Why the HR tech stack keeps re-consolidating

The pattern is now twenty years old. A monolithic HR suite — Workday, Oracle HCM, SAP SuccessFactors — gets called brittle, expensive, and slow. A wave of point solutions arrives, promising to do one thing (performance, learning, recognition, scheduling) better than the suite ever did. Buyers buy them. Three years later, the same buyers complain about integration costs, identity sprawl, contract overhead, and the fact that no two systems agree on what a “manager” is. The suite then absorbs the most successful point solutions, or rebuilds the category in-house, and the cycle resets.

This has happened with applicant tracking. It has happened with learning. It happened with performance management twice. It is happening now with skills inference and with AI-assisted coaching. There is no reason to expect the pattern to stop.

The structural reason

The reason point solutions win individual category battles and lose the procurement war is that HR buyers do not buy software. They buy headcount reduction in their own IT function. A specialist tool that requires its own SSO integration, its own data residency review, its own SOC 2 renewal, its own quarterly business review, and its own vendor relationship is — net of CIO time — almost never cheaper than the same feature as a checkbox inside an existing suite.

The suite vendor knows this. The suite vendor’s pricing power is therefore not about feature parity. It is about integration debt amortisation. Once a company has wired Workday or its equivalent into payroll, identity, finance, expense, learning, and analytics, the marginal cost of adding another module from that vendor is approximately zero. The marginal cost of adding any other vendor is one new line in a procurement review, indefinitely.

Why the disruptors keep arriving anyway

The disruptors arrive because the suite features are, individually, bad. Workday’s learning module is worse than the median dedicated LMS. SuccessFactors performance management is worse than the median dedicated performance tool. Oracle’s recognition module is barely a product. This is not a temporary state of affairs — it is the equilibrium. The suite has no incentive to invest deeply in any single module because the lock-in does not depend on per-module quality. So each module is built to a “good enough to not lose the deal” standard, and a market opens for whoever can build the actual best version.

That best-of-breed vendor then sells against the suite. They win some deals on feature quality. They lose more deals on integration friction. After three or four years they either get acquired by the suite (most common), get absorbed into a competing suite, or pivot into a more defensible adjacent category.

What this means for buyers right now

The current AI wave is no different. The skills-inference startups, the AI coaching tools, the autonomous talent agents — they will all face the same procurement gravity well within thirty-six months. Buying them is fine. Building a strategy that assumes they will exist as independent vendors in 2029 is not.

The buyers who get this right treat point solutions as a temporary capability rental. They negotiate short contracts. They avoid deep customisation. They assume the eventual landing place is either the suite or absorption. The buyers who get it wrong sign three-year deals at premium pricing for capabilities that are about to ship inside the suite they already own.