Data-driven study 2026: 4 in 10 companies now analyze data, but the gap by company size is widening

39.85% of companies in the European Union with 10 or more employees analyzed data in 2025, either with their own staff or through an external provider, according to Eurostat. Two years earlier, the figure was 33.25%. The progress is real but uneven: among large companies it reaches 82.03%, and among small ones, 35.06%.

Being a data-driven company means making decisions based on measured evidence, not just intuition or experience. In this study we look at how many companies do it, what data they use, who does the analysis and what research says about the results.

39.85%

EU companies that analyze data

with 10 or more employees, 2025

82.03%

Large companies that analyze data

250 or more employees, 2025

47 points

Gap between large and small firms

it was 43 points in 2023

30%

Data-driven US plants in 2010

up from 11% in 2005

Source: Eurostat, 2025; Brynjolfsson and McElheran, 2016

View the data as a table

EU companies that analyze data	39.85% | with 10 or more employees, 2025
Large companies that analyze data	82.03% | 250 or more employees, 2025
Gap between large and small firms	47 points | it was 43 points in 2023
Data-driven US plants in 2010	30% | up from 11% in 2005

Methodology

This study combines three public sources, each with its own definition of "data-driven":

Source	What it measures	Coverage	Period
Eurostat, survey on ICT usage in enterprises	Companies that perform data analytics (in-house or external) and which data they analyze	EU companies with 10 or more employees, excluding the financial sector	2023 and 2025
Brynjolfsson and McElheran, US Census Bureau	Plants that make data-driven decisions, under a strict definition	Representative sample of US manufacturing plants	2005 and 2010
Stanford HAI, AI Index 2026	Organizations that use artificial intelligence	Organizations surveyed worldwide	2023 to 2025

The figures are not directly comparable: they show the same phenomenon from different angles. Differences and gaps were calculated by InfoByData from the published data.

Finding 1: data analytics is growing, especially in large companies

Companies of every size advanced between 2023 and 2025, but large ones started higher and kept rising faster. The gap between large and small companies went from 43.03 to 46.97 percentage points.

• 2023
• 2025

Small (10 to 49)28.78%35.06%

Medium (50 to 249)51.43%59.84%

Large (250 or more)71.81%82.03%

Total33.25%39.85%

0%50%100%

Source: Eurostat, isoc_eb_das

View the data as a table

	2023	2025
Small (10 to 49)	28.78%	35.06%
Medium (50 to 249)	51.43%	59.84%
Large (250 or more)	71.81%	82.03%
Total	33.25%	39.85%

It's the same pattern we saw in our study on AI in business: in Latin America, a large company is four times more likely to use AI than a small one. Data and AI advance together, and small businesses fall behind on both.

Finding 2: sales and customer data are the most analyzed

Companies mainly analyze what they already record in their daily operations: transactions and customer information. External sources, such as government open data or satellite data, are still a minority.

Transactions (sales, payments)26.19%

Customers (purchases, preferences, reviews)17.62%

Social media11.44%

Web data10.75%

Government open data7.28%

Location of devices or vehicles7.06%

Sensors or smart devices5.92%

Satellite data3.58%

Source: Eurostat, 2025

View the data as a table

Transactions (sales, payments)	26.19%
Customers (purchases, preferences, reviews)	17.62%
Social media	11.44%
Web data	10.75%
Government open data	7.28%
Location of devices or vehicles	7.06%
Sensors or smart devices	5.92%
Satellite data	3.58%

For a marketing team, that translates into metrics such as customer acquisition cost, conversion rate or customer lifetime value, which we explain in our guide to marketing KPIs. For an SEO team, web data is what makes it possible to measure the real effect of changes such as Google's AI Overviews, which cut clicks from 15% to 8% of visits, according to Pew Research Center.

Finding 3: most companies analyze data in-house

At every size, more companies analyze data with their own staff than with external providers. The difference is largest among large companies, which have more resources to hire specialized talent.

• Own staff
• External provider

Small27.86%12.68%

Medium54.47%17.64%

Large78.84%29.17%

Source: Eurostat, 2025

View the data as a table

	Own staff	External provider
Small	27.86%	12.68%
Medium	54.47%	17.64%
Large	78.84%	29.17%

A company can use both options, so the percentages don't add up to the total.

Finding 4: data-driven decisions are linked to higher productivity

The most cited evidence on impact comes from the US Census Bureau. Economists Erik Brynjolfsson and Kristina McElheran analyzed a representative sample of manufacturing plants and found that data-driven decision making nearly tripled, from 11% to 30%, between 2005 and 2010, according to their article in the American Economic Review.

In a companion Census study covering more than 18,000 plants, the same authors found that value added was about 3% higher in plants that adopted these practices than in those that didn't. The improvement appears after adoption and not before, which the authors consider consistent with a causal effect.

Their definition of "data-driven" is demanding: a plant had to have data available, use it to make decisions, track 10 or more key indicators and set both short- and long-term targets. The study also found that the benefit is greater when the company already has technology and skilled workers.

Finding 5: AI raises the bar

Artificial intelligence depends on well-organized data to work. Stanford's AI Index 2026 reports that 88% of surveyed organizations used AI in 2025, up from 55% in 2023. By contrast, only 19.95% of European companies used AI that year, according to Eurostat. That distance shows that trying AI is one thing and building it into business decisions is another.

Public data is also raw material. In our article on crypto in numbers, every price comes from CoinGecko's open API: an example of how an analysis can be built entirely from public sources.

Limitations of the study

• There is no comparable statistic for Chile or Latin America. Eurostat only covers the European Union, and we did not find a regional survey that measures data analytics with the same methodology.
• Definitions vary by source. For Eurostat, it's enough for a company to analyze data; the Census study requires systematic use in decision making.
• The productivity study covers US manufacturing from 2005 to 2010. Its results are a reference, not a prediction for other sectors or countries.

How to move toward a data-driven company

1. Start with the data you already have: sales, payments and customers are the sources companies use most, and they don't require buying external information.
2. Define a few indicators and a target for each. The Census study required 10 or more KPIs with short- and long-term targets; a small business can start with five, like the ones in our marketing KPI guide.
3. Decide who will do the analysis: in-house staff is the most common option, but an external provider can help you get started.
4. Measure before and after every change. It's the same logic as our on-page SEO best practices: each change is evaluated with click and ranking data.
5. Organize your data before adding AI: without clean, accessible data, AI tools deliver less.