The First 90 Days as a Data Leader: What to Decide, Defer and Leave Alone
A new data leader is asked for a strategy before they know what they have inherited. The first quarter is better spent learning where authority sits, settling the few things that cannot wait and declining to promise the rest.
A newly appointed head of data or chief data officer arrives to three things at once: an estate nobody has fully described, colleagues with different ideas of what the role is for, and an expectation that a strategy will follow within weeks. Each pulls towards early commitment. The argument of this article is that the first ninety days should produce few decisions, chosen carefully, and that being clear about what is being deferred is part of the job.
What follows sorts the quarter’s choices into three groups: what to settle, what to put off and what to leave as it is. It ends with how to present the result to an executive team. The reasoning is DII’s, supported where indicated by published guidance written for and by government data leaders. It describes no individual’s experience.
Learn before deciding
The first month is for finding out, and the order matters: people before systems.
Start with what colleagues need. The Federal Chief Data Officers Council’s playbook, which is advice from serving chief data officers to new ones, begins with relationships. It recommends a stakeholder analysis “starting with leadership”, followed by interviews “to identify their mission needs and challenges”1. Its reasoning is practical: a leader’s engagement “should focus on forging relationships before zeroing in on data governance activities that may risk appearing solely compliance driven”1. A round of conversations in which the new leader mostly listens will show which problems the organisation believes it has. That is the list the role will be judged against.
Find out where authority sits. A data leader’s formal position may not match the reach the role needs. US federal guidance sets the standard explicitly for its own chief data officers: they “shall serve in a central leadership position, with visibility into relevant agency operations”, placed highly enough to engage regularly with the rest of the leadership2. The same guidance is realistic about the limits. The officer must have the necessary authority, and yet “some of these responsibilities are shared with others in the agency”2. A new leader needs to know, early, which decisions are theirs, which are shared and which belong to someone else entirely. The relationship with the chief information officer is the one to understand first; the playbook asks new leaders to develop it “even where there is no direct reporting relationship”1.
Read the estate for what is missing. A complete inventory is not achievable in a month and is not needed. Three readings are. The first is who owns what: an ownership register, if one exists, shows by its gaps where nobody answers for data, a point developed in how to assign data owners, stewards and custodians. The second is what is trusted: which reports the executive team relies on and which it argues about, the subject of what makes business intelligence trustworthy. The third is what the effort goes on. The UK National Audit Office found in 2019 that “people do not monitor the time or costs involved in sorting poor-quality, disorganised data”, and that some departments put the share of time spent cleansing data at between 60% and 80%3. Those figures are departments’ own estimates and describe government in 2019. The useful point for a new leader is the first half of the finding: the cost of working around poor data is unlikely to have been counted, so asking for it will be informative.
Take a quick reading of maturity. A formal assessment belongs later. An informal one belongs now. The US Federal Data Strategy’s governance playbook starts the same way, advising agencies to “discuss and informally assess the current state of data and related infrastructure maturity” before choosing a model for a fuller exercise4. The Data Maturity Quick Scorecard serves that purpose. What a fuller assessment can and cannot establish is covered in what to assess in a data strategy maturity assessment and how to act on the result.
Learn how money moves. The playbook tells data leaders to understand the organisation’s budget process so that resources can be planned into future budgets1. A leader who learns in month five that next year’s bids closed in month three has lost a year.
What to decide
A small number of decisions cannot wait, because leaving them open shapes everything else by default.
Decide what the role is for, and say it. The playbook is direct about the risk of leaving this vague. Chief data officers who do not use data as a strategic asset face “the risk of becoming a compliance driven position”1. Within the first weeks the leader should be able to state, in two or three sentences, what the role exists to change and what it does not cover, and should test that statement with the person they report to. Disagreement at this point is cheap. Discovered a year later it is costly.
Decide which one or two problems to solve first. The same guidance holds that new leaders “must demonstrate early value generation”1. The choice of problem matters more than its size. A good first problem has a business owner who wants it solved, data that is within reach, and a result that can be shown inside the quarter or shortly after. It should also leave something behind that will be used again: an agreed definition, a cleaned dataset, a working relationship. The playbook advises weighing “the commitment, focus, and priorities of senior leadership” when choosing1. A problem the executive team does not care about will earn no credit however well it is solved.
Decide what happens to inherited commitments. There will be projects under way, promises made by a predecessor and purchases half-completed. Each needs one of three answers: continue, pause for review, or stop. Deciding nothing is itself a decision to continue, taken without the information. The reviews can be brief. The point is that the new leader has chosen and has said so to those affected.
Decide who to rely on. Before any restructuring, identify the handful of people inside and outside the team whose judgement is sound and who know how the organisation works. This is a private decision and it shapes the quality of everything learned afterwards.
What to defer
The pressure in the first quarter is to produce large things quickly. Several should wait, and the leader should say that they are waiting and why.
The strategy document. A strategy written before the leader understands the organisation’s priorities, its budget and its appetite will be either generic or wrong. What can be offered at ninety days is a statement of direction and a date for the strategy. What a strategy has to settle, and why those choices need other people’s agreement, is set out in what a data strategy is and the decisions it has to make.
Reorganising the team. Structure should follow the work, and the work is not yet defined. Changing reporting lines early also spends goodwill that will be needed later. The exception is a vacancy or a role conflict that is stopping delivery now.
Platform and tooling choices. A new leader will be shown a procurement that is nearly complete, or asked to endorse a platform direction. Unless a contractual deadline forces the question, these should wait until the uses they must support are known. Tools chosen before the requirement is understood constrain the strategy that follows.
An organisation-wide governance launch. The playbook lists this among its pitfalls in plain terms: “Do not just start with data governance for the sake of data governance, partner on key questions to drive value”1. Governance is necessary. Launched before anyone has seen a benefit, it is received as control.
A full maturity assessment. It will be more useful once there is a strategy to measure against and an owner for the result. The UK guidance for government departments recommends that a first assessment be aimed at one strategically important area and not the whole organisation5.
Deferral only works if it is explicit. A list of what will not be decided this quarter, with the reason and the date each will be revisited, is one of the most useful documents a new leader can produce. It prevents the silence being read as indecision.
What to leave alone
Some things should not be touched in the first quarter at all.
What works. Every organisation has data practices that function without attention: a reconciliation that a finance team has run reliably for years, a reporting cycle people trust. They may be inelegant. They are not the priority, and disturbing them creates opponents.
Other people’s territory, until invited. Security, privacy, records management and technology have their own leaders and their own obligations. The federal guidance acknowledges that a data leader’s responsibilities are shared and should be carried out in consultation with those officials2. An early attempt to redraw those boundaries will be read as a claim on someone else’s job.
The predecessor’s reputation. Criticising what was inherited is tempting and gains nothing. The people who built it may still be in the building.
Arguments about definitions that no decision depends on. There will be long-running disputes about terms. Unless one is blocking a decision that matters this quarter, it can stay unresolved a little longer.
Setting expectations with the executive team
At or near day ninety, the leader owes the executive team a short account. It is not the strategy. It has four parts.
- What was found. A candid reading of the estate and of how data is used, including what is better than expected. Three or four findings, each with an example.
- What will be done in the next two quarters. The one or two problems chosen, who the business owner is for each, and what will be shown by when.
- What is being deferred. The list above, with dates, and in particular when the strategy will be brought forward.
- What is needed. The specific decisions, people or funds the leader requires from the executive team, stated as requests.
The fourth part is the easiest to leave out, and it is the test of whether the role has been given the means to succeed. If the requests are refused, the leader has learned something important early. Why data strategies stall after approval describes what follows when that conversation is postponed until after a strategy has been signed off.
The quarter at a glance
The first ninety days: aims and outputs by period
| Period | Aim | What should exist at the end |
|---|---|---|
| Weeks 1 to 4 | Learn: people, authority, estate, money | Interview notes; a map of who decides what; a quick maturity reading; the budget calendar |
| Weeks 5 to 8 | Decide the few things that cannot wait | A statement of the role agreed with the line manager; one or two first problems with business owners; a disposition for each inherited commitment |
| Weeks 9 to 13 | Show early progress and set expectations | First visible result or a dated plan for it; the list of deferred decisions; the ninety-day account to the executive team |
The periods are a guide and will overlap. A leader joining during a budget round or a crisis will have to compress the first of them.
The Data Leadership 90-Day Plan sets these periods out as a working plan with interview prompts and a template for the ninety-day account. The Chief Data Officer Checklist is a shorter reference for the responsibilities of the role itself.
Sources
- Federal Chief Data Officers Council (United States Government). CDO Playbook: Advancing the Federal Data Strategy. 2021.
- Office of Management and Budget (Executive Office of the President, United States). Memorandum M-19-23: Phase 1 Implementation of the Foundations for Evidence-Based Policymaking Act of 2018: Learning Agendas, Personnel, and Planning Guidance. 10 July 2019.
- National Audit Office (UK). Challenges in using data across government (HC 2220). 21 June 2019.
- Federal Data Strategy (United States Government). Federal Data Strategy Data Governance Playbook. July 2020.
- Central Digital and Data Office (UK Government). Delivering the Data Maturity Assessment: objectives and resources. 27 March 2023.
This article is general professional information. It is not legal or regulatory advice.