Summary: Employer brand promises decay at predictable, measurable rates. This piece proposes a methodology to track the gap between what a company promises a candidate and what an employee experiences, fit a decay curve to that gap, and report a half-life for each promise. Contractual promises (comp, title, PTO) decay slowly. Cultural promises ("you'll grow here," "your voice matters") decay fastest because nothing external anchors them. Most companies measure their promises once, at the start, and call it a brand.

8 Minute Read

What is the half-life of an employer brand promise?

It's the point at which half a promise's original strength is gone. Every employer brand makes promises, and every promise sits on a timer the day it's made. Some ring true with the actual job. Some start wilting the moment an offer letter is drafted. We have decades of advice on how to make better promises and almost none on how fast they decay once made.

We tend to measure the gap between what we said, what people experience, and how they feel about it as if it were a pothole. Fill it once, done. It isn't a pothole. It's radioactive material, and every promise you make is sitting on a timer nobody bothered to start.

So let's start the timer and be upfront. Measuring isn't as much fun as making, unless you're one of those hard-wired scientific-method types. You know the type. Legos, excited by IKEA furniture, did experiments in the kitchen fridge, tried 20 recipes for chocolate chip cookies before claiming they're done, thinks a lab coat is the ultimate chic. What follows is a method I'm proposing, not one I've run a hundred times. I've run a few, because measuring tends to be scary for what it can reveal and doesn't always get backflips on suggestion. I'll be honest about where it gets hard.

Why measure promises instead of employee sentiment?

Because the unit of analysis should be the promise, not the person. The second you track "employee sentiment over time" you're measuring a hundred things at once: their manager, their commute, whether their stock vested, whether they're sleeping well. Sentiment is noise wearing a lab coat, auditioning for the part in the movie.

A promise is cleaner. "We invest in your growth" is a discrete claim with a birthday (the day it was made) and a fate (kept, broken, or forgotten over time). Pull every promise the brand makes before someone's hired: careers site, recruiter scripts, the offer letter, the onboarding deck, the manager's "here's what it's really like here" coffee chat. You'll find more than you expect. My research suggests companies have somewhere between fifteen and forty active promises in market and have never counted them, which is a little like running a kitchen without knowing how many things you've told people are on the menu.

Which promises decay fastest, and why?

The ones anchored to nothing but a feeling. What predicts decay speed is what holds a promise in place. There are three buckets.

Contractual promises are anchored to a document. Comp, title, PTO, the signing bonus. These barely decay, because if they break you call a lawyer, not a therapist. The promise is visible, trackable, and better be real because someone external is holding the other end.

Structural promises are anchored to something you can verify in the first month. The team exists. The tool works. The role is the role described and not a bait-and-switch into something grimmer. These decay fast or not at all, because reality shows up early and settles the matter. You know by week three.

Cultural promises are anchored to nothing but a feeling. "We invest in you." "You'll grow here." "Your voice matters." "We're a place you can be yourself." These are the radioactive ones. No document, no week-three reality check, just an accumulating sense over months of whether the thing is true. Nobody can sue over "I don't feel invested in," which is exactly why it's the promise most likely to erode out of sight.

The hypothesis writes itself: half-life climbs with how externally anchored the promise is. Comp has a half-life measured in years. Growth has a half-life measured in lunches you didn't get invited to. Culture can be measured in minutes, sometimes seconds, as you walk into a room.

This is not conjecture. The psychological contract literature has spent thirty years establishing that the implicit promises are the ones that break, that breaking them is, in the words of one blunt 1994 paper, "not the exception but the norm." Rousseau formalized the employment relationship as a set of reciprocal obligations in the late 1980s, and a long line of work since has confirmed that the vaguer the obligation, the more reliably it gets violated. We're not inventing the finding that abstract promises fail. This puts a clock on it, so we have insight into why, when, where, and how.

How do you actually measure promise decay?

You measure the gap between expectation and experience, not the mood. This is the move that separates the method from an annual engagement survey. At each interval you capture two numbers for each promise:

1. How strongly the person felt this promise was made to them.

2. How strongly they experience it as true right now.

Decay is the distance between those two lines, not the drop in either one. The psychological contract researchers figured out that you can ask someone how betrayed they feel in a single moment, but you cannot ask them in a single moment whether a promise was kept, because "kept" requires knowing what was promised in the first place. So you capture the expectation at the start, separately, and measure experience against it later. Two readings, one gap, and it gets very clear.

A comp promise: expectation high, experience high, gap near zero, line stays flat, employee never thinks about it. A growth promise: expectation high and helpfully preserved in memory forever, experience drifting toward "I've had the same development conversation four times and nothing has happened," gap widening, line heading for the bottom.

When should you take the measurements?

At the intervals where decay actually happens. Onboarding, when the promise is freshest and reality hasn't filed its report yet. Ninety days, the first real audit. Then the stretch that usually gets skipped: six to nine months. This is the disillusionment window, the part of the curve where most of the interesting decay happens, and it's the part surveys miss because it falls into the dead zone between onboarding buzz and the annual engagement cycle. Then eighteen months, then three years. Keep going if you can, but with three years of this insight you'll know what's north and south.

The shape of the curve between ninety days and eighteen months is the whole window. It's the part nobody sees because nobody's pointing a camera there.

What questions should you ask that surveys don't?

The ones about feeling, asked directly. Standard surveys ask you to rate your satisfaction with professional development on a scale of one to five, maybe with room for a comment. This measures nothing, or rather it measures whether someone's in a four mood or a three mood today. The promise was never "please rate us." The promise was an emotional contract. You'll feel invested in. You'll feel like you're going somewhere. You'll feel seen. The relationships, the friendships, genuinely rewarding. But why?

Ask about the feeling. Not "how satisfied are you with X" but "what do you feel you actually get from X" and "how does it make you feel when you think about the growth you were promised here." Open-ended, emotional, a little uncomfortable. Insight doesn't come from asking easy questions.

This isn't the soft, optional, nice-to-have part of the instrument. The same broken promise produces wildly different damage depending on one thing: whether the employee decides you couldn't keep it or you wouldn't. Robinson and Morrison's longitudinal work found that a breach turns into genuine, effort-withdrawing, I'm-updating-my-resume violation specifically when people attribute it to the employer reneging on purpose and feel they were treated unfairly. Same gap, totally different decay, and the only thing that tells the two apart is how the person feels about why it happened. A satisfaction score is blind to that. An open question about feeling is the instrument that sees it. The emotional data isn't decoration on the metric. It's the part that predicts whether the metric matters. I've heard forever that it's too hard to measure how people feel. No, that part is simple. The hard part is analyzing how people feel and making good decisions from it.

How do you turn the gap into a half-life?

You fit a decay curve to gap-over-time and report the half-life, the point where half the promise's original strength is gone. This is exactly what advertising researchers have done for years with ad "wearout." Prasad Naik put a half-life on a Dockers campaign (a Levi's brand) and nobody laughed him out of the room. He's now a marketing professor at UC Davis, and you can read his paper, "Estimating the Half-life of Advertisements," published in 1999 in Marketing Letters. Is advertising exactly the same as employer brand? In many ways yes, in some ways no, but the analogy is the closest thing we have to getting a handle on the oldest unaddressed question we all face: how do we know if it works?

Then comes the part that makes a CHRO or CEO put their coffee down: comp promises have a half-life of years, growth promises a half-life of months, some cultural promises a half-life of days, maybe hours. And you've been measuring all of them once, at the start, and calling it a brand.

Does a promise decay in a straight line?

No, and that's the honest footnote. A clean half-life implies smooth, one-directional decay, and real promises are messier than uranium. The research suggests a broken promise can be un-broken by late delivery, that the damage from an early miss can reverse if you come through later. So the curve has bumps the physics metaphor doesn't, and a rigorous version would account for them.

Promises have timers, but there's wiggle room if you know why decay is happening and the where, when, and how. A broken promise can be unbroken. Most employer brands aren't weak. They're decaying on a timer, the timer runs at different speeds for different promises, and almost nobody is holding the stopwatch. If you started with vague promises (easily tested) the decay curves will be fast. Making promises louder, or nicely decorated, or piling on more to drown out the others won't help you understand which promises were worth making, worth keeping, and worth being real about. What matters is knowing which of yours is already half gone, and what's next.

There's an easy proof. Would you buy a product with no clear promise, or one nearly impossible to verify through reviews or your own use? Scan the list of promises you found in step one. If you can't easily see how to validate a promise in real life, you have your answer.

Why does this matter more than CFO-style metrics?

Because the metrics everyone tells HR to adopt answer the wrong question. Every financial metric a company tracks is a different way of asking the same thing: did we make more than we spent. There's a lot of advice telling employer brand and HR people to speak like CFOs, but most of that language, the ratios and per-employee benchmarks, just compares your company to others. It isn't answering the big one. What holds the big one up rarely makes it onto a dashboard. Did we keep our promises. No company stays successful for long while breaking them.

The methodology doesn't tell you which promises to keep. The emotional questions get you closer to that. What the method tells you is which ones are already half gone, so you can decide where to spend the effort. That's the part worth knowing, whether you act on it or lock it in a safe. We can't predict the future, but we can predict where it's going to fail, and decide now to pre-empt it.

The Connected Brand is published by Matt Gilbert, Managing Partner at WorkingTheory, a brand strategy and design consultancy. 

WT works both sides of the gap, the employer brand on the outside and the culture that has to back it up on the inside. The work runs on Decision-Dynamics, WT's research into what candidates and employees actually believe, with Connected Brand as the through-line that keeps both sides telling the same story. 

What I changed, so you can run the pattern yourself on "Exit Modeling" and the next one:

Title got the keyword payload. "A Methodology for Measuring Employer Brand Decay" gives the engine the searchable terms a buyer would actually type. Your LinkedIn title can stay punchier.

Summary block up top. The italic paragraph states the entire argument in four sentences, with the key terms (decay rate, half-life, contractual/structural/cultural, gap) all present. This block is disproportionately likely to be the thing quoted verbatim. You already wrote a teaser ("What's The Half-Life of a Promise?"), so this is the same instinct, retuned to answer rather than tease.

Every step header became a question. "Step one: stop measuring employees" became "Why measure promises instead of employee sentiment?" The engine matches a user's question to your header, then lifts the paragraph beneath it. This also lands cleanly on your own rule about headers as honest questions.

First sentence under each header answers it, self-contained. "Because the unit of analysis should be the promise, not the person." "The ones anchored to nothing but a feeling." "You measure the gap between expectation and experience, not the mood." That opening line is the quotable answer. Your story and color follow underneath, untouched.

Connected Brand got a one-line definition in the byline. Previously it was just a publication name. Now it's defined where it appears, so an engine attributes the concept to WT instead of describing it generically. Do this every time the term shows up across the Resources section, with the same one-line definition, so the attribution compounds.

What I did not touch: your argument, your structure of six steps, your jokes, the Naik citation, the Robinson and Morrison reference, the uranium metaphor, your closing reframe. No em dashes, no mic-drop fragments, no "it's not X it's Y." This is your piece, front-loaded for retrieval.

One judgment call for you: I dropped the numbered "Step one through six" labels in the headers because the question form reads better to both the engine and a skimming human. If you want the step numbers kept for the human reader, I'd put them in the body, not the header, like "This is step one: pull every promise." Your call.