The Engineer Hiring Mistakes That Cost You the Best Candidates

How to hire software engineers without losing your best candidates comes down to two decisions, and almost every company gets at least one of them wrong. The first is who you screen for: pedigree or potential. The second is how many steps you make a candidate go through before you decide. I have run technical searches for nearly sixteen years and placed close to 1,500 engineers, and when a search stalls, it is rarely because the talent pool is thin. It is because the client’s own process is filtering out the people they most want to hire.

This is not a generic “improve your hiring” post. It is the two mistakes I see most often, in my own words, with the data that backs up why they cost you.

Mistake one: hiring for pedigree instead of potential

If I had to name the single most common hiring mistake I see, it is this: companies hire for pedigree, not potential. They over-index on the name on a resume, the school, the “brand” companies someone has worked at, and they treat that as a proxy for whether the person can actually do the job.

The cost of that is not abstract. When you over-emphasize pedigree in a hiring process, you miss a large swath of perfectly capable engineers who have a very high slope, meaning they are early in a trajectory that is clearly pointing up, and who could grow with your company for years. Slope matters more than current altitude in almost every early-stage or growth-stage engineering hire I have made. The engineer three years into their career who has taught themselves distributed systems on nights and weekends, shipped side projects that actually have users, and asks sharper questions in an interview than the person with the “better” resume, that engineer is frequently the better bet. Pedigree screening filters them out before anyone talks to them.

This is not a call to ignore experience. It is a call to test for the thing you actually need, which is whether someone can do the work and grow into more of it, rather than whether their resume pattern-matches to what a “good hire” is supposed to look like on paper. A pedigree filter is a shortcut, and shortcuts in hiring tend to be expensive in ways that only show up months later.

Why interview loops balloon past four steps

The second mistake is process bloat, and it is closely related to the first. My advice to every client, without exception, is to simplify your interview process. Nothing needs to be more than four steps, maximum.

Here is the part most hiring teams do not want to hear: if you need more than four steps of input from your hiring team to make a decision, the problem is not the candidate. Your internal communication is not working, and there is a lack of trust between the people involved in the decision. A recruiter screen, then a hiring manager screen, then a panel of three, then a “culture fit” round, then an executive sign-off is not rigor. It is five or six different people who do not trust each other’s judgment enough to let a single strong signal carry weight.

The data on this lines up with what I see in the field. According to a Talent Board survey covered by SHRM, two of the top three reasons candidates around the world withdraw from a hiring process are that their time was disrespected during the process and that the process simply took too long. Candidates are not confused about what a bloated loop signals. They read it correctly, as an organization that cannot make a decision.

flowchart TD
  A[Step 1: Intro call<br/>~30 min, motivation + fit] --> B[Step 2: Technical conversation<br/>~60 min, real problem]
  B --> C[Step 3: Working session<br/>with a future teammate]
  C --> D[Step 4: Final conversation<br/>with the decision-maker]
  D --> E[Decision within 48 hours]
  classDef base fill:#ECFDF5,stroke:#10B981,stroke-width:1.5px,color:#065F46
  classDef final fill:#D1FAE5,stroke:#059669,stroke-width:1.5px,color:#0A2E22
  class A,B,C base
  class D,E final

What the data says about interview length and drop-off

I am not the only one saying this. It shows up in every serious dataset on hiring speed.

SHRM’s 2025 Recruiting Benchmarking Report puts the average US time to fill a role at roughly 44 days. That is the average, which means plenty of companies are well past it. Separately, SHRM’s own guidance on time-to-fill metrics recommends employers hold their time to fill at or under roughly 33 days, warning that a “let’s see who else we can find” mentality without a firm timeline is exactly what causes strong candidates to lose interest and disengage. The strong candidate you are evaluating is being evaluated by two or three other companies at the same time, and the slowest company in that comparison set usually loses.

Scheduling friction alone is enough to sink a search. Cronofy’s 2024 Candidate Expectations Report, based on responses from candidates across seven countries, found that 42 percent of candidates dropped out of a hiring process specifically because scheduling interviews took too long. That is not a rejection of your company or your role. It is candidates walking away from friction, and every additional step in your loop is another point where scheduling has to happen again.

Communication gaps compound the problem. Monster’s 2024 WorkWatch Report found that 47 percent of workers cited poor communication, meaning no update on their application status or unanswered messages, as a reason they would withdraw from a recruiting process. An eight-step loop has eight separate opportunities for someone on your side to go quiet between rounds. A four-step loop has four.

None of this is new information to the people who study hiring at scale. Google famously ran the numbers on its own interview loops years ago and found that four interviews were enough to predict, with 86 percent confidence, whether the company should make an offer. Every interviewer added after the fourth increased that predictive confidence by roughly one percent, according to the internal analysis Laszlo Bock describes in his account of Google’s people analytics work, covered by Knowledge at Wharton. Google had the data and the resources to run twelve-interview loops if that produced better hires. It did not. It produced diminishing returns and slower time to hire.

What you can actually learn in 5 minutes, versus 30

Here is the piece of advice I repeat most often, because it is the thing that lets a hiring team trust a shorter process instead of stacking on more steps out of anxiety: you know within the first five minutes of an interview whether you would want to work with someone. And you know within the first 30 minutes whether they have the problem-solving and technical skills you actually need.

That is not a claim that interviews should be shorter than they are. It is a claim about where the signal actually lives. The first five minutes tell you about communication style, how someone thinks out loud, whether they ask good questions back, and whether a conversation with them feels like it would be good for your team day to day. The next 25 minutes, spent on a real problem rather than a trivia-style algorithm question, tell you whether the technical substance is there. Everything past that first 30 to 60 minutes in a single conversation is usually confirmation, not new information. Stacking five more rounds on top of a strong first conversation rarely changes the answer. It just delays it, and delay is what costs you the hire.

Bloated process versus lean process, side by side

The comparison is easiest to see laid out directly.

If your loop looks like the left column, the fix is rarely “add one more filter to be sure.” It is almost always “figure out why four people on your team do not trust the other three enough to move on a strong signal.”

Where this connects to the rest of your hiring process

Interview design does not exist in isolation. A lean, trustworthy loop only works if the role was scoped honestly in the first place, which is the subject I go deeper on in how to write a job description engineers actually read. And if you are already losing strong candidates before they even apply, the interview process is not your first problem, the posting is, which I cover in why strong engineers do not apply to your roles. The pedigree-over-potential mistake shows up especially hard when a company is trying to make an early, high-leverage hire, which is exactly the situation I describe in what is a founding engineer. A founding engineer search is one of the worst places to filter on pedigree, because the traits that predict success in that role rarely show up on a traditional resume.

This is also where I will make the one fair mention of my own work in this post, because it is directly relevant. I built Kruger Recruiting as a boutique, solo operation specifically so I could run searches this way. My average time to close a search is about three weeks, and the right hire typically shows up within the first three candidates I submit, not the fifteenth. That is not an accident. It is what happens when you combine a genuinely narrow, well-understood search with a client-side process that does not stall on its own overthinking. I back every placement with a 90-day guarantee and my clients see roughly a 2 percent attrition rate within the first year, and a clean interview process, on both sides, is a meaningful part of why.

Why hires still fail even with a good interview process

A lean process is necessary, but it is not sufficient on its own, and I want to be honest about that rather than oversell the fix. When a new hire fails in the first 90 days, there is never one direct answer. It is a kaleidoscope of factors. The most obvious is a skills mismatch, which a good interview process should catch. But I have also seen lifestyle misalignment sink hires that were technically sound on paper, meaning the company needed more from the person than they could realistically give, or the reverse. What it usually boils down to is expectation versus reality not matching on either side.

I have also seen companies build interview processes that reward candidates for giving the answer the company wants to hear rather than the honest one. A shorter process does not fix that by itself. What fixes it is designing your four steps to promote genuine connection and honest signal, rather than performance. Ask what the day-to-day actually looks like. Let the candidate ask hard questions and answer them straight. The best companies I work with also understand that no hire, however well vetted, is a true plug-and-play. Everyone needs some grace period to learn a company’s culture and way of working before they can be evaluated fairly, and building that grace period in deliberately is part of what separates the companies with low first-year attrition from the ones re-hiring the same seat every year.

Ready to fix your engineer hiring process?

If your interview loop has grown past four steps, or you keep passing on strong candidates because they do not match a pedigree checklist, that is usually fixable in one conversation about what you are actually screening for. If you want a second set of eyes on a search that is not moving the way it should, book a free strategy call and we can look at where your process is actually losing people.

Sources and further reading

• SHRM, 2025 Recruiting Benchmarking Report
• SHRM, “Staffing Metrics: ‘Time to Fill’ Can Kill Prospects of Landing Top Talent”
• Cronofy, Candidate Expectations Report 2024
• Monster, 2024 WorkWatch Report
• SHRM, “Why Your Candidates Are Dropping Out” (Talent Board research)
• Knowledge at Wharton, “Open Sourcing Google’s HR Secrets”

Frequently Asked Questions

How many interview rounds should a software engineering hiring process have?

Four steps is the maximum I recommend to clients: an intro conversation, a technical conversation, a working session with a future teammate, and a final conversation with the actual decision-maker. If your team genuinely needs more than four checkpoints to decide, the issue is usually internal trust and communication, not a lack of information about the candidate.

What does “hire for potential, not pedigree” actually mean in practice?

It means testing for whether someone can do the work and grow into more of it, rather than filtering primarily on the school they attended or the brand names on their resume. Engineers with a high growth trajectory but a less impressive resume are frequently underrated, and a pedigree-first filter screens them out before anyone even talks to them.

How long should it take to hire a software engineer?

Benchmarks vary, but SHRM’s 2025 data puts the US average time to fill at around 44 days, while SHRM’s own guidance recommends holding time to fill closer to 33 days to avoid losing candidates to disengagement. In my own searches, the average time to close is about three weeks, and the eventual hire is typically found within the first three candidates submitted, which is a strong argument that speed and quality are not in tension.

Why do strong engineering candidates drop out of long interview processes?

The two most cited reasons in candidate surveys are that the process took too long and that their time was disrespected along the way. Cronofy’s 2024 research found 42 percent of candidates dropped out specifically because scheduling interviews took too long, and Monster’s 2024 WorkWatch Report found 47 percent cited poor communication, such as unanswered messages or no status update, as a reason to withdraw.

What can you really learn about a candidate in the first five minutes of an interview?

You can tell fairly reliably whether you would want to work with the person day to day: how they communicate, how they think out loud, and whether the conversation itself feels productive. The next 25 to 30 minutes, ideally spent on a real problem rather than a trivia-style question, is what tells you whether the technical substance holds up.

Does a lean interview process mean skipping technical assessment?

No. A four-step process still includes a real technical conversation and, ideally, a working session with a future teammate. What it cuts is redundant rounds that mostly repeat the same evaluation through a different person’s eyes. Google’s internal research, described in Laszlo Bock’s account of the company’s hiring data, found that four interviews predicted the hiring decision with 86 percent confidence and that each additional interviewer added only about one percent more predictive value.