How a Data Analytics Company Runs With Zero Data Scientists

A data analytics company with no data scientists sounds like a restaurant with no cooks. But TextQL, a $17 million agentic data analytics platform, operates with roughly 40 people and zero dedicated data roles. No data scientists. No data analysts. No data team at all.

Mark Hay, CTO and co-founder of TextQL — which serves enterprise customers including Amazon, Dropbox, and the NBA — didn’t arrive at this structure by accident. It’s a deliberate bet on where the data profession is heading.

“The interesting thing is we don’t actually have a data team,” Hay says. “It’s a shared responsibility.”

What “Shared Responsibility” Looks Like in Practice

At TextQL, data analytics isn’t a department — it’s a capability distributed across the entire company. Sales uses the product to track pipeline. Marketing runs campaign analysis. Finance builds reports. Engineers contribute to the shared ontology as they ship features.

The company uses its own product as the canvas for running the business. When the team launched a new feature they called “the feed,” everyone — not just a data team — used TextQL’s analytics to track adoption, identify usage patterns, and decide what to build next.

This works because the product itself eliminates the traditional bottleneck. In most companies, a business user has a question, submits a request to the data team, waits for an analyst to write a query, and gets results days later. TextQL’s platform lets anyone ask the question directly, in plain English, and get an answer immediately.

The Ratio Problem This Solves

The traditional model has a structural flaw: demand for data analysis always outstrips supply. One former data team leader (who later evaluated TextQL as a customer) described the ratio at her startup as one data scientist for every six product managers. Every data professional was permanently backlogged.

Hay describes the data scientist’s traditional role using a prospecting metaphor: someone who uses technology to find “million-dollar decisions” buried in the data. The problem is that the search is expensive — each analyst can only dig in so many places. Most companies have orders of magnitude more questions than their data team can answer.

The TextQL model doesn’t eliminate the skill of asking good questions. It distributes the tool to answer them. The people closest to the business problem — the salesperson, the marketer, the finance lead — ask their own questions and get their own answers.

What This Means for Data Roles

Hay is careful not to declare data roles extinct. His prediction is more nuanced: the role evolves from answering questions directly to nurturing the systems that answer questions at scale.

Instead of writing queries all day, data professionals build and maintain the ontology — the shared definitions, relationships, and business logic that make AI-generated queries accurate. They shift from being the bottleneck to being the quality layer.

At TextQL, engineers add to the shared ontology as they ship features. There’s no separate data governance process. The ontology grows incrementally, driven by actual usage rather than top-down planning.

The implication for hiring is real. Companies building data teams today might consider whether they need ten analysts writing queries, or two data engineers building systems that let everyone else write their own.

FAQ

Can a company run without a data analytics team?

TextQL operates with approximately 40 people and zero dedicated data scientists or analysts. Data analytics is distributed as a shared responsibility across sales, marketing, finance, and engineering. This works because their AI platform eliminates the query-writing bottleneck — anyone can ask business questions in plain English and get immediate answers.

How does TextQL use its own product internally?

TextQL uses its own analytics platform as the primary tool for running the business. When they launched a new feature called “the feed,” the entire company — not just engineers — used the product to track adoption and identify usage patterns. Engineers contribute to the shared ontology as they ship features.

What is the typical ratio of data scientists to business users?

Traditional organizations often have ratios of one data scientist for every five to ten business stakeholders. One startup reported a ratio of one data scientist for every six product managers, creating a permanent backlog of unanswered questions. AI-powered analytics platforms aim to collapse this ratio by letting business users query data directly.

Are data scientist roles disappearing because of AI?

Data scientist roles are evolving, not disappearing. The shift is from answering individual questions (writing SQL queries, building dashboards) to nurturing the systems that answer questions at scale — building ontologies, defining business logic, and maintaining data quality. Fewer analysts are needed for query work, but the systems design role grows.

How do you build a data ontology incrementally?

Instead of defining every column and metric upfront, build the ontology through usage. Start with no definitions. When a query fails or returns incorrect results, add the specific semantic definition that fixes it. Over time, the ontology grows to cover the most-used patterns. TextQL’s approach is to solidify what works and let the system repeat it.

What does a flat data team structure look like?

In a flat data structure, analytics responsibility is shared across functional teams rather than centralized in a dedicated department. Business users ask their own questions using AI tools. Engineers maintain the shared ontology as they ship features. Data governance happens incrementally through usage rather than through a separate planning process.

How does AI change data team hiring decisions?

Companies may need fewer query-writing analysts and more data engineers who build systems. The hiring question shifts from “can you write SQL?” to “can you design the ontology that makes AI-generated queries accurate?” TextQL’s 40-person company with zero data roles represents one extreme — most enterprises will likely land somewhere between traditional teams and fully distributed analytics.

What is the biggest bottleneck in enterprise data analytics?

The bottleneck is human throughput: data teams receive more questions than they can answer. Every analyst writes one query at a time. AI-powered platforms remove this bottleneck by letting business users query data directly. The remaining bottleneck shifts from human bandwidth to compute capacity — infrastructure that can handle AI agents generating 100-1,000x more queries than humans.