Source mix and keyword lists shape regional gaps in our listings
LinkedIn feeds supply 34.0% of listings in the Americas and 93.3% in Africa, so a regional gap in discipline mix or pay text is first a source effect.
The Hired Creative listings export of 2026-11-04 holds 21,020 listings, first seen between 2026-06-03 and 2026-11-03. LinkedIn feeds supply 9,840 of them (46.8%), but the share runs from 34.0% of listings labelled Americas to 93.3% of those labelled Africa. The LinkedIn feeds for the US and Europe were also searched with a different keyword list from the other four. A gap between regions in discipline mix, pay text or level is therefore first a source and query effect, and not yet a market fact. This piece shows how to read the data. It does not show that the data are unrepresentative of any market.
Key points
- About a third of listings in the Americas and Europe come from LinkedIn feeds, against 72.6% in Asia-Pacific, 83.1% in the Middle East and North Africa, 93.3% in Africa.
- The US and Europe LinkedIn feeds have no AI or software keywords: 0.3% of rows (n=2,868) are AI, software or engineering titles, against 19.2% (n=6,839) in the other four.
- Our LinkedIn rows carry almost no pay text or job type (12 of 9,840 have pay text), so pay comparisons between regions mostly compare which boards are read.
- Until source and query are controlled, a regional gap in discipline mix, pay text or level is not yet a market fact.
Where the listings come from
Grouped by source name, 46.8% of the 21,020 listings come from LinkedIn feeds (95% interval 46.1 to 47.5). Employer career-site systems such as Greenhouse, Ashby and Workday supply 32.0% (6,735 listings). Remote, tech and design boards supply 16.2% (3,412), public-sector sites 2.3% (483) and other boards 2.6% (550). Three sources, Greenhouse, linkedin-asia and remoterocketship, hold 36.8% of all rows.
The export has six main LinkedIn feeds, each with more than 1,000 listings. Each searches a fixed set of 8 to 10 locations and 2 result pages per keyword, so its row count reflects search effort, not the number of live jobs.
LinkedIn share by region
In the Americas and Europe, employer career sites and boards supply about two thirds of listings. In Asia-Pacific, the Middle East and North Africa, and Africa, Hired Creative sees the market mostly through LinkedIn feeds.
Show the numbers
| Region | % | Listings (n) |
|---|---|---|
| Americas | 34% | 7,880 |
| Europe | 35.1% | 4,462 |
| Asia-Pacific | 72.6% | 3,544 |
| Middle East and North Africa | 83.1% | 2,415 |
| Africa | 93.3% | 1,062 |
Source: Hired Creative listings export, 2026-11-04 (n=21,020). Two further region labels in the export, worldwide and other, are not charted. All rows count in every total.
Two keyword lists
Hired Creative's feed configuration gives the US and Europe LinkedIn feeds one keyword list, called the main list here. The other four feeds (Middle East and North Africa, Africa, Asia, Latin America) use an extended list that adds AI, machine-learning, front-end, React and web-developer keywords.
The effect shows in titles. Titles containing a term that appears only in the extended list are 2.6% of rows in the two main-list feeds (95% interval 2.1 to 3.3) and 27.3% of rows in the four extended-list feeds (26.3 to 28.4). Titles classed as AI, software or engineering follow the same pattern.
Show the numbers
| LinkedIn feed | % | Listings (n) |
|---|---|---|
| US (main list) | 0.1% | 1,466 |
| Europe (main list) | 0.6% | 1,402 |
| Middle East and North Africa (extended list) | 18.5% | 1,758 |
| Africa (extended list) | 15.5% | 1,120 |
| Asia (extended list) | 19.6% | 2,720 |
| Latin America (extended list) | 22.6% | 1,241 |
Source: Hired Creative listings export, 2026-11-04 (n=21,020). Rows are collection feeds; the region label in the export differs from the feed for some rows.
The near-zero share in the two main-list feeds is produced by the keyword list, because the list contains no such keywords. It is not a finding about the US or European markets. Other sources in the same regions that Hired Creative reads, not collected with these LinkedIn keywords, show AI, software or engineering titles in 13.2% of Americas rows (n=5,197) and 11.2% of Europe rows (n=2,894). Those sources are a different set, skewed to tech and design employers. They are not a measure of the market's mix.
Among the four extended-list feeds, the AI, software and engineering share runs from 15.5% (Africa) to 22.6% (Latin America), and the share of PR and communications titles from 4.7% (Latin America) to 11.2% (Africa). These differences were not tested, and they cannot be separated from location-specific search effects. The US and Europe feeds cannot be compared with these four on discipline mix, because their lists differ.
What LinkedIn rows carry
Our LinkedIn rows carry almost no pay text or job type. Of 9,840 LinkedIn rows, 12 (0.1%) have pay text and none has a job type. Of 11,180 other rows, 1,599 (14.3%) have pay text and 5,141 (46.0%) have a job type. The level field is filled for 32.3% of LinkedIn rows and 49.3% of other rows.
Pay text is concentrated. Of 1,611 rows with pay text, 1,013 come from remoterocketship, 423 from Ashby and 109 from Pinpoint, together 95.9%. None of the three is a LinkedIn feed. Within the Americas, pay text appears in 0.4% of LinkedIn rows (10 of 2,683) and 23.8% of other rows (1,235 of 5,197). The same source gap sits inside one region, so a comparison of pay text between regions mostly compares which boards Hired Creative reads. This piece reports no regional pay or transparency gap. The missing job type in LinkedIn rows may reflect what the collection captures rather than what LinkedIn ads contain. The data do not show which.
What outside studies show
A 2014 technical report from the Georgetown University Center on Education and the Workforce states that online job ads overrepresent openings for college graduates and underrepresent openings for less-educated workers. A 2024 OECD benchmark of Lightcast vacancy data against official statistics covered Australia, Canada, the United Kingdom and the United States. It found regional representativeness considerably better than occupational and sectoral representativeness, and also that vacancy shares across sectors and occupations closely follow official data. It is cited here only as evidence that coverage varies by dimension. Neither study covers the Middle East, Africa or Asia, and neither measures Hired Creative's coverage.
Method
Data: the Hired Creative listings export retrieved 2026-11-04, 21,020 rows, first seen 2026-06-03 to 2026-11-03. All rows are in the source and region tables.
Source: a row is a LinkedIn row when its source name starts with "linkedin". The six main LinkedIn feeds are linkedin (US), linkedin-europe, linkedin-mena, linkedin-africa, linkedin-asia and linkedin-latam. Other LinkedIn-named sources are small and are included in every LinkedIn total. All other source names are assigned to a family (employer career-site systems, remote, tech and design boards, public-sector sites, other boards) by name-matching rules. The employer career-site family also holds bank, university and company career portals. The LinkedIn total does not depend on these rules.
Region: the region field of the export. Some rows from one regional feed carry the label of a neighbouring region, so regional shares are approximate.
Fields: a row has pay text when the has_salary_text flag is set, a job type when the job_type field is non-empty, and a level when the level field is non-empty.
Keyword lists: taken from the Hired Creative feed configuration. A title contains an extended-list-only term when it contains a keyword present in the extended list and absent from the main list.
Title classes: each title gets one class by keyword rules: design core, AI/software/engineering, social and content, video/motion/3D, PR and communications, or unclassified. For precision, 40 random titles per class were read by hand (seed 20261105). The share judged correct was 90.0% to 97.5% across the five classes. An earlier round on looser rules scored lower on two classes, so those rules were tightened and a fresh sample read. Recall was not measured.
Intervals: Wilson 95% intervals for shares. No permutation or bootstrap test was run, and no result here is a correlation across countries.
Places: country placement is rule-based, from the free-text location. Of 21,020 listings, 19,052 (90.6%) could be placed in a country, and 49 countries have at least 30 listings. Countries with fewer than 30 listings, and any whose separate reporting could identify a sensitive location, are pooled.
Minimum sample: 30 listings per country. The smallest LinkedIn feed has 1,120 rows. The Africa non-LinkedIn group has 71 rows and is not split further.
Limits
- Query and region are confounded. Only the US and Europe LinkedIn feeds use the main list, so no LinkedIn comparison can separate a market difference from a keyword-list difference between the US and Europe and the rest. The fair comparisons on discipline mix are among the four extended-list feeds, or between the US and Europe feeds.
- The non-LinkedIn comparison group is biased toward employers that use career-site systems and toward tech and design roles. It has 409 rows for the Middle East and North Africa and 71 for Africa.
- Row counts are not job-market sizes. New sources were added during the collection period. The export is a snapshot and shows no trend.
- About 10.5% of rows share a title and company with another row. The export is not de-duplicated across locations or sources.
- The title classifier is a keyword method. Forty titles per class is a small sample, so each precision figure has a wide interval. Recall is not measured: 2.0% of main-list feed rows, 2.7% of extended-list feed rows and 13.2% of non-LinkedIn rows are unclassified, including some real creative roles. The AI, software and engineering class includes some hybrid creative-AI roles.
- Country placement is rule-based, and the US state and city inference was not checked by hand. A listing placed in a country is not confirmed to be in the right one.
- Not shown: whether a regional gap is real once source and query are controlled. For most regions those controls are not available.
Sources
- Carnevale, Jayasundera and Repnikov, "Understanding Online Job Ads Data: A Technical Report", Georgetown University Center on Education and the Workforce, April 2014
- Tsvetkova, D'Amico, Lembcke, Knutsson and Vermeulen, "How well do online job postings match national sources in large English speaking countries?", OECD LEED Papers 2024/01
- Hired Creative listings export, 2026-11-04 (n=21,020; first seen 2026-06-03 to 2026-11-03).