Request a briefing

Essay · Decision Rights

Brakes That Let You Go Faster: Designing Governance for Agile Organizations

How governance designed as enabling constraints gives teams room to act and gives leaders better control.

Brakes That Let You Go Faster is an essay on agile governance for executives and operating leaders whose compliance, risk and security controls slow the changes their teams try to make. It shows how decision rights matched to the situation, constraints that enable action and controls designed into daily work give leaders better control while teams deliver with less delay.

A team can finish new software in days and still wait weeks to put it in customers' hands. Until release, finished software is inventory: money already spent, waiting in a queue and earning nothing. Many organizations that adopt agile delivery keep their older approval boards and final review gates, so faster building only lengthens that queue. In our experience, shortening the wait starts when leaders treat agile governance as a design choice, one that lets teams release working software to customers on a steady cadence while leaders keep the control a regulated business needs.

Agile governance decides who may decide what, and where

Agile governance, in our view, is the design of how power flows through an organization, so each decision is made where its context sits, with only as many limits as the risk calls for. In operating governance, authority comes down the reporting lines and decision rights settle the questions that run across them.

We see undesigned power do its damage at the borders between functions. Product teams are measured on shipping and learning, while security, legal and finance are charged with keeping the wheels on. Without designed decision rights, the two sides settle priorities through a Game of Thrones contest of rank and stamina, and decisions wait while they do.

A risk-free system is a dead system

Good governance prices each control: what it protects and what it costs in flow. Airport security shows the trade-off: each screening rule about shoes, liquids or laptops is meant to lower one risk, and each one lengthens the line. Screen out every risk, and nobody boards. Airports also run a second design, in which travelers vetted in advance through the Transportation Security Administration's PreCheck program keep their shoes, belts and light jackets on, because their risk was judged before they reached the checkpoint.

Limits can also make action possible. Brakes let a driver travel fast, because a car that could not stop would be safe only at walking pace. Teams that know the limits they may not cross can act without asking permission. Limits of this kind are enabling constraints: each one fixes a boundary and leaves the choices inside it open.

We find rules serve an organization better when people treat them as texts to interpret and improve. Leaders who propose new governance often hear that policies are carved in stone, and Jewish tradition offers a corrective. The Hebrew name for the Ten Commandments, Aseret ha-Dibrot, means "the ten utterances." By the rabbinic count, the Torah has 613 commandments, and the medieval commentator Rashi, following Saadia Gaon, taught that all 613 are contained in those ten utterances. Even the carving became a lesson: Pirkei Avot asks that harut, the word for engraved on the tablets, be heard as herut, freedom. A company policy likewise interprets a law, a risk or a choice about how the business runs, written for a context that may since have changed.

Teams learn to work around distant approval gates

Preventive controls, approvals required before a change, often become queues when placed far from the work. People whose organizations required an outside body, such as a change approval board, to approve changes were 2.6 times as likely to fall among the low performers at software delivery in a 2019 survey of almost 1,000 technology professionals. Formal approval in that survey showed no link to a lower rate of failed changes. A survey shows a pattern, so leaders should check what their own board catches and delays.

In our view, the change approval board has hardened into orthodoxy. Information technology service management frameworks teach formal approval as the route to stable releases, and regulators in finance, healthcare and government often require that someone other than a change's author approve it. Peer review and automated checks can meet that requirement in lighter ways, as the survey report describes. In our experience, a board still earns its place for a change that cannot be tried in small steps or rolled back, or one whose consequences reach across teams.

Locking work down tends to produce workarounds, and workarounds hide risk. We have watched teams answer a new restriction with a private spreadsheet or an unapproved tool, and we treat each workaround as a sign that the official route does not fit the work.

We often see governance, risk and compliance specialists scored on whether others follow their controls. Full compliance and poor decisions about customers can sit side by side, and a compliance-only scorecard will not show the cost in delays, rework or unreported exceptions.

The right decision tool depends on how much anyone knows

Decision-rights charts, naming who recommends, agrees, performs, gives input and decides, settle routine questions quickly where cause and effect are known and decisions repeat. Where nobody can see cause and effect in advance, we have seen those charts give false confidence. There, small trials run with the people closest to the work produce better decisions.

Shared information can give decentralized decisions the context they need. For three years I worked at a company that opened its financial data to every employee, although proprietary numbers could have reached a competitor. To my knowledge, the openness produced no adverse incident in those years, and people could see how their local choices affected the company's results. One company is a small sample.

Compliance belongs inside the work, close to the people who do it

Compliance designed into daily work gives faster feedback and spares teams a scramble before each audit. Automated checks in the flow of work can record who changed what, who reviewed it and which tests passed, giving auditors a record made as the work happened. The record helps only where its checks cover the obligations and exceptions reach someone able to act.

Many company policies we review are what rabbinic tradition calls a fence around the Torah: a rule added to a law to keep people well clear of breaking it. A fence can also do harm: the same tradition warns that one taller than the field it encloses can fall and cut down the young plants. We change governance by walking those fences one at a time, asking which risk each guards against, what delay it adds and whether the regulation requires it. A sweeping overhaul announced from the top tends to meet the organization's antibodies, so we make small changes, which cost less and are safer to try. Before each one, we agree what it must protect and which result would reverse it. We track incidents and exceptions alongside delay, because a shorter queue with more exposure is a worse design.

In our experience, governance specialists do their best work inside the teams they serve, keeping the independence their review needs. We judge them by those teams' results for customers and by the risks their review catches. Leaders who bring specialists inside get brakes their teams trust, and teams that trust their brakes can go faster.


The workshop Strategy Design and Its Management Systems helps leadership teams design the decision rights, boundaries and controls their organization runs on. Request a briefing.

Questions leaders ask about agile governance

What is agile governance?

Agile governance is the design of who may decide what, where, and within which limits, so teams can act without waiting while leaders keep the control they need. That design places each decision with the people who have the context to make it, sets boundaries teams may not cross, and puts compliance checks into daily work, where they give feedback as the work happens.

Do change approval boards reduce risk?

Respondents whose organizations required an outside body, such as a change approval board, to approve changes were 2.6 times as likely to fall among the low performers at software delivery in a 2019 survey of almost 1,000 technology professionals. Formal approval showed no link to a lower rate of failed changes. In our experience, a board still earns its place for a change that cannot be tried in small steps or rolled back.

How should a company start changing its governance, risk and compliance processes?

Change one policy at a time. Treat each company policy as a fence added around a law, and ask what it protects, what it costs in delay and whether the regulation itself requires it. Small changes are cheaper and safer to try and easier to roll back. Track incidents and exceptions alongside delay, so a shorter queue does not come with more exposure.

What are enabling constraints?

Enabling constraints are limits that make action possible. Each one fixes a boundary and leaves the choices inside it open, so a team that knows what it may not cross can act without asking permission each time. Brakes are a familiar case: a car that could not stop would be safe only at walking pace, and brakes let a driver travel fast.

References (16)

This essay is developed from William Haas Evans's manuscript Designing Resilience: The Strategy, Structure, and Spirit of Enterprise Agility.

  • Babylonian Talmud, tractate Makkot 23b. The count of 613 commandments, taught in the name of Rabbi Simlai. https://www.sefaria.org/Makkot.23b
  • Rashi (Shlomo Yitzchaki), commentary on Exodus 24:12, citing Saadia Gaon's Azharot. All 613 commandments counted as included in the Ten Commandments. https://www.sefaria.org/Rashi_on_Exodus.24.12.3
  • Pirkei Avot 6:2, from the sixth chapter (Kinyan Torah), a later teaching added to the tractate. The teaching that harut, engraved, be heard as herut, freedom. https://www.sefaria.org/Pirkei_Avot.6.2
  • Bereshit Rabbah 19:3. Rabbi Hiyya's teaching that a fence should not stand taller than what it protects, lest it fall and cut down the saplings. https://www.sefaria.org/Bereshit_Rabbah.19.3
  • Mishnah, tractate Avot (Pirkei Avot) 1:1. The instruction of the Men of the Great Assembly to make a fence around the Torah. https://www.sefaria.org/Pirkei_Avot.1.1
  • Robert Simons, Levers of Control: How Managers Use Innovative Control Systems to Drive Strategic Renewal, Harvard Business School Press, 1995. Boundary systems, and the image of brakes that let a car go fast (quoted from this book in Kesler and Kates, 2011).
  • Robert Simons, "Control in an Age of Empowerment," Harvard Business Review, vol. 73, no. 2, March to April 1995, pages 80 to 88. https://store.hbr.org/product/control-in-an-age-of-empowerment/95211. Boundary systems as limits on the search for opportunity.
  • Gregory Kesler and Amy Kates, Leading Organization Design: How to Make Organization Design Decisions to Drive the Results You Want, Jossey-Bass, 2011, chapter 8, "Design for Operating Governance." https://www.oreilly.com/library/view/-/9780470589595/. Operating governance as the allocation of power across units and levels.
  • Stanley Salthe, Development and Evolution: Complexity and Change in Biology, MIT Press, 1993. The term enabling constraints, as credited in Juarrero (1999).
  • Alicia Juarrero, Dynamics in Action: Intentional Behavior as a Complex System, MIT Press, 1999. https://openlibrary.org/books/OL37173M. Constraints that make new, coordinated behavior possible.
  • Nicole Forsgren, Dustin Smith, Jez Humble and Jessie Frazelle, Accelerate: State of DevOps 2019, DevOps Research and Assessment (DORA) and Google Cloud, 2019, pages 48 to 52. https://dora.dev/publications/pdf/state-of-devops-2019.pdf. Change approval, segregation of duties and peer review.
  • Paul Rogers and Marcia Blenko, "Who Has the D? How Clear Decision Roles Enhance Organizational Performance," Harvard Business Review, January 2006. https://www.bain.com/insights/who-has-the-d. The RAPID decision roles: recommend, agree, perform, input and decide.
  • David Snowden, "The Social Ecology of Knowledge Management," in Charles Despres and Daniele Chauvel, editors, Knowledge Horizons: The Present and the Promise of Knowledge Management, Butterworth-Heinemann, 2000. The first use of the name Cynefin, according to David Snowden, "The evolution of Cynefin over a decade," The Cynefin Company, 2010. https://thecynefin.co/the-evolution-of-cynefin-over-a-decade/
  • Cynthia F. Kurtz and David J. Snowden, "The new dynamics of strategy: Sense-making in a complex and complicated world," IBM Systems Journal, vol. 42, no. 3, 2003, pages 462 to 483. A later form of the Cynefin framework, with different domain names.
  • David J. Snowden and Mary E. Boone, "A Leader's Framework for Decision Making," Harvard Business Review, November 2007, pages 68 to 76. https://hbr.org/2007/11/a-leaders-framework-for-decision-making. Choosing how to decide by the relationship between cause and effect, under the Cynefin domain names simple, complicated, complex, chaotic and disorder.
  • Mary Poppendieck and Tom Poppendieck, Lean Software Development: An Agile Toolkit, Addison-Wesley, 2003. Partially done work as waste, behind the opening's point that unreleased software is inventory.

William Haas Evans

Founder and chief strategy officer of Fugue Strategy Advisors. He works with leadership teams on strategy design, strategic foresight and organization design. More about William

Read next

Request a briefing

What is standing between your team and its goals this year?

Tell us in a 30-minute briefing, and we'll help you decide where to start.

Request a briefing