How Has Your Industry’s Recruiting Evolved? 25 Outdated Strategies to Avoid

Recruiting in supply chain has always been complicated, but it’s become even more complicated now because of AI.

For example, a supply chain manager can mean completely different things depending on the company. In one organization, you might need someone who wears a lot of hats, someone with experience in procurement, logistics, operations, warehouse management, demand planning, maybe even customer service. But in another company, a supply chain manager might be someone who’s really focused on procurement. So you can’t just look at the title and assume you know what the person does.

Now you also have AI in the conversation. Some companies are starting to look for people who have AI experience or at least know how to work with AI. So I think it’s becoming even more important to understand the industry and the actual role.

And I think that’s one thing that used to work but doesn’t work as well anymore, which is relying heavily on job boards and applications. You can post a role and get hundreds or even thousands of applications, but a lot of them are not actually a fit. They might have the right keywords because AI has made it easier to optimize resumes and get through an ATS, but that doesn’t mean they can actually do the job.

So now I think you have to be much more proactive. Instead of just waiting for candidates to apply, you have to go out and find them, especially for specialized and senior roles.

And again, it comes back to understanding the role and the industry. If you don’t understand what the company actually needs, it’s very easy to match someone based on a title or keywords and end up with the wrong person.

“I would say the clearest change is how much less a resume tells us. Two years ago a well-written application/resume was a reasonable first filter. Now nearly every candidate uses AI to write and tailor their applications. I don’t think that’s wrong; it’s a tool that’s available to them. But the applications we see for one role read far more alike than they used to, and the most polished ones don’t reliably come from the strongest people. So the strategy that stopped working for us is screening on the resume and using the interview to confirm what it said. We put that weight on a first screening interview instead: our recruiters ask candidates to walk through a specific piece of their own work, why they made the decisions they made, what they would change now. You can’t prepare a generated answer for a follow-up question you didn’t see coming, and that live conversation is where you find out whether the judgment belongs to the person. It has also made the AI question easier to answer when clients raise it.”

—Maria Jose Lorenzo, Director of Talent Acquisition, Hire With Near (https://www.hirewithnear.com)

Bio: Maria Jose Lorenzo is Director of Talent Acquisition at Hire With Near, where she leads a recruiting team of more than 40 people placing Latin American professionals into remote roles with US companies.

In remodeling, recruiting has shifted from “find a good carpenter” to “find someone who can think like a builder, communicate like a project manager, and respect a lived-in home.” On high-end kitchens, baths, aging-in-place work, and full-home renovations, the technical skill matters, but so does cleanliness, client communication, sequencing, and problem-solving.

One thing that used to work was hiring mostly through word-of-mouth and assuming a skilled tradesperson could just “figure it out” on site. That breaks down fast when you’re doing complex remodels where the homeowner is living there, selections are coming through a showroom/design process, and small mistakes ripple into schedule and budget issues.

Now I look harder for attitude, reliability, and the ability to work inside a system. For example, on bathroom remodels with aging-in-place features, I’d rather train a detail-oriented person on our way of doing things than hire someone who has talent but won’t follow layout, waterproofing, or communication standards.

My best recruiting strategy now is to show candidates the actual standard before hiring them. Walk them through a real project, explain how we protect the home, handle change orders, coordinate design selections, and see if they respect the process or just want to swing a hammer.

The biggest change I’ve seen is that the C-suite used to be judged primarily on what they’d accomplished. Now we’re looking much harder at how they accomplished it—and whether that approach will still work five years from now.

AI is obviously a big part of that. The technology itself matters, but I’m less interested in whether an executive can talk intelligently about AI than whether they can lead through a period of enormous change. Can they rethink a business model? Can they make good decisions when the old assumptions no longer hold? Can they get a team to actually adopt new ways of working?

One strategy that used to work extremely well but doesn’t anymore is selling the executive’s track record as the whole story. That easily becomes a liability in a fast-paced world, especially if someone is too attached to the playbook that created it.

The executives who stand out now are the ones who have demonstrated adaptability without becoming reactionary. They know what to hold onto, what to throw out, and when they need to learn something completely new.

That’s where I think the AI conversation gets interesting. The executives most likely to thrive aren’t necessarily the ones who know the most about AI. They’re the ones who are comfortable saying, I’ll figure it out.

In cross-border corporate services, the first thing that died was the take-home writing sample. An essay prompt now measures which AI a candidate uses, not how they think, so any funnel built on written answers is filtering noise.

Volume moved just as fast. Greenhouse’s March 2026 hiring-benchmark report (greenhouse.com/recruiting-benchmarks), drawn from more than 640 million applications, found applications per job up 111% and applications per recruiter up 412% between 2022 and 2025. Recruiting is now a verification problem.

We hit both walls. I’m chief executive of TKEG Expat, a corporate-services firm helping companies enter foreign markets; our remote contributor pool spans many countries, and essays were our natural screen. Between 2021 and November 2025 the funnel took in 11,838 applications, 11,381 of them for remote internships. We trained a classifier to grade the essays; it tested at 77.9% precision against our 94% bar, so we demoted it to advisory and moved the real screen to short timed assessments, mostly three-to-seven-minute tasks that are hard to outsource in real time. Roughly 49% of applicants washed out on their own — timeouts, incomplete work — before any human looked, and 119 were approved, about one percent.

For anyone hiring in Europe, the EU AI Act treats recruitment AI as high-risk; deployer duties kick in December 2, 2027 under the Digital Omnibus amendment: human oversight, six months of logs, telling workers you’re using it. That deadline is half the reason the model stayed advisory.

Keith Yunxi Zhu, Chief Executive, TKEG Holdings / TKEG Expat.

The thing that stopped working is the generic job post. For a while I could throw up a listing on the usual hiring sites, say “help wanted, weekend setup crew,” and get a stack of applicants by Friday. That was enough to keep my trucks staffed.

Those posts get ignored now. Everybody is posting the same thing, so my listing just sits in a pile with a hundred others that all sound identical. I’d get clicks and no real people, and my trucks would leave short-handed on a Saturday.

What works now is showing the actual job on video. I film my crew wrestling a 30-foot bounce house into place at the Plano Balloon Festival, or tearing down after a Cowboys event at midnight. I post those clips, and people apply because they already saw what they’re signing up for.

That honesty changed everything. Fewer folks quit after one hot afternoon, because nobody was fooled into thinking this is easy money. My good applicants started coming from people who watched a clip, tagged a friend, and said “we could do that.”

The strange part is my best hires now find me before I even post. They’ve already seen my crew online and they want in. A boring listing used to be my whole plan, and now it’s the one thing I know will get me nowhere.

The biggest shift for us has been what we look for in the first ten hires on a regulated consumer-health team. Two or three years ago I’d have hired a QA or formulation background first, weighted toward GMP fluency and regulatory instinct. That worked when we were manufacturing-forward. It stopped working once the brand side had to carry growth—we ended up with specialists who couldn’t translate a probiotic strain claim into something a real customer reading a label on her phone would trust, or a CX rep could repeat back.

What we hire for now is range plus translation ability—someone who understands structure-function claim limits AND can write to a customer, brief a designer, or sit in a supplier call. In a small DTC team, one role touches marketing, ops, and compliance in the same week.

The cost of the old approach was real: rewrites, slower launches, and me still stuck as the translation layer between lab language and the landing page. We interview for that skill directly now—a short exercise turning a clinical sentence into consumer copy tells us more than a resume does.

When I started the business, hiring meant posting on a big job board and picking from a decent stack of local applicants. That is the strategy that no longer works. The last time I tried it we received over 200 applications, most from people who had plainly never read the advert, a growing share written by AI, and almost none from anyone with an interest in electric vehicles. Sifting them cost more time than it saved.

What changed in our corner of retail is that the work itself moved. Warehouse and admin roles now carry a digital layer, stock systems, listings, support software, so we need people comfortable with data and tools, not just reliable hands. And because we sell to EV drivers, product curiosity matters. A colleague who follows the charging world spots problems and opportunities that a disengaged one never will.

So we recruit where the interested people already are: EV owner groups, our own customer base, local part-timers who want to grow into a role. Our best hire in recent years came from a customer email pointing out an error in one of our buying guides. He cared enough to write in, and he now helps run our content. Small companies win on fit, not reach.

Over my ten years in AI, from scaling platforms at Leboncoin to building our autonomous agent OS at AGO, the engineering profile we hire for has completely flipped. A few years ago, we spent most of our recruiting cycles hunting for pure data scientists who could train custom models from scratch. Today, the foundation models are largely there. What our space actually needs are systems engineers who understand how to orchestrate those models, ground them in real company knowledge bases, and make them execute actions securely without hallucinating.

Because the day-to-day work changed, our old technical screening methods stopped being useful. We used to give candidates a clean, structured dataset and ask them to train a simple model over a weekend to hit a specific accuracy threshold. That strategy just doesn’t reflect the reality of building AI products today. Now, I care much less about someone’s ability to tune a model in a vacuum, and much more about how they design guardrails for when an autonomous support agent encounters an unpredictable user. Instead of a standard data science test, we typically drop candidates into a simulated, broken agent deployment and ask them to look at the system telemetry to figure out exactly why it failed.

The biggest change is that candidates now research us harder than we research them. Before anyone takes a first call they’ve read our reviews, checked Glassdoor, and searched the founder. Whatever your culture actually is, it’s already public.

What stopped working was volume hiring. Financial services sales floors ran for years on one model: bring in a large class, hand everyone a phone and a commission plan, and let attrition sort it out. That assumed a steady supply of people willing to absorb the risk. That supply is gone, and the model always cost more than the spreadsheet showed, because every washout burned a manager’s time and a piece of the team’s morale.

Now we hire fewer people, spend longer training them, and are specific about compensation in the posting itself. New York’s pay transparency rules pushed everyone in that direction, but it was the better approach before it was required. Being vague about comp only ever filtered for candidates who weren’t paying close attention.

Cover letters. We’d read those first before anything in the file. We used to feel good that those got sorted on top.

We knew this for a long time: someone who has been willing to spend forty minutes writing up a unique paragraph about us is inherently filtered out of the 99% of those just blast faxing resume.

Work, in this case, was a proxy for intent. Now that proxy has melted to nonexistence; an email that names product and position and references our blog from last Tuesday now takes the same ninety seconds to write up whether you actually want the job, or if it is just filler when there’s nobody else around. These letters have also gotten good, and that has been part of the problem; when your average application makes pretty much any sense at all, reading cover letters only lets you know who had the word document up. Underlying this is a shift in volume: applications which used to bring several hundred (some few thousand applicants in my time) are sometimes bringing in low double digits, in fact none that effectively filter themselves out of the running early on.

Instead, what we now do is shift skills screening to the start, removing names from all those pages up front to ensure those are done. Only later are cover letters ever read, only at the tail end when the top two to three people can do the job but I need to discover the way they think.

Over the past few years, AI has changed both what we’re hiring for and how we evaluate candidates. Technical execution still matters, but I’m increasingly looking for judgment, problem-solving, communication and the ability to use AI effectively without becoming dependent on its output. Someone who can recognize when an AI-generated solution is wrong or mediocre is becoming more valuable than someone who is simply fast at producing work.

At the same time, the traditional resume, portfolio and take-home assignment have become much less useful as screening tools. It’s now extremely easy to present polished work without being able to tell how much of it actually reflects the candidate’s ability.

Because of that, a short paid, on-the-job trial has become almost a necessity for us. We give someone real work in a controlled setting and evaluate how they communicate, ask questions, use available tools, respond to feedback and ultimately deliver. You learn more from watching someone work for a few days than you can from several rounds of interviews.

The recruiting strategy I’d say no longer works is hiring primarily based on someone’s ability to interview well and present an impressive portfolio. AI has made the gap between appearing capable and actually being capable much harder to identify without seeing someone operate in a real environment.