Over the last four years we have sat on about twenty panels for first-data-leader hires at mid-market clients. Twelve have worked out. Eight have not. The pattern of the failures is consistent enough to write down.
The two common archetypes and their failure modes
The enterprise veteran
Twenty years at a bank or a big retailer. CIO-adjacent background. Presents beautifully. Talks fluently about transformation, target operating models, and the importance of culture. Arrives and proposes a programme that would take two years and cost £8m. Is frustrated by the mid-market's inability to fund it. Leaves at 18 months.
The rising specialist
Very strong technical background. Senior data engineer or senior data scientist at a tech-forward firm. Gets the hiring bar right, gets the first platform right, gets the first model right. Struggles with the exec conversation. Cannot turn a board's hand-wave into a quarterly plan. The work is strong but not visible. Leaves at 24 months, often amicably, for a deeper technical role elsewhere.
What the successful hires had in common
- Mid-market scar tissue. They had operated in an organisation of 500–10,000 staff before. They knew the cadence, the politics, the budget reality.
- Technical credibility without technical self-indulgence. They could read a dbt project and interview a senior engineer; they did not insist on writing the code.
- An opinion on sequencing, delivered in two pages. Within six weeks, they had a one-page strategy and a one-page sequencing plan. Not a strategy document — a two-pager.
- Comfort with "no". They declined use cases that were not ready, turned down tools that did not fit, and said so to the board.
- Boundary around the role. They did not try to own analytics, BI, data engineering, ML, and AI ethics as a single monolith. They built the spine and let the adjacent functions (finance analytics, customer analytics, marketing analytics) stay where they sat.
The title does not matter; the scope does
CDO, Head of Data, Director of Data & AI, VP Data — the title is immaterial. The scope and reporting line are not. The pattern that works at mid-market:
- Reports into CFO, COO or CTO, depending on the organisation's centre of gravity
- Owns: platform, governance, core data engineering, analytics engineering, and AI capability
- Partners with: finance, sales, operations, customer-service analytics teams
- Does not own: domain BI teams in the business (they stay in the business)
Trying to own everything leads to a hire who spends their days arbitrating rather than delivering.
What to interview for
Technical interview
Not a whiteboard SQL test. A technical conversation: read a real dbt project or a real data-platform architecture, critique it, explain what they would change and why. Ask about a past programme; probe until you hit the weakest part. If they will not talk about the failures, they have not had any useful ones.
Exec interview
Present them with the actual problem the organisation has. Not a hypothetical; the real one. Ask for a two-page plan within a week. Read the plan. Ask what they would stop doing. The answer to that question is diagnostic.
Board interview
One hour with two non-exec directors, unrehearsed. Can they talk about risk appetite, AI governance, and TCFD reporting without jargon? If not, they will not land the board-level conversation when they arrive.
Compensation honesty
Mid-market clients regularly try to hire this role for £120–160k. The market floor for the profile above is, as of 2026, £180–250k base in the UK outside London, more in London, plus a meaningful equity or bonus component. Hiring below market is where a lot of the failures come from; the pool is good, but not at £120k.
The transition plan
Do not hand the new hire a blank page. On day one, they should inherit: a one-page current-state from the outgoing leader or consultancy; a list of the three most pressing decisions; a budget envelope; an introduction to the five most important stakeholders. The first ninety days are theirs to reshape, but they should not have to excavate.