Guides · Govern

IT steering committee: terms of reference that make it decide

Most IT steering committees listen to updates and defer decisions. This guide shows how to design one that decides: a written mandate, thresholds, papers that pass six tests, an agenda that puts decisions first, and minutes that record who does what by when.

Why most IT steering committees disappoint

Most IT steering committees become status meetings. Each project presents an update, papers arrive the night before with one option, and the decision that matters is deferred to next month. Members stop coming, send delegates, and the committee loses the authority it was set up to have. Real decisions then move to corridors and one-to-one meetings, where nobody records them.

The cause is rarely the people. It is the design: no written mandate, no thresholds, no rules for papers, and an agenda that follows the projects instead of the decisions. All four can be fixed in a quarter.

The test of a good committee is simple: at the end of each meeting, can the secretary list the decisions taken, with an owner and a date for each? If the honest answer is "mostly updates", the committee needs a reset, not more members.

The mandate: what it decides, and what it does not

Write the mandate as a list of decisions, not a list of topics. "Oversees the IT portfolio" invites updates; "approves investments between two thresholds" invites decisions.

The committee decides:

  • Technology investments between the lower and upper threshold, and investments below the lower threshold that fall outside an approved budget.
  • The ranking of discretionary demand against the year's capacity and budget.
  • Stopping, pausing or re-planning projects that breach their tolerances.
  • What happens after go-live when benefits fall short: continue, re-scope or stop.
  • Time-bound risk exceptions, within the tolerance the executive committee has delegated to it.
  • Technology policies and standards that apply inside IT and to projects.

It does not decide:

  • Investments below the lower threshold within an approved budget: the budget owner decides, with the CIO.
  • Investments above the upper threshold, risks beyond its delegated tolerance, and policies that apply to all employees: it recommends them to the executive committee or board.
  • Operational matters handled in service reviews, and personnel decisions inside the IT organisation.

Align every amount with your company's delegation of authority. A committee that approves something the delegation reserves for the board creates a governance problem, not a faster decision.

Terms of reference: the nine sections

Terms of reference should fit on two pages. Each section answers a question members will otherwise argue about in the room.

SectionWhat to write
1. PurposeOne sentence: why the committee exists and what it is accountable for.
2. MandateThe decisions it takes, the thresholds, and what it recommends upward.
3. MembershipRoles, not names: chair, vice-chair, voting members, attendees without a vote, secretary. Rules on delegates.
4. QuorumHow many voting members must be present, and who must be among them.
5. Urgent decisionsHow a decision is taken between meetings, by whom, and how it is ratified at the next meeting.
6. Meetings and papersFrequency, length, paper deadline, pack deadline, minutes deadline, the paper template.
7. Risk and complianceWhen legal, compliance or security must give an opinion before a decision.
8. ReportingWhat goes to the executive committee or board after each meeting, and how.
9. ReviewWhen the committee reviews its own effectiveness, and who approves changes to these terms.

Membership and quorum

  • Chair a business executive, often the chief operating officer or chief financial officer. A committee chaired by the CIO is seen as IT's meeting; one chaired by the business is the company's.
  • Keep the voting members to six to nine: the heads of the largest business units, finance, HR or operations as relevant, and the CIO. Security and portfolio management attend without a vote.
  • Give the secretary real authority: usually the head of portfolio or the PMO. The secretary returns papers that are not ready.
  • Make the quorum mean something: the chair or vice-chair attending, finance present or represented by a named deputy, and more members attending themselves than sending a delegate.
  • Allow named delegates only, who can commit their unit. A delegate who has to "take it back" turns every decision into a deferral.

Decision rights and thresholds

Thresholds move most of the decisions out of the meeting and keep the committee's time for the ones that need it. An illustrative set, in thousands of your currency:

DecisionWho decidesTrigger
Investment below 250Budget owner, with the CIOWithin approved budget; outside it, the steering committee
Investment from 250 to 1,000Steering committeeAny new commitment in the band
Investment above 1,000Executive committee or board, on the committee's recommendationAny new commitment above the band
Project re-planSteering committeeCost forecast 10% or more over baseline, or go-live 60 days or more late
Early warningPortfolio report, no decisionCost 5% over, or 30 days late
Risk exceptionSteering committee, time-boundWithin the tolerance delegated by the executive committee; beyond it, escalated

Amounts are whole-life cost: one-off plus recurring costs over the contract or asset life, so that subscriptions and phased projects are not judged on their first year. Splitting a project to stay under a threshold is not allowed. Set the amounts against your own size and delegation of authority. The principle matters more than the numbers: every decision has one place where it is taken, and everybody knows which.

Papers that let the committee decide

Most deferred decisions trace back to the paper. A paper should fit in four pages and pass six tests before it is tabled. The secretary returns any paper that fails one.

TestPassed when
Decision askedThe first line says what the committee is asked to decide.
Sponsor signedThe business sponsor has read and signed the paper.
OptionsAt least two real options, including doing nothing or doing less.
Cost and benefitFor each option, with the confidence of the estimate.
RisksThe main risks of each option, with the opinion of legal and compliance where regulation applies.
RecommendationOne option is recommended, with the reason.

Fix the deadlines and keep them: papers to the secretary seven days before the meeting, the pack to members six days before, minutes within four days. The secretary tests each paper on arrival and returns any that fail the same day, so the sponsor has a chance to fix it. A pack that arrives late is not read, and a committee that has not read the papers cannot decide.

The agenda: decisions first

Put decisions at the top of the agenda, when attention is highest, and updates at the end, or in the pack only. A 120-minute meeting might run like this:

TimeItem
10 minActions from the last meeting: done, late, at risk
60 minDecisions requested, one paper at a time, each closed with a recorded decision
25 minPortfolio exceptions and benefit reviews: only projects outside tolerance or short of their benefits, each with the decision asked
15 minRisk exceptions, forward plan and next meeting
10 minAnything new, and how the meeting went

Track the share of meeting time spent on decisions. A committee that spends less than half its time deciding is still a status meeting.

Minutes and follow-through

Minutes record decisions, not discussion: the decision, its conditions, the owner, the date and any dissent. Keep two logs across meetings: a decision log and an action log. Open each meeting with the actions that are late. A committee that never checks its own decisions teaches the organisation that its decisions do not matter.

Review the committee every year

After the first three meetings of a reset, take ten minutes to ask what is working. Then, once a year, ask members to rate the committee on a short survey (does it decide, are papers ready, is the time well used) and look at the facts: attendance, pack and minutes on time, the share of time on decisions, decisions deferred, actions late. Agree two or three changes, and update the terms of reference.

Questions

How often should an IT steering committee meet?

Monthly works for most mid-size and large companies, with no meeting in the quietest month. Quarterly is too slow for portfolio exceptions; decisions then move outside the committee. Fortnightly meetings usually signal that thresholds are too low.

Should the CIO chair the IT steering committee?

Usually not. A business executive in the chair makes the committee the company's forum for technology choices rather than IT's meeting, and puts the CIO in the role of proposing and explaining. The CIO remains a voting member.

What is the difference between an IT steering committee and a project board?

A project board steers one project inside the mandate it was given. The steering committee decides across the portfolio: which projects start, which stop, and how money and capacity are shared. Project boards escalate to it when they breach their tolerances.

What if members keep sending delegates?

Allow only named delegates who can commit their unit, count delegates separately in the quorum, and report attendance to the chair. Persistent delegation is usually a sign the meeting is not worth attending: fix the agenda first.

Toolkits on this topic

Related