> ## Content Index > Fetch the complete content index at: https://wi.senterprises.com/llms.txt > Use this file to discover other available public pages before exploring further. # Give the Council Something to Decide - URL: https://wi.senterprises.com/articles/mdm-give-the-council-something-to-decide/ - Published: 2026-10-26T08:00:00.000Z - Updated: 2026-11-06T03:12:08.000Z - Description: A governance council is usually built on its charter. It lasts on two things the charter cannot supply: a room of people who can actually decide, and a steady supply of decisions it is required to make. - Author: Jeffrey Shabel - Tags: General MDM, Data Governance Operating Models, 2026, October 2026, 2026-W44, AI: Contextual, #wi-art_01KVZKVTFN6VKVPEN1N7FAV0QM *Weaving Intelligence (House AI Style), with Jeff Shabel’s Experiential Commentary* A data governance council is usually designed on paper first. It gets a charter, a membership list drawn from the org chart, a decision-rights matrix and a standing slot in everyone's calendar. Asked about a council that worked, I did not start with the paper: > Councils work in business in the way councils work everywhere: through relationships and alliances. They work when the overriding opinion is the consensus opinion—whether that opinion is best for the Enterprise or not is another matter. The block quotations in this piece are my answers to the questions put to me. The prose around them is my argument from those answers and the sources cited; where a passage goes past both to something I am proposing or forecasting for your council, it opens with "my reading". The quotation above is a description, not a recommendation. A council produces the view its members can agree on, and the view its members can agree on is shaped by who owes whom, who needs whom next quarter and which alliances already existed before the first meeting. Whether that view is the best one for the enterprise "is another matter," and the clause is meant: a council that works is one whose decisions stick, which is not the same as one whose decisions are right. A council lasts when two conditions hold together: the room contains people who can actually decide, and the council is owed a steady supply of decisions it is required to make. Take away the first and it becomes a debating society whose conclusions someone else overrules. Take away the second and it becomes a status meeting, then a calendar habit, then nothing. The claim forbids two things a reader can check against any council they sit on: a council whose members cannot commit the people with authority, yet whose decisions hold, and a council with no queue of required decisions that is still doing governance a year later rather than hearing updates. I have said before that councils, like stewardship models and maturity assessments, are correct and oversold, and that each lives or dies on the weight of the champion behind it ([Rules of Decreasing Strictness](https://wi.senterprises.com/articles/edg-rules-of-decreasing-strictness/)), because it asks people to take time out of their day for something that does not touch their own deliverables. I have also said that councils add bureaucracy where it is least useful: they are good for signing off proposed changes and poor at setting standards or direction, where they suffer from too many cooks ([Keep the Policy, Keep the Score, Move the Council](https://wi.senterprises.com/articles/keep-the-policy-keep-the-score-move-the-council/)). This piece is about how to build a council that lasts, and it rests on the pattern across the councils I have seen rather than on one council's story. The stakes are not small. Gartner predicted in February 2024 that "By 2027, 80% of data and analytics (D&A) governance initiatives will fail due to a lack of a real or manufactured crisis" [\[1\]](#ref-1). That is an analyst forecast, and the release gives no method behind the number, so it is evidence of where the analysts' attention is rather than a measurement. Gartner's prediction concerns governance initiatives as a whole, not councils, and the release does not say what a crisis supplies. My reading is that one thing a crisis supplies is decisions that cannot wait, and a council without them has nothing it is required to do. ## How a council actually decides The formal account of a council is its decision rights. The Data Governance Institute names the prior question behind any governance rule or data decision as "who gets to make the decision, and when, and using what criteria?" [\[2\]](#ref-2). Robert Seiner's non-invasive framework says the strategic level is often labeled "the Data Governance Council" and gives it the role of the body that "acts as the ultimate decision maker for resolving issues that cannot be resolved at a lower level of the organization" [\[3\]](#ref-3). That account is accurate as far as it goes. The informal account is the one that decides whether the formal one works: > I have regularly seen councils break down when groups with opposing opinions dig in, even if there is a tiebreaker who can keep the work moving. Councils need members with significantly greater soft skills than hard skills to be successful as the decisions are rarely solely based upon the technical details. Both halves of that answer cut against the usual design. The first says a tiebreaker does not rescue a council whose factions have dug in. My reading is that a tiebreaker can close a vote but cannot make the losing faction carry out the result, and a council whose decisions are not carried out has stopped being a council in everything but the calendar. The publisher's description of Jeffrey Pfeffer's *Managing with Power* makes the same point from the other side: "a decision by itself changes nothing" [\[4\]](#ref-4). What changes things is the political support behind the decision. That fits the rule I have given before for contested definitions: the decider decides, meaning whoever holds the authority and, more importantly, the power to enforce ([Rules of Decreasing Strictness](https://wi.senterprises.com/articles/edg-rules-of-decreasing-strictness/)). A tiebreaker with that power is the decider, and the council is where the decision gets its sign-off. A tiebreaker without it is a chair. It also squares with a line I wrote earlier, that an escalation path ending in a council ends nowhere ([Before It Had a Name](https://wi.senterprises.com/articles/edg-governance-frameworks-before-the-name/)). That is true of a council with nobody in it who can enforce, which is exactly the council this article is trying to prevent. The second half of the answer says the members a council needs are the ones who can work a room. A disagreement about how to define an active customer is never only a disagreement about a definition. It is also about whose reports will change, whose numbers will move and whose team will do the rework, and a member who can only argue the technical merits will lose to one who can see what each faction needs in order to agree. The survey evidence points the same way. When Gartner asked data and analytics leaders what stood in their way, "The top six roadblocks to D&A reported in the survey are all human-related challenges" [\[5\]](#ref-5). Its account of what effective chief data and analytics officers prioritise names "critical leadership behaviors such as relationship building, influencing, communication and building an engaged team" [\[6\]](#ref-6). Neither study is about councils. Both are about the people councils are made of. There is a cost to admitting that councils run on alliances, and it should be stated plainly. A council that works produces the coalition's answer. Paul Rogers and Marcia Blenko of Bain put the risk in one line: "Consensus is a worthy goal but can be an obstacle to action or a recipe for lowest-common-denominator compromise" [\[7\]](#ref-7). That answer is usually durable, because the people who must live with it agreed to it, and it is not guaranteed to be right. The design question is therefore not how to keep politics out of the council, which cannot be done, but how to make the coalition argue about evidence it has all read, against criteria it agreed before the argument started. The first three meetings exist to do that. ## Who is in the room > Who was in the room? The actual decision makers or their proxies and a very small number of technical experts whose role was advisory. Each part of that answer rules something out. "The actual decision makers" rules out a council staffed by whoever had time, because the people with time are rarely the people with authority. John Kotter's study of failed transformations made the same observation about the coalitions that lead change: "Senior managers always form the core of the group," and "groups without strong line leadership never achieve the power that is required" [\[8\]](#ref-8). Seiner's framework describes the council's members as people who "report directly to and are delegated by the Executive level" [\[3\]](#ref-3), which is the same answer with the proxy built in. "Or their proxies" allows delegation, with a condition that follows from the decider rule, and my reading is that a proxy has to be able to commit the principal in the room. My reading is that a proxy who must take every question back to the principal turns the council into a relay, and a relay decides nothing at its own meetings; only a proxy who can commit the principal is a decision maker at all. "A very small number of technical experts whose role was advisory" rules out a failure data governance councils are especially prone to: filling with the people who understand the data and emptying of the people who own the budget. The experts answer whether something can be done, what it will cost and what it will break. They do not vote on whether it should be done, because that question is about the enterprise's priorities, and the people who set those priorities are the decision makers. This does not excuse the decision makers from understanding the data. MIT Sloan's account of data literacy is that "Leaders need to trust and understand data well enough to make good decisions," and MIT Sloan's Rama Ramakrishnan says the goal for a leader should be to become "a fast but effective consumer of analysis that is produced by my organization" [\[9\]](#ref-9). That is the bar for a council member: able to consume the experts' analysis, not able to produce it. Size follows from the same logic. Aaron De Smet of McKinsey puts the threshold bluntly: "More than six decision makers with a vote starts to become a problem" [\[10\]](#ref-10). My reading is that a council drawn from ten domains does not need ten votes: vote with the few who hold the budget and the power to enforce, and bring each domain lead in as an adviser who answers what can be done, what it costs and what it breaks, with their case carried in the pre-read. A council of twenty is not twenty decision makers. It is an audience with a few decision makers in it, and the audience is where the status meeting comes from. ## The first three meetings The build sequence is short, and every step of it is about making the first decision possible. This is what I would put on the first three agendas: > Meeting 1: Council organization, roles and introductions. What the council is for, why it has the members it does, what decisions it’s expected to make and using what criteria and member introductions. That’s usually enough for a half hour. Half an hour is usually enough. My reading is that this is because nothing is decided in it: a first meeting that tries to decide something decides it before anyone knows what they are deciding by. The item that matters most is the one that sounds most like paperwork: what decisions the council is expected to make, and by what criteria. It is the Data Governance Institute's question again, "using what criteria?" [\[2\]](#ref-2), asked before any decision is on the table. That timing matters because of how people react to decisions that go against them. Joel Brockner and Batia Wiesenfeld's review of the research on reactions to decisions, which the abstract says covers 45 independent samples, found that the fairness of the procedure "is more positively related to individuals' reactions when outcome fairness or valence is relatively low" [\[11\]](#ref-11). Put plainly: when people lose, the process is what they judge. My reading is that a member who loses a vote at the sixth meeting is more likely to judge the result by criteria agreed at the first, before anyone knew which way they would cut. Fix them then. The second meeting reviews the current state and the third brings proposed changes and the first decisions. The order matters. My reading is that a council asked to decide before it has seen the current state will either defer, which teaches it that deferring is what it does, or decide on whatever its loudest member already believed. Michael Mankins and Richard Steele's advice to executive teams is the same sequence in their words: "Design and conduct strategy sessions so that participants agree on facts related to each issue before proposing solutions" [\[12\]](#ref-12). My reading is that the current-state review gives every member the same facts before the first proposal arrives, so that the relationships and alliances in the room argue about the proposal rather than about the facts. The third meeting is where the council does what it is useful for: signing off, or declining to sign off, on changes put to it. It does not draft the standard. Someone brings the council a proposal, and the council decides. > For Meetings 2 and 3, make sure the documents (and there should be documents) are available as early as possible before the meeting so the Council Members can review them and come ready with questions. "And there should be documents" carries more weight than its parentheses suggest. My reading is that a council that receives its material in the room spends the meeting reading, and a decision made on a first reading is a decision made on whatever the presenter chose to emphasise. Bain's committee guidance sets a floor: "circulate the agendas and supporting documents at least 48 hours beforehand" [\[13\]](#ref-13). One well-known practice points the other way on the mechanism, and it is worth knowing why. Amazon writes "narratively structured six-page memos" and then, in Jeff Bezos's words, "We silently read one at the beginning of each meeting in a kind of 'study hall'" [\[14\]](#ref-14). That guarantees the reading by spending meeting time on it rather than trusting people to arrive prepared. It agrees with the goal, which is that everyone decides from the same document read in full, and disagrees about how to get there. My reading is that either way works. What does not work is a council that sees the material for the first time when the presenter shares a screen. ![Three meetings in sequence: organisation and roles, a current-state review, then proposed changes and first decisions; after them a running rhythm of required reports and required decisions; without work, the council is re-absorbed.](https://wi.senterprises.com/assets/diagrams/FIG014_MDM_The_First_Three_Meetings_v1_0.png) Figure 1: The first three meetings build to the council’s first decision; after them, required reports and required decisions keep it working. **The first three meetings, then the rhythm that keeps a council alive** Beneath the title a key reads: How to read this: (practitioner) — the practitioner’s account quoted in the article; (this publication’s reading) — the article’s own analysis. A pattern, not a measurement. Three boxes run left to right, joined by arrows. **Meeting 1: Organisation, roles and introductions**. What the council is for; why it has these members; which decisions it will make, and by what criteria; introductions. About half an hour (practitioner). **Meeting 2: Current-state review**. Documents out as early as possible, so members arrive having read them, with questions (practitioner). **Meeting 3: Proposed changes and first decisions**. Documents out as early as possible. The council signs off, or does not, on proposals put to it (practitioner). An arrow from the third meeting drops into a full-width box outlined in amber, headed **After that: the running rhythm**: Regularly scheduled reports it is required to receive, and decisions it is required to make (practitioner). Each decision recorded with its date and owner (this publication’s reading). Without work, it atrophies and is re-absorbed (practitioner). (The last sentence is set in grey.) ## Give it something to decide Asked what kept people attending once the novelty wore off, I gave the unflattering answer first: > What kept people attending after the novelty wore off? Recurring meetings and inertia, mostly. The only thing I would suggest adding is regularly scheduled reports that are required and decisions that need to be made. Without work, councils, like muscles, atrophy. Keep them busy and the council will persist—starve it and it will be re-absorbed. The first sentence is the uncomfortable one, and it should be. Inertia is a real force, and a council can run on it for a long time, but a council running on inertia is already the status meeting the next section is about; it has simply not been noticed yet. Steven Rogelberg and colleagues found that "perceived meeting effectiveness was found to have a strong, direct relationship" with employees' job attitudes and well-being [\[15\]](#ref-15). My reading is that this fits the worry here: whether a meeting feels worth having is tied to how people feel about their work, and a council kept alive by habit may keep its attendance while losing its members' goodwill. The study did not test councils or attendance, so that last step goes beyond its finding. The recommendation in the rest of the answer is the remedy, and it comes in two parts that do different jobs. Required reports keep the council informed whether or not anything is wrong. For example, a quality scorecard, a register of open exceptions, the list of standards due for review: these arrive on a schedule because the council is owed them, not because someone thought of raising them. They are the council's evidence. On their own they do not keep it alive, because a council that only receives reports is a reading group. Required decisions are what keep it alive. A standard proposed by a domain and waiting for sign-off, an exception to a policy that only the council can grant, a dispute between two groups that neither can settle alone: each is a decision with a date by which it must be made, and each is a reason for the people who can make it to be in the room. Bain's screen for committees is the same test applied from the top: "A committee should exist only if it plays an explicit role in decisions that have a material impact" [\[13\]](#ref-13). The same brief describes an insurer's executive committee that spent "40% of its time to 'informing' members of new developments and only 20% considering decisions" [\[13\]](#ref-13). That committee spent twice as long informing its members as deciding. ![Required reports, proposed changes and disputes feed a council of decision makers, which produces decisions that the person with enforcement power acts on; with nothing to decide it atrophies, and when groups dig in it breaks down.](https://wi.senterprises.com/assets/diagrams/FIG015_MDM_What_Feeds_A_Council_v1_0.png) Figure 2: Three kinds of work feed a council; its output is decisions, enforced by whoever holds the authority. Beside it, the two ways it fails: starved of decisions, or stuck because its groups have dug in. **What feeds a council, what starves it, and where it breaks** Beneath the title a key reads: How to read this: (practitioner) — the practitioner’s account quoted in the article; (this publication’s reading) — the article’s own analysis. A pattern, not a measurement. Across the top, three boxes side by side: **Required reports**: Regularly scheduled and required (practitioner). Owed to the council whether or not anything is wrong (this publication’s reading). **Proposed changes**: Put to it for sign-off, which councils are useful for (practitioner). **Disputes between groups**: Questions the groups cannot settle between themselves (this publication’s reading). Their arrows join into one that points down to a box in the centre, outlined in amber, headed **The council**: The decision makers or their proxies, with a few technical advisers (practitioner). It works through relationships and alliances: the consensus view prevails (practitioner). Arrows lead from the council to a box on each side. To its left, a box headed **Starved: nothing to decide**: With nothing to decide it drifts into status updates (this publication’s reading); without work it atrophies and is re-absorbed (practitioner). To its right, a box headed **Stuck: groups dig in**: Opposing groups dig in and the council breaks down, even with a tiebreaker (practitioner). An arrow from the council points down to a box headed **Decisions**: Its deliverable (practitioner). Recorded, dated and owned (this publication’s reading). An arrow from it points right to a box headed **Enforcement**: Rests with whoever holds the authority and the power to enforce (practitioner). My reading is that the two failures in that figure are different, and that confusing them leads to the wrong repair. A starved council has nothing to decide; adding members, a better charter or a more senior chair does not help it, because the missing ingredient is work. A stuck council has plenty to decide and cannot, because the groups in it have dug in; giving it more decisions only adds to the pile. The first needs a supply of decisions. The second needs the person who can enforce one. ## How to stop it turning into a status meeting > How do you stop a council turning into a status meeting? You give it work. It needs to have a requirement to generate decisions as that’s their ‘Deliverable.’ In order to do that, they need things to Decide first. "They need things to Decide first" is the half of that answer easiest to skip. A council cannot generate its own supply of decisions without becoming the standard-setting body it is bad at being. The decisions have to come to it: proposals from the domains, exceptions to policies, disputes between groups, standards that have come up for review. My reading is that a new council's first queue is the backlog it was created to clear: the open definition disputes, policy exceptions and standards awaiting sign-off that existing teams have not been able to settle. Whoever chairs the council collects that backlog before the first meeting and lists it under the decisions the council is expected to make; after that, nothing reaches the agenda without a named owner and a date by which it must be decided. My reading, again, is that the supply after that comes from a rule, not on request. Name the changes that cannot proceed without the council's sign-off, and have the executive sponsor say so at the first meeting. The rule governs what reaches the council; the relationships and alliances in the room govern what it decides. The rule holds only while the sponsor's weight is behind it, so a thinning queue is a sponsor problem before it is a council problem. If no sponsor will make sign-off a gate, do not form a decision council at all: call it an advisory forum, because a council with nothing it is required to decide is the status meeting this section is about. And if a sponsor's weight lapses after the council exists, go back to the sponsor first; if the gate is not restored, convert the council to an advisory forum rather than let it keep meeting as a decision body. My reading is that decisions as the deliverable give a test anyone can apply to an agenda. Read each item and ask what the council will have decided when the item closes. An item that ends in "noted" is status. An item that ends in "approved", "declined" or "sent back with these conditions" is governance. A council whose agenda is mostly the first kind has already become a status meeting, whatever its charter says. De Smet's rule is shorter: "If it's a decision meeting and you don't have a decision to be made just cancel it" [\[10\]](#ref-10). Senior teams can lose strategic time to operational concerns. One second-hand report of Michael Mankins's survey of 187 companies, in Stanford Social Innovation Review, found that executives "spent an average of 21 hours each month in team leadership meetings," "of which 18 hours were consumed by operational concerns, leaving only three hours for important strategic issues" [\[16\]](#ref-16). My reading is that if an executive team, whose members own budgets and run businesses, drifts toward operational concerns, a council whose members attend as a side duty may drift faster. My reading is that three habits follow from the test. Write every agenda item as the question the council will answer. Send the status in the pre-read, so the meeting spends no time on it. And record every decision with its date, the option chosen and who carries it out, so the council can see its own output. The Data Governance Institute's framework treats the same record as part of governance itself: knowing "WHO made WHAT decisions, and when" [\[2\]](#ref-2). A council that cannot point to what it decided last quarter will struggle to argue it should meet next quarter. ## The case against councils Three serious objections deserve an answer, because each would change the build if it were right. The first is that councils are the wrong vehicle altogether. Zhamak Dehghani's data mesh describes how "traditional data governance attempts to achieve that through centralization of decision making," and replaces it with "A decision making model led by the federation of domain data product owners and data platform product owners," backed by "automated execution of decisions by the platform" [\[17\]](#ref-17). Vendors of automated governance make a sharper version of the case. Atlan, which sells metadata and governance software and so has a stake in the answer, describes a council with no accountable owner as "a committee with a charter and no consequence for ignoring it," and names "governance theater," where "councils produce policy nobody enforces," as a recurring pattern [\[18\]](#ref-18). I agree with the diagnosis and not with the remedy. A federation of domain product owners deciding global rules is a council by another name, made of decision makers, which is what this article argues for. Automated execution is enforcement, which is what a council without a decider lacks. Neither removes the need for someone to decide what the platform should enforce. Seiner's reply to the "governance is dead" argument makes the point without mentioning councils at all: governance is "the execution and enforcement of authority over the definition, production, and use of data," and so "Data governance cannot die; it's just not possible" [\[19\]](#ref-19). The second objection is aimed at consensus. Bain's RAPID model gives every decision a single "D," "the single point of accountability who commits the organization to action" [\[7\]](#ref-7), and Bezos describes Amazon's norm of "disagree and commit," in which a leader with conviction but no consensus asks the room, "Look, I know we disagree on this but will you gamble with me on it? Disagree and commit?" [\[20\]](#ref-20). Both say a council should not need consensus to act. This is the objection that comes closest to the claim, and it does not refute it, for a reason the objectors' own evidence supplies. "Disagree and commit" is a norm, and a norm is a relationship: it works where the dissenter trusts the person asking. RAPID's single decider is the person with enforcement power under another name. And McKinsey's survey of decision making found that "quickly committing to big-bet decisions without reaching consensus among leaders does not beget success," and that "having only clearly defined roles, accountability, or decision rights is not sufficient to win" [\[21\]](#ref-21). The first quotation says consensus matters for the decisions that matter. The second says decision rights alone, the formal account, are not enough. My claim was never that councils ought to run on consensus. It is that the ones that work do, and the evidence from the people arguing against consensus fits that description better than it fits their prescription. The third objection is that soft skills are not enough, and that council members need real technical depth. The evidence for this is a qualification, not a refutation: leaders need enough data literacy to consume analysis [\[9\]](#ref-9), and Gartner's priorities for data leaders include data literacy alongside the leadership behaviours [\[6\]](#ref-6). In the governance literature and the Gartner, McKinsey and Bain material I read, I found no source arguing that council members need deep technical skill more than they need the ability to work a room. The advisers supply the depth; the members need enough of it to know when the advisers disagree. One further piece of evidence cuts against the word "re-absorbed," and it is worth taking seriously. Herbert Kaufman's study of United States federal executive-branch bureaus found that of 175 in his 1923 sample, "No less than 148 of them (nearly 85 percent) were still going in 1973" [\[22\]](#ref-22). Formal bodies, once created, tend to persist. Kaufman called his own method "illustrative, not probative," and federal bureaus have statutes and budgets that internal councils lack, so this is an analogy rather than a study of councils. But it suggests that a starved council does not always disappear. It may simply keep meeting, hollow, as a status meeting that nobody has the standing to cancel. That is not a refutation of "re-absorbed"; it is the form re-absorption often takes. The work goes back to the organisation the council was carved out of, and the meeting stays on the calendar. ## Something to decide The usual question about a governance council is how to design it: the charter, the membership, the decision-rights matrix. Those matter less than two things the charter cannot supply. The room has to hold people who can decide, and the council has to be owed decisions it is required to make. My reading is that if the first three meetings go right, the council can make its first decision at the third. Keep the decisions coming and it has a better reason than habit to be meeting next year. Stop sending them and it is likely to drift into status updates and then, quietly, be absorbed back into whatever it was carved out of. ## References \[1\] Gartner, Inc., "Gartner Predicts 80% of D&A Governance Initiatives Will Fail by 2027, Due to a Lack of a Real or Manufactured Crisis," press release, Stamford, Conn., 28 February 2024 (Saul Judah and Anurag Raj quoted). Read in full 2026-10-25\. An analyst prediction; the release states no method behind the figure. Cited for the sentence quoted. \[2\] The Data Governance Institute, "Framework Component #6: Decision Rights," datagovernance.com (no publication date shown; page last modified 15 May 2024). Read in full 2026-10-25\. Cited for the decision-rights question and the record of who made which decisions. \[3\] Robert S. Seiner, "Non-Invasive Framework for Data Governance Implementation — Part 1: Levels and Components," TDAN.com (no date shown). Read in full 2026-10-25\. Cited for the strategic level as the data governance council, its role as ultimate decision maker for unresolved issues, and its membership. \[4\] Jeffrey Pfeffer, *Managing with Power: Politics and Influence in Organizations* (Harvard Business School Press, 1993, as the cited page dates it). Book not read; publisher's description only, on the Stanford Graduate School of Business faculty page, read 2026-10-25\. Cited for the description's phrase. \[5\] Gartner, Inc., "Gartner Survey Reveals Less Than Half of Data and Analytics Teams Effectively Provide Value to the Organization," press release, 21 March 2023\. Read in full 2026-10-25\. Cited for the sentence on roadblocks. \[6\] Gartner, Inc., "Gartner Identifies Top Three Priorities of Effective Chief Data & Analytics Officers," press release, Stamford, Conn., 19 January 2023\. Read in full 2026-10-25\. Cited for the leadership behaviours quoted and, in the closing objections, for data literacy among the priorities. \[7\] Paul Rogers and Marcia Blenko, "Who Has the D? How Clear Decision Roles Enhance Organizational Performance," *Harvard Business Review*, January 2006, as published by Bain & Company, 28 June 2006\. Read in full 2026-10-25 on Bain's page. Cited for the sentence on consensus and the definition of the D role. \[8\] John P. Kotter, "Leading Change: Why Transformation Efforts Fail," *Harvard Business Review*, March–April 1995\. Read in full 2026-10-25 in a course copy hosted by UCSF Family and Community Medicine (hbr.org could not be fetched). Cited for the two sentences on the guiding coalition (Error #2). \[9\] MIT Sloan School of Management, "What is data literacy?," Ideas Made to Matter (no author or date shown). Read in full 2026-10-25\. Cited for the sentence on leaders and the phrase from Rama Ramakrishnan. \[10\] Aaron De Smet and Leigh Weiss, "To unlock better decision making, plan better meetings," McKinsey & Company, 9 November 2020 (interview). Read in full 2026-10-25\. Cited for the two sentences quoted. \[11\] Joel Brockner and Batia M. Wiesenfeld, "An Integrative Framework for Explaining Reactions to Decisions: Interactive Effects of Outcomes and Procedures," *Psychological Bulletin* 120, no. 2 (1996): 189–208\. Abstract only, read 2026-10-25 at the copy Columbia Business School hosts; the 45 samples are as the abstract states them. Cited for the finding quoted. \[12\] Michael C. Mankins and Richard Steele, "Stop Making Plans; Start Making Decisions," *Harvard Business Review*, January 2006, as summarised on Bain & Company's page. Read in part 2026-10-25: Bain's page carries the summary and recommendations, not the full article. Cited for the recommendation quoted. \[13\] James Hadley and Jenny Davis-Peccoud, "Committees that work," Bain & Company, Decision Insights brief, 8 August 2012\. Read in full 2026-10-25\. Cited for circulating documents in advance, the screen for whether a committee should exist and the insurer's executive committee. \[14\] Jeffrey P. Bezos, "2017 Letter to Shareholders," Amazon, published 18 April 2018\. Read in full 2026-10-25\. Cited for the six-page memos and the silent reading at the start of meetings. \[15\] Steven G. Rogelberg, Desmond J. Leach, Peter B. Warr and Jennifer L. Burnfield, "'Not Another Meeting!' Are Meeting Time Demands Related to Employee Well-Being?," *Journal of Applied Psychology* 91, no. 1 (2006): 86–96\. Abstract and introduction read 2026-10-25; the body was not. Cited for the abstract's finding on perceived meeting effectiveness. \[16\] Ivor Holloway, "How to Waste Time and Lose Sight of Priorities," *Stanford Social Innovation Review*, Spring 2005\. Read in full 2026-10-25\. A second-hand report of Michael C. Mankins's survey of 187 companies with Marakon Associates and the Economist Intelligence Unit, published as "Stop Wasting Valuable Time," *Harvard Business Review*, September 2004, which was not read. Cited for the figures as the review reports them. \[17\] Zhamak Dehghani, "Data Mesh Principles and Logical Architecture," martinfowler.com, 3 December 2020\. Read in full 2026-10-25\. Cited for the three phrases quoted from the section on federated computational governance. \[18\] Emily Winks, "Why Do Data Governance Implementations Fail?," Atlan, 19 August 2026\. Read in full 2026-10-25\. Vendor content: Atlan sells metadata and governance software. Cited for the two phrases quoted. \[19\] Bob Seiner, "Data Governance is Not Dead — It's Not Possible," Solutions Review, 11 April 2024\. Read in full 2026-10-25\. The author's own argument, on a contributor outlet. Cited for his definition of governance and the sentence quoted. \[20\] Jeffrey P. Bezos, "2016 Letter to Shareholders," Amazon, published 17 April 2017\. Read in full 2026-10-25\. Cited for "disagree and commit". \[21\] Iskandar Aminov, Aaron De Smet, Gregor Jost and David Mendelsohn, "Decision making in the age of urgency," McKinsey & Company, 30 April 2019\. Read in full 2026-10-25\. Cited for the two findings quoted. \[22\] Herbert Kaufman, *Are Government Organizations Immortal?* (Washington, D.C.: The Brookings Institution, 1976). Read in part 2026-10-25 (the front matter and the chapter with the survival data; the concluding chapter was not reached), in a scanned copy hosted at gwern.net. Cited for the survival figure (p. 34) and his description of his method.