What employers actually name as skills
Python leads, English is second and Excel outranks SQL. What company career pages name when they list a skill, and what that leaves out.
Jobzyl3 min read
Figures on this page are current as of
Every "most in-demand skills" list you have read was built from a survey of people who were asked to guess. This one is built from what employers wrote down, across 148,263 open roles at 985 companies, read off their own career pages.
The result is less flattering to the industry than the survey version, and more useful.
What gets named
Show the numbers
| Category | Value |
|---|---|
| Python | 15704 |
| English | 12869 |
| AWS | 9596 |
| Excel | 9413 |
| SQL | 8170 |
| SAFe | 8167 |
| Project Management | 7411 |
| Salesforce | 7126 |
| Machine Learning | 7023 |
| Agile | 6633 |
| Kubernetes | 6548 |
| CI/CD | 6506 |
| Azure | 6457 |
| CRM | 6291 |
| Business Development | 6151 |
Data as of
Python leads, which surprises nobody. The two highlighted bars are the story.
English is the second most-named skill on the list, ahead of every framework and every cloud platform. It appears because a large share of these employers are hiring across borders and feel the need to say so, which means that for a sizeable slice of this market the qualifying requirement is a language rather than a technology.
Excel then outranks SQL, by a clear margin, on the career pages of companies that run their own engineering on Greenhouse and Ashby. The spreadsheet is the most commonly named analytical tool here, ahead of the thing everybody recommends learning instead of it.
Below those two the list is unremarkable and that is worth saying: AWS, SQL, Salesforce, Kubernetes, Azure, CI/CD. The infrastructure names cluster. If you are choosing what to learn next, the chart's middle is a flat field rather than a ranking with a winner.
What the list is not
This counts what employers wrote, not what a job needs. Those come apart in three ways, and all three matter more than the ordering does.
A posting that says "strong communication skills" contributes nothing, because that is not a skill name our extractor recognises. So the chart is biased toward things with proper nouns, and it systematically under-counts everything a recruiter would call a soft skill. The absence of teamwork from this list is evidence about our parser, not about employers.
SAFe sits higher than most engineers would predict. That is a real signal about who is hiring, since scaled agile frameworks are named by large organisations with formal delivery processes, and those organisations post a lot of roles. It is not a signal that you should go and learn it.
And a posting naming a skill is not a posting requiring it. "Nice to have" and "must have" read identically to a text extractor. Nobody should read this chart as a list of gates.
65%
of postings name at least one recognised skill
The rest named none our extractor knows, which is not the same as naming none at all. Adzuna and Jooble truncate their descriptions, so a posting can lose its skills section before we ever see it.
Data as of
That is the denominator, and it is the caveat that matters most. Roughly a third of this census contributes nothing to the chart above, largely because we could not read a long enough description to find anything.
The useful conclusion
The gap between what this market asks for and what gets written about it is measurable, and it is wide. English and Excel are the second and fourth most-named skills on the career pages of modern technology companies, and neither appears on any "skills of the future" list we have read.
If you are writing a CV, the practical move is to name the plain tools you actually use, because the postings do. If you are choosing what to learn, the middle of that chart says the marginal return on the eighth cloud platform is lower than it feels.
You can see the skills named on individual postings in data engineer roles in the United Kingdom or remote machine learning engineer roles.
Method
Distinct postings, collapsed on title, company and location before anything is counted. One posting names several skills, so the bars do not sum to the census and were never meant to.
Skills are extracted from the posting text against a known vocabulary. That vocabulary is the whole limitation of this post and it cuts one way: a skill we do not know about cannot appear, so this chart can only understate the variety of what employers ask for.
The census covers companies publishing on Greenhouse, Lever, Ashby and SmartRecruiters. Their whole boards are read with no search term, so within that set the mix is a census rather than a sample. Which companies are in the set is decided by our own discovery and is not random, so this describes these companies and not "the market".
Related
- Why entry-level jobs are so hard to findOf the roles that state a level on a company careers page, the overwhelming majority ask for senior. Junior and intern are a rounding error.
- Job boards overlap far less than you thinkThe same posting rarely turns up on more than one board. The repetition you notice is one board relisting, not boards copying each other.
- What 985 companies are actually hiring forA census of 148,071 open roles across 985 company career pages. Software engineers lead, and sales outnumbers every data and AI role combined.
Every figure here comes from the jobs Jobzyl has indexed, and carries the date it was pulled. Subscribe by RSS, or run the search yourself.