Your Company Knows Everything Except Its People

Revenue gets a dashboard, a forecast, a weekly review, a board pack, and a CFO. Risk gets a dashboard, a limit framework, daily alerts, stress tests, and a CRO. Security gets real-time monitoring, incident response, and a CISO.

People get an annual survey. Sometimes they get a quarterly performance review, usually written after the important events have already happened.

This is an odd arrangement. When a bank loses one basis point of margin, someone is asked to explain it before lunch. When a bank loses one of its most valuable people, the organisation often waits for the resignation letter before it begins investigating.

The financial cost can be significant. SHRM’s turnover-cost tools explicitly distinguish direct hiring costs from the broader costs of vacancy, replacement, onboarding, and lost productivity. Estimates vary substantially by role, which is exactly why companies should model their own exposure rather than repeat a single universal multiplier.

For a senior producer earning £400,000, a departure can easily become a seven-figure business problem once recruitment, ramp time, lost relationships, delayed work, and institutional knowledge are included.

That £1.2 million figure is not a universal accounting rule. It is a plausible scenario. The important point is that most companies cannot tell you their own number with confidence.

This is the largest observability gap in modern business. It is also one of the few that technology could plausibly close, if the technology is designed to support people rather than monitor them. Seven observations explain why.

Revenue per employee is useful because it reveals something about operating leverage. It is not useful because it tells you why the leverage exists.

A company may produce more revenue per employee because it has:

• Higher prices.
• A capital-light business model.
• Stronger intellectual property.
• A temporary commodity advantage.
• Fewer employees classified inside the reporting entity.
• Better systems.
• Better people.
• Or some combination of all six.

That distinction matters. It is tempting to say that the entire spread in revenue per employee is a function of individual performance. It is not. The metric is shaped by industry, accounting choices, acquisitions, capital intensity, outsourcing, and business model.

How much value is the organisation getting from the people whose decisions matter most?

In a professional-services firm, investment bank, consultancy, law firm, or technology company, a small number of senior people often control a disproportionate amount of revenue, client trust, institutional knowledge, and future opportunity.

Averages conceal this. A company can have healthy aggregate productivity while quietly exhausting the people responsible for its most valuable relationships. It can also have excellent engagement scores while under-utilising people who are already preparing to leave.

The CFO needs the aggregate number. The manager needs to know which person is changing.

The modern Chief People Officer does not lack software. They have an HRIS, an applicant-tracking system, a performance-review platform, a learning system, a pulse-survey tool, an absence system, a compensation tool, and increasingly a people-analytics platform.

These systems answer important administrative questions:

• Who works here?
• Who has been hired?
• Who has been promoted?
• Who has completed the training?
• Who is absent?
• Who received which rating?
• Who answered the survey?

They are much less capable of answering:

• Who is approaching unsustainable load?
• Who is disengaging despite still performing?
• Who is under-utilised and at risk of leaving?
• Which team is losing recovery time to meetings?
• Which manager is generating avoidable capacity loss?
• Which intervention is likely to help this person now?

The problem is not that engagement surveys are worthless. They provide useful organisational information. The problem is timing and aggregation.

A survey can tell you how a population felt when it answered the survey. It usually cannot tell you what changed last week, why it changed, or what a manager should do on Monday morning.

That is not proof that a particular dashboard failed. It is evidence that the existing management system is not producing enough usable intervention.

The signal is not missing. It is fragmented.

A senior producer already generates a rich stream of information:

No single source explains a person. A low recovery score could mean illness, travel, a new training block, a faulty sensor, or a perfectly normal variation. A crowded calendar could mean overload. It could also mean a week of unusually valuable client meetings. A slower response time could indicate burnout. It could also indicate focus.

The signal becomes meaningful only when it is combined with context and interpreted cautiously. That is the difference between data collection and coaching.

The ethical version of this system would not infer everything it can infer. It would collect only what is necessary, explain what it is doing, and give the individual control over the data and the recommendations.

The goal is not to create a score for every employee. The goal is to help a person make a better decision before a problem becomes expensive.

Every people-analytics vendor eventually arrives at the same product demo. Here is the dashboard. Here are the filters. Here are the colour-coded teams. Here is the engagement trend. Here is the attrition risk. Here is the heat map.

Then someone asks the question that matters: what should the manager do? The answer is usually another meeting.

A dashboard is useful when a person is already equipped to interpret it and act on it. It is not a substitute for interpretation. The right layer would consume the relevant signal and produce a bounded, reviewable action:

• Move Person A’s optional off-site. Her workload has risen for three consecutive weeks, her recovery window has narrowed, and she has a critical client delivery on Friday.
• Give Person B the Q4 stretch project. His output is stable, his workload is below his normal range, and his recent internal applications suggest he is looking for a larger challenge.
• Do not schedule Person C for another transatlantic trip this month. The last two trips were followed by delayed delivery and a sharp increase in sick days.

The point is not that an algorithm should make employment decisions. It should not. The point is that a system can help a human notice a pattern, understand its limits, and choose an intervention.

The reason this layer has not already been built is not simply technical. It is institutional.

Employees reasonably worry that data collected to support them could later be used to rank them, discipline them, deny them promotion, or identify them for removal. Those concerns are not hypothetical. Research on AI-enabled workplace monitoring finds a recurring tension: employees may recognise potential benefits while remaining concerned about accuracy, privacy, purpose, and how the data will be used.

A system that tells your manager you slept badly is not a wellbeing product. It is surveillance with a softer colour palette.

The architecture has to be different:

• The individual owns the raw data.
• Participation is explicit and voluntary.
• The system explains what it collects and what it infers.
• Raw personal data is not exposed to the employer by default.
• Employers receive only the outputs the individual has chosen to share.
• The system cannot be used for discipline, redundancy selection, or automated promotion decisions.
• Data has a retention limit.
• The individual can pause, export, correct, and delete it.
• A human remains accountable for consequential decisions.

This is not a compliance appendix. It is the product. The central question is not whether the company can obtain the data. It is whether the person would willingly give it to the system because the system demonstrably works for them.

If the answer is no, the company does not have a coaching layer. It has an extraction layer.

A dedicated human performance director for 200 senior employees could be extraordinarily expensive. But the original £60 million estimate assumes one £300,000 performance director per person. That is not a defensible baseline for most organisations. It makes the comparison dramatic, but it also gives a sceptical CFO an easy reason to dismiss the entire argument.

The financial case should be built from a company’s own baseline: value created equals avoidable departures, plus recovered capacity, plus reduced absence, plus faster ramp time, minus programme cost.

For a 200-person senior population, the relevant question is not whether software is cheaper than £60 million. It is whether a system can help prevent one avoidable departure, shorten one prolonged absence, or recover a small amount of capacity from the people who carry the most leverage.

If one senior departure costs £1.2 million in a particular business, the system does not need to transform the whole workforce to justify a meaningful investment. It needs to produce measurable change.

This is a better argument because it can be tested. A company should be able to run a controlled pilot, define the outcomes in advance, and stop if the outcomes do not improve.

This product is not competing for the same budget as gym memberships, meditation apps, or fruit in the office. Those products may have value. They are simply not designed to solve the problem described here.

The relevant budget sits closer to:

• Retention.
• Workforce planning.
• Executive coaching.
• Absence reduction.
• Manager effectiveness.
• Operational resilience.
• Productivity.
• Risk management.

The economic opportunity is large, but the familiar “$250 billion talent-loss cost” is too imprecise to carry the argument. A more defensible version is Gallup’s estimate that low engagement cost the global economy $8.9 trillion in lost productivity, based on its 2024 global workplace analysis. Gallup’s subsequent reporting estimated that the fall in engagement during 2024 alone represented $438 billion in lost productivity.

These are global estimates, not a company’s addressable market. They should be used to establish scale, not to imply that a single product can capture the entire number.

The Chief People Officer walks into the boardroom with something more useful than “engagement is up three points.” She has a view of capacity by function, team, and seniority. Not a single magic score. A set of carefully bounded indicators:

• Where workload is rising faster than recovery.
• Where high performers are becoming under-utilised.
• Where managers are absorbing unsustainable coordination work.
• Where travel and meeting patterns are eroding delivery.
• Which interventions were recommended.
• Which interventions were accepted.
• What changed afterwards.
• What the system got wrong.

The CFO does not receive an employee surveillance report. The CFO receives a business report: retention in critical roles, time to replace, time to productivity, absence patterns, delivery reliability, capacity recovered, intervention cost, and return on investment.

The system should show uncertainty. It should say when a recommendation is based on weak evidence. It should let the person disagree. It should create a record of what was suggested and why, not a hidden score that follows someone around the company.

That is what responsible instrumentation looks like. Human capacity becomes visible without becoming a target.

At Sloane, we are building the operating layer for human capacity. It connects the signals a person chooses to share. It understands the context around them. It remembers what has worked before. It turns a crowded set of observations into one useful next action.

For the individual, that might be: protect the two hours before tomorrow’s client presentation.

For the manager: remove the optional project from this person’s week.

For the company: stop treating preventable capacity loss as an inevitable cost of doing business.

The aim is not to know everything about people. The aim is to know enough to help them, and to give them control over what “enough” means.

Your company already has a nervous system for revenue, cost, risk, security, and customers. The missing layer is not another HR dashboard. It is a trusted way to understand the human capacity that makes every other layer work.