The August jobs report surprised almost everyone. Reuters reports that nonfarm payrolls rose by 162,000 in August, the largest gain in five months and nearly three times the 56,000 economists had forecast. Read the composition, though, and the picture for technical hiring is very different: the information sector shed 23,000 jobs, professional and business services added only 10,000, and government plus leisure and hospitality accounted for more than 60% of the gain.
The turnover data says the same thing more bluntly. The Bureau of Labor Statistics JOLTS release for July shows job openings little changed at 7.3 million and hires little changed at 5.1 million, with hires in professional and business services down 188,000. Economists quoted by Reuters still describe the market as "slow-hire, slow-fire." If you lead IT, security or infrastructure at an operating company and you are budgeting for sourcing software this quarter, that is the market you are buying for: openings exist, but every hire is slow, deliberate and expensive to get wrong.
This guide is written for the CIO, CISO or VP of IT Operations who owns a hardware, firmware, test, network, cyber, SRE, field-service, ERP or data-center hiring plan and has been asked to justify a software purchase. It covers what a slow-hire market changes about the evaluation, the questions to put in the RFP, and the red flags that should end a demo early.
What does a slow-hire market change about buying sourcing software?
In a fast market the constraint is volume: you need more candidates, faster, and almost any tool that widens the top of the funnel looks good. In a slow-hire market the constraint flips. Requisitions are fewer, each one carries a longer approval trail, and the cost of a wrong hire is felt for a year. The software question becomes: can this tool find the specific person for a specific technical role, and can my own recruiters prove it did?
The applicant side has not slowed down. Axios reports, citing recruiting-software company Ashby, that recruiters now process roughly 291 applications per hire, up from about 100 in early 2021, and that, according to Indeed, it takes almost 25% longer to fill a role than before the pandemic. More inbound and fewer hires means the inbound pile is now mostly noise. A tool that mainly helps you sort noise faster is solving last cycle's problem.
Identity has also become a screening problem. The Wall Street Journal reports that employers are asking remote candidates to remove virtual backgrounds and pan the camera around the room, because of AI-assisted answering, deepfakes and remote workers misrepresenting where they live. For a buyer hiring into network, security or data-center roles, that is not a curiosity; it is a control requirement.
Which roles should the evaluation be built around?
Do not evaluate on a generic "software engineer" requisition. The roles that stay hard to fill in a slow-hire market are the ones where the candidate has to have touched the physical or operational system: embedded and firmware engineers, hardware test and QA, network and infrastructure engineers, security operations and OT-security specialists, SREs, field-service technicians and leads, ERP administrators and data-center operations staff. The JOLTS data supports the emphasis: durable goods manufacturing was one of the few sectors where openings rose in July, up 76,000, while August payrolls added 16,000 in manufacturing and 22,000 in construction even as the information sector contracted.
Pick two or three of those roles you actually have open, in the U.S. or Canadian markets where you hire, and make every vendor run its search against them during the pilot. The output you want is a shortlist your hiring manager can interrogate, not a count of profiles found.
What should be in the RFP?
Five questions separate a sourcing platform your recruiters will operate from a tool that becomes shelfware:
- Who operates it? The answer should be your own recruiting team, with the vendor providing the software rather than a service desk that runs searches on your behalf. Insourced hiring at an operating company only works when the people accountable for the requisition control the search.
- Where does the candidate data come from, and under whose license? Ask whether the platform works through credentials your company already holds for sourcing databases and professional networks, or whether it collects data in ways you could not defend in an audit. Your CISO should read this answer.
- How does it evaluate technical evidence? Ask the vendor to show how it distinguishes a firmware engineer who owned validation on a shipped product from one who lists the keyword. Keyword density is not evidence; system ownership, test coverage, incident response and field outcomes are.
- What is the identity and consent posture? Given the fraud environment described by the Journal, ask what the platform does to keep synthetic or misrepresented profiles out of the shortlist, and how it records candidate consent for outreach.
- What is the audit trail? Every search, score and message should be reconstructable. In a slow-hire market, procurement and legal will ask why a candidate was surfaced or excluded, and "the model decided" is not an acceptable answer.
How should you run the pilot?
Keep it short and adversarial. Give each vendor the same two or three live requisitions, the same recruiter, and a fixed window of two to three weeks. Measure four things: how many of the surfaced candidates the hiring manager agreed were genuinely qualified; how many replied to outreach; how much recruiter time the tool consumed versus saved; and whether every step could be explained afterward from the platform's own records. A vendor that cannot expose that record during a pilot will not expose it in production.
Two comparisons are worth making explicitly. First, compare the tool against your own inbound funnel on the same roles; if the platform's shortlist is not clearly better than what 291 applications per hire already gives you, it is not worth the license. Second, compare it against the agency fee you would otherwise pay for the same role. In a slow-hire market with fewer requisitions, the license only pays back if it replaces spend you would actually have incurred.
What are the red flags that should end a demo?
- The vendor cannot say which data sources it uses or under what license.
- Scores are presented without the underlying evidence, or the evidence is a restatement of the résumé.
- The demo is run only on software-developer requisitions and the vendor cannot switch to a network, firmware or field-service role on the spot.
- The pricing model is tied to volume of profiles viewed or messages sent rather than to roles filled or seats operated; volume pricing rewards the noise you are trying to escape.
- There is no answer to the fraud question at all.
Where does UPPER fit?
UPPER is sourcing software that an operating company's own recruiters run for hardware, embedded, firmware, test, QA, cyber, network, SRE, field-service, ERP and data-center roles across the United States and Canada. It works through the sourcing credentials the customer already licenses, keeps an audit trail of every search, score and message, and returns an evidence-backed shortlist rather than a profile count. That is the standard this guide asks you to hold every vendor to, including us. Run the same two or three requisitions through UPPER and through whatever else is on your list, and let the hiring manager decide.
What is the bottom line for IT leaders budgeting this quarter?
August's 162,000-job headline does not mean technical hiring got easier. Openings are flat at 7.3 million, hires are flat at 5.1 million, the information sector is shrinking, and applicant volume per hire has nearly tripled since 2021. In that market the right sourcing software is the one your recruiters operate themselves, that finds evidence rather than keywords for the roles you actually have open, and that can defend every shortlist to security, legal and the hiring manager. Buy for the market you are in, not the one the headline describes.