Belonging Is Not a Feeling. It Is a System Failure You Have Not Measured Yet.

When was the last time you could define belonging operationally?

Not describe it. Not put a slide together about it. Define it the way you would define turnover. With a numerator, a denominator, and a threshold that tells you something is wrong.

If you cannot do that, you are not leading a belonging strategy. You are hosting a belonging conversation. And there is a difference that costs organizations real money and real people.

I have been in a lot of rooms where belonging was on the agenda. Conference rooms. All-hands. Leadership offsites where someone printed the word in a nice font and put it on the wall next to the catering. And in almost every one of those rooms, the word did the emotional work of meaning something without anyone being required to define what it meant. That is not fearless HR leadership. That is borrowed language in place of a plan.

I am an HR thought leader working out of Southeast Wisconsin. I have spent fifteen years inside some genuinely difficult organizational situations, including active CBA negotiations in a 1,200-person manufacturing environment, where belonging was not a concept anyone had time to leave undefined. It was either present or it was not, and you could see where it broke down before a grievance ever got filed. Not because I was especially perceptive. Because belonging, when it collapses, produces signals. And if you are trained to look for those signals, you stop needing the survey.

The Measurement Problem That Nobody Wants to Name

Here is the contrarian take I will defend: belonging is the most consequential unmeasured variable in your people strategy, and HR has allowed it to stay unmeasured because measuring it would make someone accountable for fixing it.

That is not a comfortable sentence. I know that.

But think about what we have done with engagement. We built tools, ran surveys, created indices, reported to the board. Engagement became something you could track. And then, quietly, a lot of organizations realized engagement scores were high while attrition was climbing, and the metric had become performance rather than signal. We measured the feeling instead of the system.

Belonging is at risk of the same fate, only earlier in its lifecycle. We are still in the phase where organizations celebrate having a belonging initiative before anyone has asked what problem it is solving.

Here is what I have found useful. Belonging, operationally, is the degree to which an employee believes their absence would be noticed, their presence is valued beyond their output, and their voice has changed at least one decision in the last six months. That is not a perfect definition. But it gives you three things you can actually ask about, three things you can disaggregate by manager cohort, tenure band, and function, and three places where a system can fail someone long before they tell you they are leaving.

When I diagnosed that 53% of exits at a healthcare organization were happening within the first 90 days, the problem was not onboarding in the technical sense. Nobody was handing people bad paperwork. The problem was that people were arriving and not becoming real to anyone. Their manager had too many direct reports and too little structural support to do the human part of the job. Their peers were cordial but busy. There was no system designed to make a new person feel like their absence would matter. That is a belonging failure. It just took a different shape than the belonging conversation usually takes.

We redesigned onboarding with that framing and measured a 22% attrition reduction over the following year. Not because we added more welcome activities. Because we changed the structural conditions that produce belonging, and then we tracked whether the conditions were being met.

The HR Status Quo This Piece Is Picking a Fight With

HR status quo thinking on belonging sounds like this: we will run a survey, segment the results by demographic, identify gaps, build programs to address the gaps, and report back in twelve months.

That approach has three problems.

First, the survey is a lagging indicator. By the time someone tells you they do not feel like they belong, they have often already decided what to do about it. You are measuring the exhaust, not the engine.

Second, programs built to address belonging gaps are almost always designed to change the employee's experience of the organization rather than the organization's design. More ERGs. More listening sessions. More visibility events. Some of those things matter. But none of them fix a manager who has fourteen direct reports and zero accountability for psychological safety on their team. You cannot program your way out of a structural problem.

Third, and this is the one nobody says out loud, organizations that treat belonging as a cultural initiative rather than an operational discipline are protecting someone. Usually middle management. Because a genuine belonging measurement system would show you exactly where belonging breaks down by manager, by team, by decision point, and that data is uncomfortable to look at and even more uncomfortable to act on.

Here is my forward-looking prediction, and I will be specific enough that you can come back and argue with me about it in three years: the middle manager role as currently configured is structurally incompatible with the belonging outcomes organizations say they want. The span of control is too wide, the coaching expectation is real but the coaching support is not, and the accountability systems are designed to surface performance failures, not belonging failures. HR is not ready for what it will take to fix that. We have been protecting the middle manager from honest metrics for a long time. That is over. The data will eventually make it impossible to look away.

AI Can Help, But Not the Way You Are Thinking

I build custom HR workflow automations using the Claude API. I am not commenting on AI in HR people operations from a distance. I am inside the architecture, thinking about context windows, token limits, and what a model can and cannot reliably do with qualitative signal.

And what I can tell you from that vantage point is this: AI is very good at pattern recognition across large qualitative datasets. Exit interview data. Engagement verbatims. Internal communication tone. Pulse survey open-ends. The kind of material HR collects and then summarizes into a four-slide deck that nobody acts on.

Where AI actually creates leverage in a belonging strategy is in surfacing the signal before it becomes a statistic. Not replacing the HR judgment call. Flagging the patterns early enough that there is still time to make one.

The problem is that most organizations are not structured to receive that signal even if it arrives. You can build a beautiful detection system and still have it feed into a manager conversation that nobody has been trained to have. The technology is not the constraint. The organizational design is.

I say all of this from Southeast Wisconsin, not from a conference keynote. The organizations I have worked with do not always have the resources of a coastal tech company, and that has made me a better diagnostician. When you cannot throw a program at a problem, you have to understand the problem more precisely. That discipline has been useful.

Belonging is not a feeling you can culture your way into. It is a system you design, a measurement you build, and an accountability structure you are willing to defend when the data points at someone inconvenient.

The question is whether you are ready to look at it that way.

Or whether the word on the wall is still doing enough work to keep everyone comfortable.