Latest / Hiring & Firing with Fexingo: Recruitment, Performance, and Workforce Decisions / How a Trial Project Beat the Traditional Interview
Transcript
- Lucas: So there's this software firm in Austin, about two hundred employees, and last year they decided to kill the traditional interview for their engineering roles. No more whiteboard coding, no more 'tell me about a time you showed leadership.' Instead, every shortlisted candidate gets a paid trial project that takes about four hours. Luna: Four hours is a big ask. Did people actually do it? Lucas: That's the thing — most did. And the ones who didn't mostly self-selected out, which the company actually saw as a feature. The trial was designed to mirror a real task from their own backlog: build a small feature, fix a bug, write a test. They paid everyone who completed it their standard contractor rate. Luna: And what happened to their bad hire rate? Lucas: It dropped by a third within a year. They were tracking it pretty carefully — hires that didn't make it through the first six months. Before the change, it was around one in four. After, it was closer to one in six. And their time to hire actually went down too, because they stopped wasting weeks on second and third interviews. Luna: So the trial was a better signal. But what about the people who are great in an interview but lousy at the actual job? I feel like that's the whole point. Lucas: Exactly. The research on structured work samples is pretty consistent — they're one of the best predictors of job performance, way better than unstructured interviews. And yet most companies still default to the hour-long chat, because it's what we know. This firm just decided to invert it: the conversation became the sanity check, and the work became the test. Luna: But how do you design a trial that's actually fair? I mean, someone with more experience might finish in two hours, someone else in six. And you're asking them to spend unpaid — well, paid in this case — but still, it's time. Lucas: That's the tension. The company paid them, which is key. They set a fixed time-box of four hours, but they didn't time it strictly — they just said 'here's the task, here's the expected scope, send it back when you're happy.' They evaluated the output, not the speed. And they used a scoring rubric that three different engineers applied independently, so it wasn't just one person's gut. Luna: So the rubric was the real tool. What did it look like? Lucas: They had four criteria: correctness of the solution, code quality, how well the candidate communicated their approach, and how they handled a deliberately ambiguous requirement. That last one was the sneaky part — the brief was written with a hole in it, and the candidates who asked a clarifying question before diving in scored higher on communication. That's a signal you almost never get from an interview. Luna: Asking a question before starting — that's a nice proxy for how someone works in the real world. Most of us sit in front of a vague ticket and immediately start guessing. Lucas: Right. And that's the beauty of a work sample — it's not about the answer, it's about the process. They could see how someone reacted to ambiguity, how they asked for help, how they structured their time. None of that comes through in a thirty-minute conversation where everyone's on their best behavior. Luna: But I've also heard the flip side — that trial projects can be a way to get free labor, or that they favor people with lots of free time. How did this company handle that? Lucas: They paid everyone, which eliminated the free-labour complaint. And they made the task genuinely relevant to the role, not a generic test. But there's a real concern about access — a four-hour block is a lot harder to carve out if you're a single parent or you're working two jobs. The company actually offered a flexible window: candidates could submit within a week, and they could pause and come back. They even had one candidate who did it in two sessions because their childcare fell through. Luna: Flexibility matters. But I'm curious about the bias angle — if the task is designed by the current team, doesn't it just reflect their own way of doing things? And could that shut out people who think differently? Lucas: That's a sharp point. The team built the task around their own codebase and their own conventions, so in a way it was testing for cultural fit as much as raw ability. But they were explicit about that: they told candidates the task mirrored their actual work, so if you didn't like the style, you were free to decline. Some people did, and that was its own filter. Luna: So the trial became a two-way street — the candidate got a taste of the job too. That's actually a big selling point for the people who went through it. Lucas: Yeah, the feedback from candidates was surprisingly positive. They said it was more stressful than a typical interview, but also more honest. Instead of rehearsing answers, they got to show what they could do. And the company found that candidates who completed the trial were more likely to accept an offer, because they already had a clear sense of what they were signing up for. Their offer acceptance rate went up by about fifteen percent. Luna: That's a big deal. But I have to ask — what about the people who are brilliant in an interview but freeze under the pressure of a timed task? Could you be screening out good people? Lucas: That's a real risk. They had a few candidates who did poorly on the trial, but then aced a follow-up interview, and they ended up hiring one of them. So they didn't treat the trial as a hard gate — it was one signal among several. But the data they collected over the year suggested that the trial was a stronger predictor than the interview. The correlation with performance reviews was much higher. Luna: So the interview became almost like a tie-breaker. That's a big shift from the usual 'we'll do three rounds and then decide.' Lucas: Right, and it's part of a larger trend. You've got companies like Automattic and Basecamp that have been doing paid trial projects for years, and they swear by them. But the interesting thing here is that this firm wasn't a tech-startup outlier. They're a mid-sized B2B company that just decided to try it after a string of bad hires in one specific team. And it worked well enough that they've now extended the model to their customer-support roles. Luna: So what does a trial project look like for a support role? You can't exactly ask someone to take a live ticket. Lucas: They created a simulated inbox with a batch of fake tickets — some straightforward, some vague, one that was a real customer complaint. The candidate had to respond to each one. It took about two hours, and they were evaluated on tone, accuracy, and how they escalated the tricky ones. The hiring manager told me it revealed more about a person's judgment in one sitting than any fifteen-minute role-play ever did. Luna: That makes total sense. And it also gives the candidate a preview of the emotional labor of the job — you're dealing with frustrated people all day. That's something you can't fake your way through. Lucas: Exactly. And that's the real lesson: the trial project isn't just a better filter for the company, it's a better experience for the candidate. They get to see what the job actually is before they say yes. That's a huge deal in a market where job descriptions are often, let's be honest, aspirational fiction. Luna: Okay, before we keep going — I have to say, this conversation is exactly why I love this show. If you're listening and you've got a hiring process that's stuck in the last decade, this is the kind of thing that might genuinely change how you think about it. And you know, if today's episode was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. It's the smallest ask, and it keeps us ad-free. Lucas: Yeah, she's right. We don't have sponsors, we have listeners. And if you've ever learned one thing from this show that saved you a headache — a coffee is the price. But no pressure, seriously. You can just tell a friend about the podcast. Luna: Alright, back to the trials. So what's the downside? There has to be a catch, because if this was universally better, everyone would be doing it. Lucas: The catch is scale. If you're hiring for a role that gets hundreds of applicants, you can't send everyone a trial project. The evaluation time would kill you. This company only used it for shortlisted candidates — after a phone screen and a quick skills check. And they still had to spend about two hours per candidate on evaluation, which is more than a typical interview. For a high-volume role, that's not sustainable. Luna: So it's a tool for the final stage, not a replacement for the entire funnel. That makes sense. Lucas: Right. And there's also a legal and ethical angle. You're asking someone to work before they're hired, even if it's paid. There are rules around that — you have to be careful not to ask for something that's the same as the real job, or you might be creating an employer relationship. The company's lawyer was involved from day one, which is why they paid everyone and kept the tasks clearly scoped. Luna: So it's not just a hiring trick, it's a legal minefield if you're not careful. But the payoff seems real — a third fewer bad hires and faster hiring overall. That's a compelling trade-off. Lucas: It is. And I think the deeper takeaway is that the trial project forces you to define what success looks like before you start looking for it. You can't design a work sample unless you know what skills actually matter for the job. Most companies never do that — they just write a job description and hope for the best. Luna: That's a really good point. So if you're a hiring manager, the first step isn't to copy the trial project — it's to sit down and figure out what the job actually requires. The trial is just the test. Lucas: Exactly. And the good news is you don't have to go all-in overnight. Start with one role, one team. Design a trial that takes two hours, pay people for it, and see what you learn. If it works, expand. If it doesn't, you've only changed one part of your process. Luna: And if you've been through a trial project as a candidate, you're probably already sold — or you have horror stories. I'd love to hear from listeners on that. You can find us on the Fexingo site or on social. We read everything. Lucas: Alright, that's a solid place to leave it. Until next time, keep thinking about what you're really testing for. The résumé gets you in the door, but the work is what tells the truth.