The Loop

I am an employment attorney and I have built people functions three times from zero. Those two facts sit closer together than most people expect, because the calls that end up on a lawyer’s desk almost never start as legal problems. They start as people decisions that were made without a record.

Take the ones my firm exists for. A termination. A reduction in force. An executive exit. In every one, the first question is the same: what was this person hired to do, what were they told, and what did we write down. And in most companies the honest answer is that the job description was written to attract someone, the interview notes are in an applicant tracking system nobody has logged into since, the onboarding plan was a spreadsheet, and the performance conversation happened over lunch.

Four systems, none of which talk to each other, and a decision that has to be defensible years later. That is not a compliance failure. It is an architecture failure, and the compliance failure is downstream of it.

The most recent function I built was at a high-growth health tech company, from fifty people to more than a hundred and fifty, through an IPO. Nothing sharpens your view of documentation like taking a company public. What you learn is that the good outcome and the defensible outcome are the same outcome. If you can say what the role was for, what you asked of the person, and what happened, you usually also made a better decision. When you cannot say it, the decision was probably worse too, and now you also cannot prove otherwise.

This is part of why I work with Nu | Erraa. It keeps one definition of the role from the business need through the first ninety days, and it comes back at the end of that window to check the hire against what was actually asked for. That check is a management tool first. It happens to also be the record you will want if the relationship ends badly.

I would rather build the structures that prevent the hard call than be the person handling it at eleven at night. Most of the time, the structure is just this: decide what the role is for, write it down once, and keep it attached to the person doing it.

None of this is legal advice, and every situation turns on its own facts and its own jurisdiction. But if you are about to make a people decision you would struggle to explain in writing, that discomfort is information. Listen to it.

Kristyn Castello is Founder and Principal of KHR Executive Advisory and a Nu | Erraa Certified Fractional Partner. A Louisiana-licensed employment attorney and SHRM-certified executive, she works with companies nationally from Baton Rouge. Reach her at kristyn@khrexecutive.com or khrexecutive.com. This article is general commentary and is not legal advice.

I have built People and Talent functions from the ground up for two decades. Different industries, different stages, companies of 30 people all the way up to 600+, pre-seed through Series C and past it, across the globe.

Here is the pattern. A company hires well at twenty people because the founder knows every candidate personally and the bar lives in their head. At forty, it still works, barely, because the first People hire is carrying it. At eighty, it stops working, and nobody can point to the day it stopped.

What actually happened is that the reason each role existed stopped traveling beyond the interview stage. At the time, the founder knew it. The recruiter knew it. The hiring manager knew it — they helped write it. Past that point, it went nowhere.

Most companies respond by buying something. An applicant tracking system, because the pipeline is a mess. Then a human resources information system, because records are a mess. Then a performance tool, because nobody wrote down what success looked like.

That is the part I have spent twenty years working around by hand. Manually carrying the reason a role was created, and the scorecard behind who got hired, into an onboarding plan built to set them up to succeed. Then carrying it again into the check-ins, so the manager and the employee had something real to talk about. Even then, a manager in month three who does not know the full range of what this person was hired to be capable of, and me going back through Slack to dig up the notes that spelled it out.

I do not think the answer is more systems. I think the answer is one definition of the role that survives the multiple handoffs, and something that comes back at ninety days to check the hire against it.

That is what caught my attention with Nu | Erraa. It is built to capture the why behind a role before you ever open the req — the success criteria, the reason the role exists — and carry that definition forward into the onboarding plan and then into the performance check-ins themselves. It is clear it is designed by someone who has lived this problem too. It is built around the whole role, not a stage of it.

The other reason it caught my attention is simpler. I do not advise. I am a fractional executive and an operator, and I get into the work, which means I need tools that hold up when someone is actually using them at speed. Most others do not survive that test.

If you are at the point where hiring has stopped feeling like it works and you cannot say why, that is usually the signal. It is rarely the recruiters. It is almost always the seams.

Sholeh Esmaili-Montoya is a Fractional People Executive & Operator | Founder at Startup Talent HR. As a fractional executive, she works with companies from 30 to 100+ people, pre-seed through Series C+, bringing experience built across organizations up to 600+. Reach her at sholeh@startuptalenthr.com or startuptalenthr.com.

I’ve taken teams from 30 to 550. Early projections almost never survive contact with what actually happens. Within months, the assumptions break down: comp frameworks shift, onboarding paths change, and half the headcount you planned for turns out to be unnecessary.

That’s not a failure. That’s just what growth looks like.

What I anchor to instead are three things: telling people the truth even when it’s inconvenient, fixing what breaks fast instead of sitting on it, and never lowering the bar on who we hire. The tools, the docs, the org chart, all of that gets rebuilt more than once, and that’s fine. I’m not attached to any of it.

Moving fast only works if a few things never move. Without that, every normal shift feels like an emergency, because nothing was ever fixed to begin with.

Staying calm follows the same logic. Something is breaking somewhere, basically always. The useful move is never reacting faster, it’s pausing long enough to tell a real fire from a Tuesday. If I panic, the room panics. If I stay level, they get to stay level too.

The place I see this most is hiring, which is why I certified on Nu|Erraa. A role gets written in Q1. By Q2 the company has moved and the job hasn’t. Nobody goes back and updates it, so someone gets hired into a role that’s already changed, and the 90-day check-in that should catch it never happens. Without a clear bar for what “good” looks like, nobody can tell the difference between someone underperforming and someone doing a job of an outdated spec.

Scaling fast means being precise about the few things that hold and letting everything else flex. Get that boundary right, and growth stops being chaos. That’s the work.

Jen Gaddini is a Nu | Erraa Certified Fractional Partner and is the founder of Marrow. Reach her at jen@buildwithmarrow.com or buildwithmarrow.com.

I have spent more than fifteen years in recruiting, most of it inside large tech companies and, for the last few years, on my own, working directly with founders. The work looks very different at those two ends. The problem I want to write about looks exactly the same at both.

A few weeks ago I sat down with a hiring manager after we had interviewed fifteen people for a customer experience role. Good conversations, smart people, and not one clear yes. Everybody in the room knew what the next move was supposed to be: open LinkedIn, go find number sixteen, keep the loop full.

I did not do that, because I have watched it play out too many times. You end up at thirty interviews, a tired panel, and an offer to whoever happened to be in the process when everyone ran out of patience. That person is rarely a bad hire on paper. They are a hire nobody can fully explain, and six months later nobody can say whether it worked.

When a search stalls that far in, I look at the role before I look at the pipeline. Usually somebody wrote a job description to attract applicants, somebody else posted it, and the hiring team never sat down and agreed on what the person is supposed to deliver. Each interviewer is scoring against a slightly different picture in their own head, so the panel keeps splitting and nobody gets to yes.

I learned this early, on niche hardware searches where there might be a few hundred qualified people anywhere in the world. You cannot source your way out of a vague brief when the pool is that small. If I walked out of an intake meeting without knowing what good looked like, I paid for it for months. Startups end up in the same place from the other direction. The applications pour in, the pool looks enormous, and the definition of the job is thinner than anyone wants to admit.

So instead of going back to sourcing, I call what I think of as a recalibration. It takes thirty minutes. I bring every scorecard from the search, and I ask three questions.

The first is who came closest, and what they were missing. I want that gap in one sentence. If the hiring manager can say it, we have something real to search for. If they cannot, we never defined the role. We defined a feeling about the role, and fifteen people have now failed to match it.

The second is what this person has done by day ninety. I am asking for outcomes a manager could check, not traits. Every one of those fifteen candidates was a strong communicator and a self-starter, so those words were never going to separate anyone. What separates people is whether they can own a queue or a process on their own by a certain week, or fix the specific thing the business opened the seat to fix.

The third is which requirement we are holding onto because of a name. A specific school, or a big-tech logo on the résumé. I cut those first, every time. Some of the sharpest people I have ever hired never went near either one.

The meeting ends with one of three calls. We change the job, we change who we are looking for, or we keep going as we are. In my experience the third is the least common, and it is the one everybody walks in assuming.

This matters more at a thirty-person company than almost anywhere else. At a large company, a stalled search burns recruiter hours. At a startup it burns founder hours, because until there is a real layer of managers, the founders usually sit in every loop. Fifteen interviews for one role can easily mean fifteen hours of a chief executive's month, plus the prep and the debriefs, before anyone counts what the empty seat is costing.

There is a second cost, and it shows up later. SHRM's 2025 recruiting benchmarking found that only about one in five organizations measure quality of hire at all. I do not think that is because people do not care. It is because nobody wrote down what good looked like at the start, so there is nothing to measure against at ninety days. The fuzzy definition that stalled the search is the same one the manager ends up squinting at in the new hire's first review.

This is part of why I work with Nu | Erraa. It starts from the business need rather than the job post. You tell Amy what the position has to get done, and that one definition is what candidates are scored against, what the onboarding plan is built from, and what the hire is checked against at day ninety. The software writes the position and scores the candidates. The call in that recalibration meeting, whether to change the job or change the search, is still a person's call, and that is where partners like me come in.

If you are a founder hiring without a recruiter, here is what I would do on your next open role. Before it goes live, write down two things in plain words: why the business needs this person now, and what they own by day ninety. Share that page with everyone on the interview panel so they are all scoring against the same thing.

Then pick your stopping point ahead of time. For most roles I would say ten to fifteen interviews. If you hit it without a yes, do not source harder. Get the panel in a room, put the page on the table and ask the three questions. Most of the time you will end up rewriting part of the role, and the next shortlist will look different because of it.

Stefanie Fackrell is the founder of dysruptiveHR and a Nu | Erraa Certified Fractional Partner. She partners directly with founders to turn business goals into practical people strategies, from defining the first hires to shaping teams and coaching leaders, drawing on more than fifteen years across startups and high-growth tech. Connect with her on LinkedIn.

Sources: SHRM Recruiting Executives Benchmarking, 2025.

I've spent more than twenty years in HR, much of it building people functions from scratch in founder-led companies. Before that, I was a professional theater performer. The first career explains more about how I do the second than you might guess.

In theater, everybody gets notes. After rehearsal the director walks through what worked and what didn't, in front of the whole cast, and nobody goes home wounded. It works because everyone agreed in advance what the show is. There's a script and there's a vision. The note is about the distance between those and what happened on stage, so it never has to be about you.

Most managers in early-stage companies are giving notes with no script.

Here's how it usually goes. The founder knows exactly why a role exists and can describe it for ten minutes without stopping. Some version of that makes it into a job post, written to attract people. Someone gets hired. Then the hire is handed to a manager, often in their first year of managing, who sat in one interview at most and who inherits the person without inheriting the definition.

That manager is rarely set up for this. Gallup's State of the Global Workplace 2026 found that 56% of managers worldwide have never had any management training. So the person now responsible for a new hire's first ninety days is often learning to manage on that same hire, with nothing written down about the job they're managing against.

By week six, something is a little off. The manager can feel it but can't name it, because there's nothing written down to hold it up against. So they do what decent people do in the absence of a standard: they stay nice. They say "great start" in the one-on-one. They fix the work themselves at night. They wait to see if it improves. Meanwhile the new hire usually senses something is wrong before anyone says a word, and silence from a manager reads one of two ways: everything is fine, or a verdict is being prepared. Neither helps them do better work on Monday.

I call this the Nice Tax. It is what a company pays when comfort wins over candor, and it compounds. The employee loses the weeks when a course correction would have been cheap. The manager starts burning out covering the gap. The founder finds out in month five, when the conversation is much harder and the employee is, fairly, blindsided.

The bill for that blindside is real. The Work Institute's 2025 Retention Report, drawn from more than 14,000 exit interviews, puts up to 40% of all turnover in the first year, and the cost of replacing someone at about a third of their base pay. In my experience, a fair number of those exits started as a week-six conversation nobody had.

When I get called in, the request is often some version of "our managers need to get better at hard conversations." Sometimes that's true, and I spend a lot of my time coaching exactly that. But skill is rarely the whole story. Most managers can deliver a clear message when they have one. What they lack is an agreed standard. And without one, feedback is just one person's opinion of another. That is why it feels personal, and why people avoid it.

So the first thing I build in a young company is rarely a performance process. I've seen too many of those: elaborate, engineered, and ultimately ineffective. I start smaller. For each role, the founder's reason for creating it, in plain words. What the person owns by day ninety. All on a single page. Then a date on the calendar when the manager and the employee sit down with that page between them.

The page is short on purpose. A manager will reread one page before a one-on-one. Nobody rereads a full competency framework after the offer letter goes out, so in practice the framework stops existing on day one.

High standards and care for people get treated as a trade-off. In my experience, each makes the other possible. Telling someone the truth in week six, against a standard they saw on day one, is about the kindest thing a manager can do.

This is why I work with Nu | Erraa. It starts where I start, with the business need, before there's a req. That one definition is what candidates are scored against, what the onboarding plan is built from, and what the hire is checked against at ninety days. In my terms, it hands the manager the script. That ninety-day check is rarer than it should be: SHRM's 2025 Recruiting Executives Benchmarking found only 20% of organizations measure quality of hire at all, down from 27% in 2022. The software can't have the conversation, and I wouldn't want it to. That still takes a person with some nerve and some practice, which is the part I coach. But it removes the most common reason the conversation never happens.

If you manage someone hired in the last ninety days, try this before your next one-on-one, not after the next missed deadline. Write down, in three sentences, what they were hired to deliver. Ask them to do the same without seeing yours. Then compare, the way a cast and a director compare notes after rehearsal: same script on the table, and the note is about the distance, not the person. The gap between the two pages is the conversation you've been putting off, and it will never be cheaper than it is today.

David Gaspin is the founder of IncubateHR and a Nu | Erraa Certified Fractional Partner. An ICF-certified executive coach with more than twenty years as a People leader, he helps founder-led companies turn the founder's vision into a working people function and develops the leaders and managers who have to carry it. Reach him at david@incubate-hr.com or on LinkedIn.

Sources: Gallup, State of the Global Workplace, 2026. Work Institute, 2025 Retention Report. SHRM Recruiting Executives Benchmarking, 2025.

Remote bought Atlas. Payoneer bought Boundless. Docebo took 365Talents. Phenom bought two AI companies at once. Read those as news and it is a busy year in HR tech. Read them as a pattern and they are all the same sentence: our product does one part of the job, and we have run out of ways to pretend otherwise.

That is not a criticism of any of those companies. It is what happens when a category is built stage by stage. Someone solves the pipeline and calls it an applicant tracking system. Someone solves the employee record and calls it an HRIS. Someone solves the review cycle and calls it performance. Each one is genuinely good at the slice it owns. None of them owns the thing that runs through all three, which is the role.

So the market does the only thing available to it. It buys the missing slice and stitches it on.

The trouble is what stitching leaves behind. When two products become one product on the pricing page, they are usually still two products in the data. The sourcing tool has its idea of a candidate, the onboarding tool has its idea of an employee, and somewhere between them is a handoff where the reason you opened the role in the first place quietly stops travelling. The logo is unified. The seam is not.

There is a different way to draw the line, and it starts by asking what the unit actually is. Not the candidate, not the employee record, not the review cycle. The role. What the business needed done, who is doing it, and whether it is getting done. Define the system around that and the handoffs stop existing, because there is nothing to hand off between.

Sources: UNLEASH coverage of 2026 HR technology acquisitions. This piece first appeared in an earlier form in The Loop, Issue 01.

SHRM’s 2026 Talent Trends research put two findings next to each other, and the combination is more interesting than either one alone. Nearly 70 percent of HR professionals report they cannot fill their full-time roles. And 80 percent say the hardest quality to hire for is judgement.

Taken separately, those read as complaints about a tight market. Taken together they are a strategy problem, because they say that the thing you most need cannot reliably be bought from outside.

Most talent spending assumes the opposite. The budget goes to the top of the funnel: more sourcing, more screening, better job posts, a faster pipeline. All of that is aimed at finding the right person. None of it helps when the right person, in the sense that matters, does not exist on the open market in the numbers you need.

Judgement is the clearest example. It is not a credential, it does not show up in a keyword search, and it is very hard to assess in four interviews. It is built in context, by someone who understands the business, over a period measured in months rather than weeks. Which means it is largely produced after the hire, not selected before it.

If that is true, then the part of the process that matters most is the part almost nobody is managing. SHRM’s benchmarking work puts the share of organizations measuring quality of hire at about one in five. Four out of five companies never establish whether the person they hired did the job they were hired to do.

Close that loop and two things change at once. You find out whether a hire worked, which is worth knowing on its own. And you get the only feedback that improves the next role definition, which is what turns hiring from a series of independent bets into something that compounds.

Sources: SHRM 2026 Talent Trends Report and SHRM Recruiting Executives Benchmarking, 2025. This piece first appeared in an earlier form in The Loop, Issue 01.

Every hiring system collects evidence. Interview feedback, scorecards, assessment notes, offer reasons, onboarding plans, performance conversations. The problem is not that companies lack data. The problem is that the data stops traveling.

The scorecard lives in one place. The onboarding plan lives somewhere else. The performance conversation happens months later, with no clean line back to what the person was hired to do. Everyone means well, but the system forgets.

Nu | Erraa was built around the opposite idea: one definition of the role, carried from the business need through hiring, onboarding, ramp and review. The point is not to collect more talent data. It is to keep the data connected long enough to learn from it.

That is the loop. Define what the role has to deliver. Hire against it. Onboard against it. Check whether it worked. Then use that answer to make the next role sharper.

Most systems collect talent data. Until now, we never connected it.