You Don't Need a Data Team. You Need a Data Decision

Most founders think they need a data engineer. The real gap is three decisions that determine whether any hire or tool will produce anything useful.

You post the job description before you know what you'd actually give the person to do. Not because you're careless — because the problem feels technical, and technical problems feel like they need technical people. But the research on why small business data projects fail points somewhere else entirely.

The University of the West of Scotland's SME governance study found that fragmented accountability and ad hoc rules were the direct cause of limited analytics use in small firms. Not missing engineers. Missing decisions about who owns what and which numbers to trust.

The hire arrives and the clock starts

First Round Review's startup hiring guide documents what happens next: early data hires spend their first months fixing broken logging and setting up foundational infrastructure. Months. On work that should have been decided before the offer letter went out. The hire is not building models or surfacing insights. They're doing remediation on a situation the founder created by not making three prior decisions.

Those decisions are not technical. They look like business questions because they are business questions.

Decision one: what is the data actually for?

Not "we want to be data-driven." Which specific business decision do you make every week or month where better numbers would change the outcome? Pricing? Retention? Which acquisition channel to fund? The AWS SME data governance guide sequences this first, before role definitions, before tool selection. If you cannot name the decision, you cannot evaluate whether any data setup is working.

Decision two: which source of numbers do you trust?

Most founders at Series A have Stripe, a CRM, Google Analytics, and a spreadsheet someone built two years ago, and all four show different revenue figures. The AMCIS review of BI adoption barriers found that decision-maker distrust of data reliability was the primary reason BI tools went unused in small firms. Not cost. Not complexity. Distrust. You need to pick one source for each critical metric and tell everyone that's the number. Not the best source. Just one agreed source.

Decision three: who owns each data asset?

The Learning Gate maturity model for SME data governance shows that firms with rich data and no explicit ownership stay stuck at the lowest governance stage regardless of what tools they add. Ownership means someone's name is attached to the CRM data, the transaction data, the product event data. When it breaks or goes stale, that person is accountable. A data hire cannot assign that accountability. Only you can.

When the hire is supposed to answer the question the founder never asked

A reasonable counterargument holds that a skilled first data hire will surface these decisions faster than a non-technical founder can, and with less risk of specifying them wrongly. Choosing a source of truth requires knowing which data sources are structurally reliable. Assigning ownership requires understanding what ownership means when pipelines break. These look like technical questions.

They're not, and here's where the counterargument breaks down.

First Round Review describes data hires spending months on broken logging. That is a no-decision environment, not a wrong-decision environment. The practitioner data warehouse guide that documents the six-month path to ROI for SMEs requires KPI mapping and CEO and CFO involvement before any technical setup begins. Not alongside it. Before. The Taylor & Francis meta-analysis of sixty-three BI adoption studies lists top-management support and organizational readiness as determinants that precede technical factors. Organizational readiness is a leadership condition. A data hire cannot manufacture it by showing up.

The hybrid view in the research is worth taking seriously: a first data hire helps founders articulate these decisions, but that person succeeds only when leadership already holds rough answers. Rough answers. Not technically precise governance documentation. A working position on what the data is for, which number to trust, and who owns what.

What readiness actually looks like

You don't need a data warehouse to make these decisions. You need a forty-five minute conversation with your co-founder or CFO where you write down one business decision the data should serve, pick one revenue source as the number, and put someone's name next to each major data asset.

That conversation is the prerequisite. Not a hire. Not a tool. When a data person arrives into an environment where those answers exist, even imperfectly, the Learning Gate model shows governance maturity advances. Without them, it doesn't matter who you hire.