You Cannot Transform HR That You Cannot See

Walk into any Fortune 100 HR function today and you will find leaders carrying different titles, different mandates, and different quarterly pressures. The CHRO owes the board a transformation roadmap and a defensible business case. A new People Analytics leader has twelve months to prove the function is worth the headcount. The mature PA team is being pushed past dashboards into real organizational influence. The Workforce Systems Leader is fighting to make a stack of seventeen point solutions behave like a platform. The AI transformation sponsor needs a charter that does not embarrass them in front of the CEO. Shared Services has been told to standardize, automate, and do more with less.

Six roles. Six questions. Most organizations treat them as six separate problems.

They are not.

Every one of those mandates requires the same underlying input: a clear, current, queryable picture of how the HR organization actually works. Not how it looks on the org chart. Not what the Workday configuration implies. Not what the playbook says the process should be. How it actually works, with every workaround, every handoff, and every undocumented exception that keeps the function running day to day.

And nobody has that picture.

The operating manual lives in people's heads

Most of what makes an HR function function is not in a system. It is in the HRBP who has been with the business unit for eleven years and knows why the comp bands were set that way. It is in the HRIS analyst who remembers which fields in the legacy system were quietly repurposed during the last re-org. It is in the shared services lead who has memorized the six real reasons a case gets escalated, only one of which appears in the published SLA.

This is tacit knowledge. It is what anthropologists, systems theorists, and organizational researchers have been telling us for decades: the work-as-imagined lives in the documentation, and the work-as-done lives in the practitioners. The gap between those two is where most transformation initiatives go to die.

When a CHRO commissions a transformation plan, the assumption is that the organization can describe itself. It largely cannot. When a People Analytics leader builds a roadmap, the assumption is that the data reflects the real process. It largely does not. When an AI sponsor maps use cases, the assumption is that the processes being automated are well understood. They largely are not.

You cannot transform what you cannot see. You cannot see what no one has written down.

The four versions of any HR process

Years ago, running People Analytics at Nike, I got asked a workforce question that on paper should have taken an afternoon to answer. The data existed. The systems were in place. The process was documented. It took three weeks. Not because the data was bad, but because three separate teams each had a different version of how the underlying work actually happened, and none of the three versions matched what the handful of people doing the work were doing every day.

That is when a framework started to take shape for me. Any HR process you can name sits at the intersection of four different versions of itself:

1. How the work actually happens today. The lived process. Workarounds, handoffs, and exceptions included. Only the people running it know this version in full.

2. How the work should happen today. The intended design, sitting behind the policy doc, the system configuration, and the training deck.

3. How the work is documented today. The handbook version. Usually drifted out of sync with the lived process years ago.

4. How the work ought to happen tomorrow. The target state, assumed whenever transformation, automation, or AI is on the table.

Most organizations assume three of the four versions are roughly the same. They are not. The gap between what people actually do, what they are supposed to do, what is written down, and what leadership wants to build toward is where every transformation initiative runs into reality.

Take something as routine as onboarding. The documented version is a clean sequence of system-generated tasks. The intended version is a well-orchestrated first ninety days. The target version is an AI-assisted experience that scales. The actual version, in most organizations, depends on which manager the new hire gets, which HRBP is paying attention that week, and whether someone remembers to flag the five system quirks that the handbook does not mention. Four versions of one process. Automate the documented one and you have just scaled the wrong thing.

This framework is one piece of what I have been calling organizational literacy: the practical capacity of an organization to read, describe, and reason about its own reality. Most HR functions do not have it. They have systems that implicitly claim to represent the organization, documentation that claims the same, and a lived reality that contradicts both. The literacy gap is the transformation gap.

When the HR function runs on people, not systems

Here is the part that rarely gets priced into transformation plans. When the operating manual lives in people's heads rather than in systems, the HR function runs on people who in turn run on “heroics”.

That sounds exciting, being a hero for a day is fine, but anyone that has been in that position for some time knows it’s a guaranteed path to burn-out. 

Individual HRBPs, HRIS analysts, and shared services leads become the connective tissue between parts of the organization that do not talk to each other through any other channel. They translate requests from the business into something the systems can process. They shepherd employees through workflows that were never designed to handle the case in front of them. They remember which regional variant to invoke, which system has the clean data, which stakeholder needs to be copied for a decision to actually stick.

That is not a process. These are the Human APIs of your organization, and every senior HR leader who has done the work knows exactly who they are.

The cost is not that the work stops getting done. It gets done. It gets done brilliantly, every day, by people who are fundamentally holding the function together with pattern recognition and goodwill. The cost is that it burns them out, the heroics go unsung, and the HR function spends its most capable people shepherding employees through the organization rather than driving business outcomes. Careers get built on knowing where the joints are. When those careers end, institutional memory walks out with them.

You cannot build an AI-enabled HR function on top of human APIs. You cannot design automation around processes whose real logic lives in somebody's head. You cannot write a credible transformation roadmap without first understanding how the work actually happens.

Why consulting has not solved this

Every consulting engagement that touches HR runs the same play. Interview people. Take notes. Produce a PowerPoint. Deliver a deck. Move on.

The notes get archived in an unreasonably beautiful PPTX file on someone’s desktop. The recordings, if they ever existed, are never opened again. The knowledge captured during the engagement has a half-life of about six months before the next firm (or same firm) is hired to run essentially the same interviews to answer the next question (or same questions). A year later, the organization has paid three times for the same discovery, and the institutional memory still lives in the same three people who are thinking about retiring.

This is not a failure of effort. It is a failure of infrastructure. The industry has treated discovery as logistics: schedule the interviews, run them, write them up, ship the deliverable. Discovery is not logistics. It is an intellectual process of capturing what an organization knows and does not yet know it knows. Treated that way, it demands different tools, different methodology, and different outputs.

A different approach: capture, structure, activate

Treat discovery as infrastructure and the approach changes. The starting premise is simple: the knowledge the organization needs is already there. It is sitting in conversations that have never happened, in the heads of people who have never been asked the right questions, in the space between what the systems report and what the people actually do.

Capture it properly. Structure it so it can be queried. Activate it against the questions that actually need answering.

In practice that means a few dozen structured stakeholder interviews, each one scoped specifically to who the person is and what every prior interview has already established. The conversations generate over 1,700 pages of structured qualitative data. That data gets processed through a five-layer architecture: raw recordings are transcribed, transcripts are cleaned for clarity and consistency, facts are extracted into structured statements, those statements are anonymized to remove individual attribution, and the result is a queryable knowledge base the organization owns.

From that knowledge base, specific deliverables get produced on demand. A CFO-ready business case for an HCM investment. An AI readiness charter. A tech rationalization roadmap. An org design assessment grounded in how work actually flows. Each draws from the same foundation. The second delivers faster than the first. The tenth delivers faster than the fifth.

The question worth asking

The question is not whether HR organizations have the knowledge they need. They do. The people doing the work know. They have always known.

The real question is what it costs to keep treating that knowledge as ambient noise rather than infrastructure. Every six months, another external team gets hired to run another round of interviews to answer another version of the same question. The notes go into another shared drive. The recordings get stored. The institutional memory continues to live in the same three people who are thinking daily about their lack of interest in heroics and retirement.

That is not a knowledge strategy. It is a tax on not having one.

The HR organizations that get transformation right in the next decade will be the ones that treat discovery as a durable capability rather than a recurring cost. They will have an operating manual. It will be queryable. It will belong to them.

Everyone else will keep paying for the same interviews, three firms at a time.

About Ikona Analytics

Ikona Analytics is an intelligent knowledge operations platform that captures, structures, and activates the tacit knowledge inside HR organizations. The Ikona Systems Diagnostic (ISD) is the foundational engagement: structured stakeholder interviews processed through a five-layer knowledge architecture, producing a persistent knowledge base the organization owns. Intelligence modules turn that knowledge into business cases, AI readiness charters, tech rationalization roadmaps, and transformation plans on demand. Validated across Fortune 100 organizations in pharma, consumer goods, automotive, and financial services.

If the question raised in this article is live at your organization, the Ikona team would like to hear about it. Start a conversation →

Read more of our perspectives at Ikona Insights.