Analysis · November 3, 2026 · data as of November 3, 2026

What a row in the Hired Creative listings export is, and is not

Hired Creative's count of creative jobs is a count of 20,976 rows from 159 source labels, about a quarter of them not creative by title, and the export cannot show which rows are open or unique.

The Hired Creative listings export read on 2026-11-03 holds 20,976 rows. Each row has its own id, and the rows come from 159 source labels. About 47% of them come from LinkedIn searches. By a title keyword method, about three quarters of rows are creative or adjacent roles, and about a quarter are not. About 3% of rows repeat another row's title, company and location text. The export has no column that says whether a listing is still open.

So a Hired Creative count is a count of rows. It is not a count of distinct, open creative jobs. This piece sets out what a row is, how titles are classed, and what the export cannot tell. Later pieces rely on it.

Key points

• The export of 2026-11-03 has 20,976 rows from 159 source labels (n=20,976). A row is a listing as one source showed it, not a distinct open job.
• By a title keyword method, about three quarters of rows are creative or adjacent and about 13% are engineering titles. The method is approximate.
• About 3% of rows repeat another row's title, company and location text. The export cannot show whether a listing is still open.
• About 47% of rows come from LinkedIn searches, so what those searches ask for shapes the mix of roles.

Where the rows come from

The export has a source column. Hired Creative sorted the 159 labels into families by name. This grouping is Hired Creative's own and approximate, and the split between single-employer and public-sector sites in particular is not exact.

Three non-LinkedIn sources supply 27.3% of rows (5,722 of 20,976): Greenhouse (2,853), Remote Rocketship (2,135) and Ashby (734). At the other end, 65 of the 159 sources have fewer than 10 rows each, 290 rows in total. A figure about "creative jobs" is therefore mostly a figure about a few large sources. Any gap between groups in a later piece has to be checked against this mix first.

Repeats, and what the export cannot show

Rows with the same title, company and location text, after lowercasing, trimming and decoding HTML entities, are repeats of each other. There are 579 such extra rows, about 2.8% of 20,976. In 305 of the repeat pairs, about 1.5% of rows, the posted timestamp is also identical while the first-seen day differs. That points to the same listing being ingested again under a new id. The export has no URL, so none of this can be confirmed. Some repeats may be separate openings, and no unique-job count is given here.

Across sources, exact repeats are rare: 4 title, company and location groups appear under more than one source. Looser overlap cannot be sized. 195 title and company groups (421 rows) span several sources, mostly with different location text. These could be one job listed twice or two real openings.

The export also has no last-seen, expiry or URL column. Among the 19,109 rows with a posted date (91.1%; 1,867 rows have none, 8.9%), 36.2% were posted more than 90 days before 2026-11-03 and 6.4% more than 180 days before. An old posted date does not show that a listing is closed. It shows only that the export cannot be read as a list of this week's vacancies.

The first-seen dates run from 2026-06-03 to 2026-11-02. They reflect when sources were added and re-ingested. They are not a trend and say nothing about market growth.

How titles are classed

Each row's title is sorted into one class by keyword, with the first matching rule winning. Hired Creative's classes are:

• Core creative: design, art direction, motion, illustration, video, photography, copywriting and similar craft titles.
• Adjacent: communications, social media, content and editorial titles.
• Engineering: developer, engineer, data and similar titles.
• Set aside: design disciplines outside digital and graphic work (game, level, industrial, interior, lighting, service design), brand-management and marketing-management titles, and a few others such as architecture and retail.
• Unmatched: everything else.

Of 20,976 rows, 63.2% are core (13,250), 13.1% adjacent (2,742), 13.1% engineering (2,758), 4.3% set aside (897) and 6.3% unmatched (1,329). Rounded, about three quarters of rows are creative or adjacent by this method, about 13% are engineering titles, and the remainder are set aside or unmatched.

These are definitional choices, not facts about the jobs. A game designer, an industrial designer and a brand manager could each be counted as creative under another definition. The UK Department for Culture, Media and Sport, for one, counts creative work by occupation and includes creative occupations inside non-creative industries, and it revises its occupation list between releases. A title keyword method is cruder than an occupation code.

The category field is not a measure

Each row also carries a coarse category field. It does not match a title-based creative definition. The clearest case is creative-tech: of its 2,788 rows, only 3.8% are creative or adjacent by title, and almost all the rest are engineering titles. By contrast, 88.3% of the 8,805 design rows and 98.7% of the 2,008 motion rows are creative or adjacent by title. For brand (1,582 rows) the figure is about half, and it depends on the choice to set brand-manager titles aside.

What a search asks for

A feed finds only what its keywords ask for. Most of the documented regional LinkedIn feeds use a 26-term list that includes AI engineer, machine learning engineer, React developer and public relations. The main LinkedIn feed and some others use a 20-term creative list. Engineering rows from LinkedIn are therefore partly an effect of the query. Across all LinkedIn rows, by title, 68.0% are core creative, 15.8% adjacent and 12.3% engineering (n=9,823). The keyword lists do not explain the whole of the 12.3%, and no comparison between regions is made here.

Method

• Data. The Hired Creative listings export read on 2026-11-03: 20,976 rows (n=20,976), each with a distinct id, 159 distinct source labels, first-seen dates 2026-06-03 to 2026-11-02. A row is one listing as one source showed it.
• Source families. Source labels were grouped by name into six families: LinkedIn feeds; shared hiring platforms (Greenhouse, Ashby, Lever, SmartRecruiters, Teamtailor, Workable, Recruitee, Personio, Pinpoint); job boards and aggregators; single-employer or university career sites; public-sector job sites; and a pooled remainder. The grouping is by label prefix and name and is approximate.
• Repeats. Title, company and location text were lowercased, trimmed of extra spaces and had HTML entities decoded. Rows sharing all three were counted as repeats, and the count is the number of rows beyond the first in each group. The stricter test added an identical, non-empty posted timestamp. Cross-source overlap used the same keys with a different source label.
• Dates. Posted dates were read as UTC, including the "YYYY-MM-DD HH:MM:SS UTC" form. Age is the number of days between the posted date and 2026-11-03. Rows with no parseable posted date are counted separately and left out of age shares.
• Title classes. Titles were lowercased, trimmed and had entities decoded. Rules apply in this order. A title with an engineering term (developer, engineer, programmer, data scientist, machine learning, devops, software, full-stack, front-end, back-end, QA, scientist, analyst, recruiter) is engineering, unless it also names a creative role such as design director, art director, creative director, UX, UI, product design, graphic, visual, motion or brand design. Next, titles with a set-aside term (for example game, level, industrial, interior, fashion, mechanical, civil, service designer, architect, retail, merchandising, brand manager or marketing manager) are set aside, unless they also name graphic, visual, motion, art director, creative director, copywriter, illustrator, UX or UI. Next, a broad list of creative-craft terms (design, designer, graphic, UX, UI, motion, illustration, animation, art director, creative, video editor, photographer, copywriter, artist, VFX, 3D artist and similar) gives core. Next, communications, social media, content, editor, writer, journalist and similar terms give adjacent. Anything left is unmatched. Keywords are mainly English with a few Spanish and French terms.
• Checks. Precision was hand-read by one reader on 40 random titles per class. Most matched their class, but samples of 40 give wide intervals. An independent re-check found false positives in the core class, such as French public-service titles matched by "animation", and false exclusions, such as product designer titles containing "retail". The net effect on the headline is about 1 point, so only rounded figures are quoted.
• Category check. Each coarse category was cross-tabulated against the title class.
• Pooling. Countries with fewer than 30 listings, and any whose separate reporting could identify a sensitive location, are pooled.
• Tests. No significance test or interval is used. Every figure is a count or share of the stated base.

Limits

• This piece does not describe the collection rules: what the collectors filter out before a row reaches the export, what the dedup step does, or when a row leaves. The export alone cannot show them.
• "Live" cannot be checked from the export, and no figure here should be read as a number of open jobs.
• Repeats are judged on text only, because the export has no URL. They are not confirmed duplicates.
• The title classifier is an approximation built on keywords and read by one person. Titles in non-Latin scripts and other languages are probably under-matched. A critic could set the class boundaries elsewhere, for example for game designers, UX research or brand managers, and get a different share.
• The source-family grouping is a convenience, not a classification of the sources by kind.
• This is a snapshot of one export. First-seen dates are not a trend.
• No pay, level, region or country comparison is made. Those fields depend on which sources are read.

Sources

• DCMS Sectors Economic Estimates: Creative Occupations Update Note, UK Department for Culture, Media and Sport, published 16 July 2026.
• Hired Creative listings export, 2026-11-03 (n=20,976).

Published by Hired Creative from its own listings and the sources above. Spotted an error? Tell us.