TLDR – The 30-Second Read
Across roughly 24,500 candidate records in our database, we count close to 13,200 different ways of writing a job title. That is nearly one new title for every two people. AI sourcing works well where talent is active online and consistently labelled, which describes software engineers far better than power and utilities engineers. For these disciplines the title is an unreliable signal, so sourcing becomes a recognition problem, not a search problem.
We went looking for a clean way to describe our own candidate base and found a mess that turned out to be the point. As at June 2026, across roughly 24,500 candidate records, there are close to 13,200 distinct free-text job titles. For every two people on file, there is almost one brand-new way of writing a role.
That matters because the loudest story in talent acquisition right now is AI. Korn Ferry’s 2026 research has 84% of talent leaders planning to use it this year, and the sourcing platforms increasingly run as infrastructure rather than novelty. Most of that promise rests on one quiet assumption: that the candidate is easy to find and easy to label. For a large part of the energy and utilities workforce, neither holds.
Where does AI sourcing actually earn its keep?
AI sourcing is genuinely good at a specific shape of problem. It works when roles are high in volume, when candidates are active and searchable, and when the signal in a profile is consistent.
Software talent is the model case. Developers leave a public trail: GitHub commits, Stack Overflow answers, conference talks and tidy skill tags. The leading “technical” sourcing platforms are built directly on that trail. SeekOut sources developers on their code contributions, while AmazingHiring and hireEZ aggregate GitHub, Stack Overflow and similar communities to rank engineers on visible output.
For that population, title-and-token matching is fast, accurate and a real-time saver. We are not arguing against the tools. We use technology heavily ourselves. The question is what happens when you point that machinery at a workforce that doesn’t behave like a software developer.
So why do power and utilities engineers break the model?
A protection engineer does not push commits to a public repo. A commissioning lead does not answer Stack Overflow questions about energising a substation. These people are on LinkedIn, but they are mostly passive, and the one field the tools lean on hardest, the job title, is wildly inconsistent.
The same work gets written a hundred different ways. That is not a hunch. It is what our own database shows, inside a clean and curated system that should flatter consistency, not expose it. If the title fragments this much in a tidy CRM, it almost certainly fragments worse on the open web.
What does the fragmentation actually look like?
We took every candidate whose title touches a given discipline and counted how many distinct title strings those people use. The pattern held across every discipline we checked.
- Power systems is about as standardised as this market gets, and even there roughly 600 engineers describe themselves with about 270 different titles
- Commissioning: around 240 people under roughly 155 titles
- Substation: around 135 people under roughly 110 titles
- Protection: around 125 people under roughly 100 titles
- High voltage: around 115 people under roughly 105 titles, close to one unique title each
- SCADA: around 60 people under roughly 55 titles
Based on Talesca candidate records as at June 2026, across roughly 24,500 non-deleted profiles. These counts are directional, and they almost certainly understate each pool for reasons the next section makes clear.
The further you move towards the scarce, specialised disciplines that are hardest to fill, the closer the ratio gets to one title per person. The roles where sourcing precision matters most are the roles where the title signal is weakest.
Why does searching the obvious keyword miss the people you need?
Here is the part that should concern anyone relying on keyword or AI title matching. Search our database for the literal phase “high voltage” and you find 31 people. The actual HV pool is closer to 115, because most of those engineers simply write “HV”. Search the spelled-out term and you surface roughly one specialist in four.
Protection behaves the same way. A meaningful slice of protection and control work sits under titles like “secondary systems”, with the word protection nowhere in sight. A couple of dozen of our candidates title themselves that way, and a keyword net for “protection” never sees a single one of them.
Our own counts undercount for exactly this reason. That is the argument, not a caveat to it.
And why does it surface the wrong people at the same time?
While the net misses the right people, it quietly catches the wrong ones. A search for “protection” also returns the occasional fire-protection, cathodic-protection, surface-coating or business-development role. The volume is small, but the effect compounds. The tool hands back a shortlist that is at once too thin on real specialists and padded with near-homonyms.
A recruiter who knows the discipline reads “secondary systems” and thinks protection. A keyword reads “secondary systems” and thinks nothing at all.
What does all this mean for how you source roles?
For these disciplines, sourcing is a recognition problem, not a search problem. The skill that matters is knowing that “secondary systems”, “HV”, “P&C” and a dozen other phrasings point at the same scarce person, then holding the network to reach them when they are not looking. That is human pattern-matching, and it is the one thing a title field cannot encode.
It also lines up with where the wider market is landing. Korn Ferry’s 2026 research has 73% of talent leaders ranking critical thinking as their top hiring priority, ahead of AI skills, precisely because someone has to judge what the tool returns and know when to override it. Deloitte’s 2026 human capital work points the same way, putting human judgement and adaptivity ahead of pure technology differentiation.
The honest position is also the useful one. Use AI sourcing where it earns its keep: active, high-volume, well-labelled roles. For the scarce technical disciplines that keep the grid standing up, a specialist who recognises the work behind the title will still beat the search box.
If you are hiring into power systems, HV, protection, commissioning or any of the disciplines where the right people rarely carries the obvious title, that recognition is the work we do. Talesca runs curated, specialist search across energy, utilities and engineering, and we hold the networks in these disciplines rather than rediscovering them keyword by keyword. If that is the kind of role you are trying to fill, we would be glad to compare notes on the market.
Frequently Asked Questions
It depends on the engineer. For software and digital roles, where candidates are active online and described consistently, AI sourcing is fast and effective. For physical-engineering disciplines like power systems, protection or high voltage, the job title is inconsistent and the candidate is usually passive, so title-and-keyword matching fragments and misses people. The tools help most where the talent is well-labelled, and least where it isn’t.
The work is specialised and the language for it was never standardised. The same role gets written as power systems, secondary systems, HV, protection and control, or a dozen company-specific variants. Across roughly 24,500 candidate records in our database we count close to 13,200 distinct titles, nearly one new title for every two people. There is no shared naming convention the way there is for, say, a frontend developer.
Because most of them never write “high voltage.” In our database the literal phrase returns about 31 people, while the real HV pool is closer to 115. The difference is engineers who simply write “HV.” A keyword or AI search anchored on the spelled-out term finds roughly one specialist in four, and quietly drops the rest from the shortlist before a human ever sees them.
No. The point is to match the tool to the role. Use AI sourcing for high-volume, active, well-labelled positions where it saves real time. For scarce technical disciplines where the title is unreliable, lean on recruiters who recognise the work behind the title and hold the network to reach passive specialists. The most reliable approach combines both, and knows which to trust for which role.
By recognising the work rather than the label. An experienced specialist recruiter knows that “secondary systems,” “P&C” and “HV” can all point to the same scarce person, and keeps direct relationships with that community so they can be reached when they aren’t job-hunting. It is pattern-matching plus a maintained network, which is exactly the signal a job-title field can’t capture.
Sources & references
- Korn Ferry, Talent Acquisition Trends 2026 (press release, 84% AI adoption; 52% AI agents). https://www.kornferry.com/about-us/press/korn-ferry-research-unveils-top-talent-acquisition-trends-shaping-2026
- Korn Ferry, Talent Acquisition Trends 2026 — full report overview (critical thinking as top skill; AI as infrastructure framing). https://www.kornferry.com/insights/featured-topics/talent-recruitment/talent-acquisition-trends
- Deloitte, 2026 Talent Acquisition Technology Trends / 2026 Global Human Capital Trends (agentic AI; human judgement over technology differentiation). https://action.deloitte.com/insight/5005/2026-talent-acquisition-technology-trends-the-new-imperative
- daily.dev Recruiter, Top AI Tools for Sourcing Developers (SeekOut Coder Score; AmazingHiring/hireEZ GitHub and Stack Overflow aggregation). https://recruiter.daily.dev/resources/best-ai-tools-sourcing-developers/
- Leonar, 13 Best Talent Sourcing Platforms Compared (2026) (SeekOut and AmazingHiring built on GitHub/Stack Overflow/code repositories). https://www.leonar.app/blog/best-talent-sourcing-platforms/
- Metaview, Top 10 sourcing tools for recruiters in 2026 (AmazingHiring for engineering; developer-community signal). https://www.metaview.ai/resources/blog/sourcing-tools-for-recruiters