discipline. Of 2,000 shuffles, 1,766 matched or exceeded the observed spread (permutation p about 0.88; about 0.79 with exact repeats removed).
In the 95 companies posting both UX or product design roles and brand or graphic roles, with each company weighted equally, the mean remote share is 33.3% for UX and 35.6% for brand and graphic. UX is higher in 11 companies, brand and graphic in 14, and 70 are tied because all of their roles carry remote wording or none do.
Show the numbers
| Measure | % | Listings (n) |
|---|---|---|
| UX and product design, pooled (n = listings) | 34.4% | 761 |
| Brand and graphic, pooled (n = listings) | 19.8% | 524 |
| UX and product design, within 95 companies (n = companies) | 33.3% | 95 |
| Brand and graphic, within 95 companies (n = companies) | 35.6% | 95 |
Source: Hired Creative listings export, 2026-10-22 (n=4,996 ATS listings). The 95 companies are a subset of the pooled listings, so the two pairs are not the same sample.
Given who the employer is, discipline adds no detectable remote signal in this sample. Differences of several points cannot be ruled out, and employer cannot be separated from the way its board words location.
Employers differ from each other
Of 134 companies with at least 10 listings in the ATS sources (78.7% of those listings), 63 had no listing with remote wording, 15 had only such listings, and 56 were mixed. Most of the 63 are on free-text sources, so "none" means no remote wording in the export, not an onsite-only employer. Company is the name string in the export, so subsidiaries count separately.
Outside figures point the same way on different measures
Two outside sources put engineering and software roles above design and arts roles. In the US Current Population Survey for September 2026, 63.7% of people at work in computer and mathematical occupations teleworked or worked at home for pay in the reference week, against 44.4% in arts, design, entertainment, sports and media occupations and 21.9% of all people at work. In Indeed Hiring Lab's remote tracker, 30.4% of US postings in its techsoftware category mentioned remote or hybrid work on 2026-09-30, against 25.7% for marketing, 23.0% for media and 19.2% for arts. Hired Creative's pooled order has the same direction, but the three measure different things, so none confirms or refutes the within-company result.
An NBER working paper (Hansen, Lambert, Bloom, Davis, Sadun and Taska, no. 31007, not peer-reviewed), covering more than 500 million vacancy postings in five English-speaking countries, reports large differences in the share of postings that offer remote work "even when zooming in on employers in the same industry competing for talent in the same occupations". It does not study creative roles and is not cited as support for the result above.
Method
Data. The Hired Creative listings export of 2026-10-22, 20,581 listings. It has no remote field, so remote status was read from location text for the sources that can show it.
Remote definition. Ashby: the location text starts with "Remote". Lever: the workplace-type suffix is "remote". Nine free-text sources (Greenhouse, SmartRecruiters, Teamtailor, Workable, Recruitee, Pinpoint, Personio, YC, designengineer): the location text or title contains "remote", "work from home", "home office" or a close equivalent, including a few non-English terms. "City or Remote" counts, so the measure is remote-eligible and a lower bound. The 11 sources give 4,996 listings from 472 companies. Other sources are not measured but stay in the totals quoted.
Disciplines. One discipline per title from keyword rules. All 65 designengineer listings count as UX and product design by source. For other sources, a list of non-creative design terms and the phrase "visual communication" are checked first, then ordered rules, first match wins: AI and machine-learning engineering, UX and product design, software engineering, motion, video and 3D, social and content, communications and PR, copywriting, art and creative direction, brand and graphic. A software engineering bucket (133 listings, not charted) is part of the 3,495 listings in the within-company test. Unclassified titles (857) and non-creative design titles (644) are excluded from it.
Statistics. Minimum 30 listings per reported discipline. Discipline intervals are 95% company-cluster bootstrap intervals (1,000 resamples); source intervals are Wilson. Within-company test: discipline labels shuffled within each company 2,000 times, comparing a chi-square-like statistic with the observed 97.6. Paired comparison: the 95 companies with at least one UX or product design and one brand or graphic listing; each company's remote share per discipline, averaged with equal weight. Discipline shares and the permutation test were rerun without exact repeats (same company, title and location text).
Outside data. BLS Current Population Survey table A-42 (September 2026) and the Indeed Hiring Lab sector file (to 2026-09-30), as published.
Limits
- Coverage. The measured set is 24.3% of listings and not the whole creative market. Over half of it is in the region value "americas", and the ten largest companies hold 24.2% of its rows. No regional comparison is made.
- Definition. Remote wording is a lower bound. Boards word location differently, so employer and board convention are one variable here.
- Disciplines. Titles are a rough guide. Generic titles such as "Designer" cannot be told apart from product or fashion design. Precision for the other buckets (26 to 28 correct in 30) was read before the last rule changes.
- Power. 70 of 95 paired companies tie. The result supports "no detectable discipline effect once the employer is known" and nothing stronger.
- Worldwide hiring. Of 1,211 ATS listings with remote wording, 236 (19.5%) say only "Remote" and 9 (0.7%, all Lever) use worldwide, anywhere or global wording. Short location text cannot show where an employer would hire, so the data do not say which employers open roles to candidates anywhere.
- Outside data. BLS counts workers, and its arts and media group includes entertainers and athletes. Indeed counts postings in its own categories.
- Time. One snapshot. No trend is claimed.
Sources
- BLS, Current Population Survey table A-42, People at work by telework status (September 2026)
- BLS, Telework definition
- Indeed Hiring Lab, remote tracker and its sector file (CC BY 4.0)
- Hansen, Lambert, Bloom, Davis, Sadun and Taska, "Remote Work across Jobs, Companies, and Space", NBER Working Paper 31007
- Hired Creative listings export, 2026-10-22 (20,581 listings).