Why a fractional data expert beats your first data hire

By Luca De Angelis · 8 min read

A founder recently told me: "We've been trying to hire our first data person for five months. Meanwhile, we're making pricing decisions off a spreadsheet nobody trusts." That sentence is why the fractional data expert model exists.

I run Satyadata as exactly that - a senior freelance data consultant working with a small number of scale-ups at a time. So yes, I have a horse in this race. Which is precisely why this comparison needs to be honest: the fractional model is a genuinely better first move for many companies, and genuinely the wrong one for some. Let's do the maths and the caveats properly.

The problem with the first full-time data hire

The classic playbook says: hit product-market fit, then hire a data analyst or "Head of Data" as employee number 20-something. In practice, three things go wrong more often than founders expect.

The hiring cycle is brutal. Across the major tech hubs - Berlin, Amsterdam, London, Zurich - I regularly see 3–5 months from opening a data role to a signed contract, plus the candidate's notice period (one month in the UK, up to three or six in Germany and Switzerland), plus 2–3 months of ramp-up. Call it half a year, often closer to nine months, before the first useful deliverable. Your pricing decisions don't wait.

Seniors rarely take the job. A genuinely senior data person - someone who has built stacks, made modelling calls, and survived their own mistakes - usually won't join as a lone first hire with no team and an undefined mandate. So the first hire is typically mid-level. Capable, but now making one-way-door decisions alone: warehouse choice, event taxonomy, metric definitions, tool contracts. These are exactly the decisions that are expensive to reverse two years later.

There isn't a full-time job yet. In the first year, most scale-ups need bursts of architecture and modelling work, then steady but modest maintenance. A full-time hire either drowns in ad-hoc requests or invents complexity to fill the week. I've inherited more than one over-engineered stack that existed mainly because someone had five days a week to build it.

The cost and speed maths, honestly

Numbers below are realistic ranges for major tech hubs as of 2026 - the lower end reflects Berlin and Amsterdam, the upper end London and Zurich - and they are deliberately conservative in my own disfavour where there's doubt.

	First full-time hire (mid-level)	Fractional data expert (2 days/week)
Time until someone starts	3–7 months (search + notice period)	1–3 weeks
Time to meaningful delivery	8–10 months (start plus 1–3 months ramp-up)	2–4 weeks - the first weeks are typically audit and orientation, not output
Base cost, year one	€70–130k salary depending on city	€75–130k in fees (day rate €800–1,400, ~95 days)
Employer add-ons	+15–25% social and health contributions, recruiter fee (15–25% of salary), equipment, tooling	None - the rate is the cost
Total costs year one	≈ €105–190k	≈ €75–130k
Months of decisions taken without data support	8–10 months of the first year	Under 1 month
Paid absence	25–30 holiday days, public holidays and sick leave, all paid	None - you pay for days delivered
Share of time on the actual work	Diluted by all-hands, performance cycles, internal projects, onboarding others	High - engagement is scoped to outcomes, not attendance
Availability	5 days a week, daily ownership	2 days a week
Depth of company context	Accumulates continuously; in every meeting	Partial by design - needs written context and clear scoping to work
Experience level	Typically 3–6 years - seniors rarely accept a lone first-hire role	10+ years, typically with exposure to all analytics areas (marketing, product, finance, business, supply)
Knowledge retention	Stays in-house - until they leave	Stays in-house if documentation and handover are delivered at the end of the engagement
Commitment and exit	Permanent contract, notice periods, morale cost of a mis-hire	Monthly notice, scope adjusts up or down

Bold marks the stronger option in each row. The employee wins three that matter: five days a week of availability, company context that accumulates in every meeting rather than having to be handed over, and knowledge that stays in-house by default. A fractional also costs more per working day - divide the fees by the days.

On the other side, three rows show most clearly what you get from engaging a fractional data expert.

The expensive number isn't in the euro columns. It's the row counting months of decisions taken without data support. Those months have a cost, it just never appears on an invoice: acquisition budget split across channels on gut feel, a pricing change nobody could evaluate, a churn problem noticed a quarter after it started, a board deck assembled by hand at midnight. I've deliberately left that out of the euro totals, because any number I put on it would be invented. But a founder who has lived through those months rarely needs convincing that it dwarfs the difference between the two columns.

You pay for days delivered, not days employed. An employee's 25–30 holiday days plus public holidays and sick leave come out of your budget whether or not work happens - roughly 15% of the year, before anyone falls ill. A fractional invoices for the days worked. If they are on holiday or sick, you simply don't have to pay them. That difference is invisible in a salary comparison and very visible in a cash flow.

A fractional doesn't need to be entertained. Employees, correctly, come with all-hands, performance cycles, career conversations, internal initiatives and the general gravity of a growing organisation - all legitimate, all consuming a real share of the week. An external senior arrives with a scoped mandate, works on the two or three things that matter most this quarter, and leaves the internal calendar alone. Two to three focused days a week routinely beat four to five diluted ones. That's a structural difference in what each arrangement is optimised for.

The overall claim stays narrow: for the first 12–18 months of a company's data journey, you get more senior judgment, sooner, for comparable total money, with an exit ramp instead of a severance conversation. A fractional still spends the first weeks understanding your business; the difference is that they've seen the pattern before, so orientation takes weeks rather than months.

You're not buying hours. You're buying the thousand mistakes someone already made at other companies, so they don't get made again on your warehouse bill.

It isn't only about the first hire

The first-data-hire case is the one people know, but most of our engagements start somewhere else entirely. Three situations come up again and again.

Parental leave and other planned absences

Someone central to your data function goes on parental leave for eight or twelve months - the analytics lead, the senior data analyst who owns half the reporting, tracking or experimentation, the one analytics engineer who understands the pipelines. A permanent replacement isn't an option, and few senior people leave a stable job for a one-year contract. The work doesn't pause either: the board still wants numbers, pipelines still break.

A fractional fits this shape almost perfectly, because the constraint is time-boxed by definition. We cover leave at two or three days a week by embedding in the team. The reporting and the ad-hoc insights keep running, one or two projects that were going to slip get picked up, and the work stays inside the team's own repos and conventions rather than in a parallel consultant workspace. When the person returns, there is nothing to hand over from outside - just a documented stack and colleagues who already know what changed. The person coming back finds their job intact rather than a backlog and a pile of undocumented changes. The same logic applies to sabbaticals, long-term sick leave, and the gap after a senior resignation.

Scaling a team: the gap between headcount approved and headcount working

This is the one most teams underestimate. You get three roles signed off - an analytics engineer, a senior analyst, a data scientist. Excellent news, and now the maths from the table above: three to seven months before each person starts, eight to ten before they deliver meaningfully - and the three searches compete for the same hiring manager's attention, so they run in sequence more often than in parallel. Realistically the third hire is contributing well over a year after the budget was approved, and every new joiner needs onboarding from a team that is already underwater.

Budget approved is not capacity delivered. In that gap, the same money that sits unspent in an open req can buy fractional senior capacity that both delivers now and prepares the ground: foundations built properly, conventions written down, an onboarding path that means each new hire is productive in weeks rather than months. The fractional isn't competing with those roles - they're being paid out of the delay the roles create anyway.

A project that needs a specialist your team doesn't have

Sometimes the need is narrow and genuinely temporary: a customer lifetime value model, a migration off a legacy warehouse, an experimentation platform, a first AI workflow. Hiring for it makes no sense - the specialism is needed for one quarter, not forever - but neither does having a capable generalist team learn it slowly and expensively on the job. Bring in someone who has done it several times, have them build it with your team rather than at them, and leave the capability behind.

Why one senior person now ships what a small team used to

There's a second shift that makes the fractional model work far better in 2026 than it did in 2019: AI-assisted tooling. My day-to-day runs on agentic workflows - Claude Code-style agents that scaffold dbt models, draft and test SQL against real schemas, write documentation, and keep analysis tickets updated while I make the judgment calls. I've written about this setup in detail in how I work as an AI-enhanced data analyst.

The honest framing: AI doesn't replace the senior judgment - deciding what to model, which metric definition survives contact with finance, when a number is too good to be true. But it compresses the labour around that judgment enormously. The boilerplate that used to justify a junior hire (pipeline scaffolding, test coverage, documentation, first-draft queries) is now largely automated. One experienced person with good tooling covers ground that took a two-to-three-person team a few years ago.

What an engagement actually looks like

A fintech scale-up (~170 people) brought us in at two days a week after four months of failed hiring. Weeks 1–2: audit and a decision memo on the stack. Weeks 3–8: warehouse and dbt foundation, a KPI tree the leadership team actually signed off, and board reporting cut from three days of spreadsheet surgery to a two-hour refresh. Weeks 9–16: the first genuinely strategic piece - a customer lifetime value model that reshaped how they set acquisition budgets per channel.

Around month five, we wrote the job spec for their first in-house analyst together, sat in the final interviews, and handed over a documented, tested stack instead of a pile of tribal knowledge.

When a fractional data expert is the wrong choice

Hiring a fractional isn't always the best choice, so here is the honest take. Don't hire one if:

• You need operational ownership every day. If pipelines feeding customer-facing features break at 2am and someone must own that pager, you need an employee (realistically, two). Fractional means days per week, not on-call coverage.
• You're ready to build a team of three or more. Past a certain scale you need full-time people in the seats: a leader doing hiring, coaching and politics five days a week, and senior technical owners living in the stack every day. A fractional senior can hold either seat while you recruit - keeping the work moving and the standards high - but the goal is to hand it over, and perhaps to join again later when the team needs extra power.
• You want a culture-carrier. Someone in every stand-up, every offsite, absorbing context osmotically. That's a legitimate need and it's a full-time property. Two days a week buys focus, not omnipresence.
• The real problem is disagreement, not data. If leadership can't agree what the company is optimising for, an outside expert becomes a weapon in someone's argument. Sort the strategy first - or scope the engagement explicitly as helping you sort it.

And one warning about the middle ground: a fractional engagement fails when it's treated as a cheap employee - pulled into every meeting, assigned tickets, managed by calendar. It works very well when it's scoped by outcomes: a stack that runs, metrics that are trusted, a model that changes a budget decision, a hire that lands well.

The decision in one paragraph

If you need daily ownership, a culture-carrier in every stand-up, or a team you can realistically staff right now - hire full-time. The fractional case is specific: you need senior data expertise this quarter rather than next year, or you have a gap with a known shape - a leave to cover, roles that will take months to fill, a specialism needed for a few months. In those situations you get senior output sooner, for roughly the total cost of a mid-level hire, embedded in your team rather than parked beside it, and with a clean exit either way.

Whichever you choose, choose it deliberately.