Agile in 2026 – What the Framework Got Right and Where Practice Stalls
After the Manifesto.
After the sprint.
Agile started as a 2001 protest against how large software shops ran projects. Nearly three decades later, the principle has migrated out of the product team and into the operating model itself, where boards now ask whether the company, not the squad, can adapt.
The pre-Manifesto decade.
Jeff Sutherland and Ken Schwaber formalized Scrum at OOPSLA in 1995, drawing on Nonaka and Takeuchi’s 1986 Harvard Business Review piece on the rugby-style product team (Takeuchi and Nonaka 137-46). Kent Beck published Extreme Programming Explained in 1999. The pieces were assembled before the Snowbird group ever met (Sutherland 21-30).
The Manifesto in one weekend.
Seventeen practitioners met at Snowbird, Utah in February 2001 and wrote four value statements and twelve principles. The four values shifted weight from process to people, from documentation to working software, from contract to collaboration, from plan to response (Beck et al.).
The principle outgrew the team.
The mechanic is moving past the product squad. Spotify codified the squads-tribes-chapters language in 2012. ING restructured 3,500 staff onto roughly 350 squads in 2015. Bayer launched Dynamic Shared Ownership in 2024, on a 90-day cycle, across 100,000+ employees. Haier’s RenDanHeYi predates them all. The cadence varies. The direction is the same (Mahadevan et al.; Bayer AG; Hamel and Zanini).
Individuals and interactions over processes and tools.
Working software over comprehensive documentation.
Customer collaboration over contract negotiation.
Responding to change over following a plan.
Source · agilemanifesto.org · February 2001 · 17 signatories
Where the practice sits in 2026
The framework took. The practice migrated. Inside the product organization, Scrum is the dominant operating cadence at the team level, and the question of whether to adopt it has long since closed (Digital.ai). The interesting question now sits one floor up. A growing list of large firms have moved Agile principles out of the squad and into the corporate operating model itself, each on a different cadence and a different vocabulary, but with the same direction of travel: push decision authority closer to the work.
The pattern is not a single playbook. It is a category of bets that boards are now placing on the operating model itself.
Four reference points, four different shapes.
Chinese white-goods conglomerate broken into roughly 4,000 self-managed micro-enterprises, each with its own P&L and contracting relationships internal and external. The earliest enterprise example of a flattened agile operating model at scale, and the conceptual ancestor of the others.
A 2012 internal whitepaper described squads, tribes, chapters and guilds as the company’s organizing language. The paper was descriptive, not prescriptive, but it produced the vocabulary that every later transformation borrowed (Kniberg and Ivarsson).
Dutch banking group restructured 3,500 headquarters staff into roughly 350 nine-person squads inside 13 tribes, and ran the alignment cadence on Quarterly Business Reviews. The first major regulated-industry transplant of the model (Mahadevan et al.).
Dynamic Shared Ownership, rolled out across 100,000+ employees on a 90-day operating cycle. Twelve organizational layers cut to six or seven. Roughly 2,000 self-directed teams of six to ten. About 12,000 positions removed. Target: €2 billion in sustainable savings by the end of 2026 (Bayer AG; Groysberg and Yucaoglu).
What the four cases share, and what they do not
None of the four runs the same cadence. Spotify did not have a planning loop on a calendar. ING aligned tribes through Quarterly Business Reviews. Bayer chose a 90-day cycle as the explicit operating heartbeat of the firm. Haier’s micro-enterprises contract with each other on terms set by the work itself, not by the calendar. The shared move is not the cadence. It is the relocation of decision authority closer to the customer, and the willingness of executive leadership to absorb the structural cost of doing so.
Most of these transformations are difficult, slow, and frequently unsuccessful. Industry estimates put the failure rate of large agile transformations between 47 and 70 percent (BAI; Robinson). The failure mode is rarely the framework. It is the unwillingness of executive sponsorship to absorb the implications of the framework, particularly the implication that the existing managerial hierarchy will need to shrink for the operating model to work.
Why this matters at the board level
For two decades, the board’s question about Agile was whether engineering should run sprints. That question is now answered and largely uninteresting. The 2026 question is whether the firm itself can adapt at the speed the market is changing, and whether the operating model is the bottleneck. McKinsey, the Business Agility Institute, and the Scaled Agile community have all converged on a phrase for this: business agility, the operating system rather than the practice (Sidky et al.; SAFe). Amazon, Citibank, and Dell are all reported to be exploring variants of the same shift (Groysberg and Yucaoglu).
What the line reads in plain English
| Sprint | Story points delivered | Cycle time (days) | What changed |
|---|---|---|---|
| S01 | 22 | 9.4 | Baseline. Mixed-discipline team running first cadence. |
| S03 | 26 +18% | 8.1 −14% | Backlog refinement tightened. Acceptance criteria written before sprint planning. |
| S06 | 34 +55% | 6.8 −28% | Demos exposed scope inflation. Team began rejecting stories without a clear customer outcome. |
| S09 | 38 +73% | 5.9 −37% | Retro pattern stabilized. Two engineering hires absorbed cleanly. |
| S12 | 41 +86% | 5.4 −43% | Velocity stable. Cycle time stable. Team rotated focus to lead-time-to-value. |
The Agile question is no longer whether engineering should run sprints. The question is whether the firm itself is built to adapt.
Pick the largest decision your company made in the last twelve months. Map the path it traveled, from the person who first identified the issue to the person who finally signed it off. Count the layers. Count the days. If the number is uncomfortable, the operating model is the bottleneck, not the team. The 2026 question for the board is not whether to scale Scrum. It is whether the company is willing to push decision authority closer to the work, and to absorb the structural consequences of doing so. Haier did it through micro-enterprises. ING did it through Quarterly Business Reviews. Bayer did it through 90-day cycles and the elimination of nearly all of its management documents. The shape of the answer will differ. The question is the same.