Popular destinations
Skill trends, comparisons, salary context, resume help and long-form guides — jump straight to what brings people back.
Skill Scarcity Index — Open Dataset
Which tech skills are genuinely hard to hire for. Every day we combine three signals per skill — how long roles stay open (time-to-fill), the salary premium employers pay over the category median and how often the same role is re-posted after failing to fill — into a single 0-100 scarcity score. Demand tells you what's popular; scarcity tells you where hiring actually hurts. Free to download, analyse and republish — just credit the source.
As of July 2026, ClickHouse is the hardest tech skill to hire for in data roles, according to the Skill Scarcity Index, a daily composite of time-to-fill, salary premium and repost rate released free under CC BY 4.0.
Source: Datamata Studios — methodology and update cadence.
Hardest skills to hire for right now
The top of the index on Jul 12, 2026. Scores are percentile-rank composites within each category, so a 90 means "harder to hire than 90% of skills in its category" — comparable across categories of different sizes.
| Skill | Category | Scarcity score | Median days open | Salary premium | Repost rate |
|---|---|---|---|---|---|
| ClickHouse | Data | 96 | — | — | 40.9% |
| SIEM | Data | 87.6 | 5d | — | 40.9% |
| System Design | AI / ML | 84.5 | 61d | -3.2% | 19.3% |
| Pandas | AI / ML | 82 | 9d | — | 21.2% |
| Incident Response | Engineering | 81 | 51d | +61.8% | 5.9% |
| PyTorch | Engineering | 80.2 | 9d | +85.4% | 13.3% |
| Elasticsearch | Data | 79.3 | 54d | +151.2% | 3.2% |
| Elasticsearch | Engineering | 79.2 | 54d | +65.2% | 4.1% |
| Kubernetes | Security | 79 | 54d | +152% | 2.6% |
| SageMaker | Data | 78.3 | 11d | +66.9% | 17.6% |
| ETL | Product | 78.3 | 39d | — | 10% |
| GCP | Product | 78.1 | 45d | — | 1.1% |
| MongoDB | Product | 76.9 | 40d | — | 2.1% |
| SQL | Product | 76.3 | 26d | +81.8% | 6.7% |
| BigQuery | AI / ML | 75.5 | 4d | — | 22.2% |
Skills companies are adopting fastest
From the same pipeline, only published here: the number of companies whose job postings mentioned a skill for the first time in the last 90 days — an early adoption signal that leads the demand curve.
The stacks employers actually hire for
Skills aren't hired one at a time. These are the most common three-skill combinations appearing together in live postings (June 2026 rollup) — whole stacks, not just pairs. Also only published here.
What's inside
The download carries the full time series — one row per snapshot date, category and skill with each scarcity component and the composite score — so the trend itself is the data. The CSV is plain tabular data; the JSON wraps the same rows with metadata (version, licence, generation timestamp). Columns:
| Column | Type | Description |
|---|---|---|
| snapshot_date | date | UTC date the snapshot was computed (YYYY-MM-DD). |
| category | string | Job category: data, engineering, product, devops, security or ai. |
| skill_name | string | Canonical skill name from the extraction taxonomy. |
| demand_count | integer | Active listings mentioning the skill on the snapshot date. |
| demand_pct | number | demand_count as a percentage of all active listings in the category. |
| median_days_open | number | Median days recently-closed listings with this skill stayed open. Blank below the sample floor. |
| salary_premium_pct | number | Median disclosed salary of listings with this skill vs the category median, in percent. Blank below the sample floor. |
| repost_rate_pct | number | Share of this skill's listings that are re-posts of an earlier identical role (a failed-hire signal). |
| scarcity_score | number | 0-100 weighted percentile-rank composite within the category. Higher = harder to hire. |
How it is built
The pipeline snapshots every active job listing daily from public company career pages and job boards. Time-to-fill is the observed lifespan of closed listings; salary premium uses only disclosed salary ranges; repost detection matches identical company-and-title postings with disjoint visibility windows. Components below their sample floor are left blank rather than estimated. For the full collection method, sources and known limitations, read the data methodology.
Query it with the API
The same series is available as a read-only JSON endpoint with open CORS, so you can fetch it from the browser or a script. Filter by category and skill.
# The whole series
curl https://www.datamatastudios.com/api/datasets/skill-scarcity-index
# One skill's scarcity trend
curl "https://www.datamatastudios.com/api/datasets/skill-scarcity-index?category=data&skill=Rust"
# CSV instead of JSON
curl "https://www.datamatastudios.com/api/datasets/skill-scarcity-index?format=csv"Responses carry the same columns as the download, plus the licence, attribution and the bulk download URLs. Please keep the attribution link when you republish.
Licence & attribution
The Skill Scarcity Index is released under the Creative Commons Attribution 4.0 (CC BY 4.0) licence. You can share and adapt it for any purpose, including commercially, as long as you credit Datamata Studios and link back to this page. It is also published on Hugging Face and Kaggle.
Suggested citation
Datamata Studios. "Datamata Skill Scarcity Index." Jul 12, 2026. https://www.datamatastudios.com/datasets/skill-scarcity-index. Licensed under CC BY 4.0.