Modern data stack

The modern data stack is a set of separate software tools, each built for one job, that a company's data team joins together to copy, store, clean and analyse the company's data. A data team is made up of data engineers, who build the systems that move data, and analysts, who use the data to answer questions. The other way to buy the same capabilities is one integration suite: a single product, from one vendor, that does all of these jobs. The word "stack" means the layers of tools, one on top of another.

Here is how a modern data stack works in general. Data connectors copy the data out of each app the company uses, such as its order system, its payment system and its customer-support tool. They load the raw data into a cloud data warehouse. A transformation tool then cleans the raw tables, fixes their formats and joins them, inside the warehouse. This step is data transformation, the T in ELT. A data catalogue lists every table and what each column means. Observability software checks that each day's data arrived on time and in full. Finally, a dashboard tool shows the answers to managers. Each of these tools is a separate product, usually rented by monthly subscription.

Why did data teams start buying this way? Around ten years ago, cloud warehouses made it cheap to store raw data and to run heavy queries on it. So teams no longer had to clean the data before loading it. They could load everything first, and pick the best tool for each later step. A small team could start with two or three tools, pay little, and add more as it grew. The cost came later. Each tool brought its own contract, its own bill and its own login. Someone had to join the tools and keep them working together. When one tool changed, the steps that depended on it could break.

The pieces are now coming back together. In 2026, the makers of two of the best-known pieces, a connector tool and a transformation tool, merged into one company. Larger technology companies have bought several other makers. So the stack is starting to look like a suite again.

Take, for example, a small company that sells handmade soaps online. Its one data engineer sets up an ELT pipeline from the shop's website and payment app into a cloud warehouse, and adds a dashboard tool, all within a week.

Consider a large insurer with hundreds of systems. It may prefer one suite that covers integration, data quality, the catalogue and master data management, so that one vendor is responsible for all of it.

Say a chart in a company's Monday meeting shows a number that is clearly wrong. With many separate tools, the data catalogue shows which tables the chart was built from, so the team can find which tool failed.

How is the modern data stack different from a data lakehouse platform? A lakehouse platform is one vendor's system that stores the data, transforms it and now also trains AI models on it, all in one place. It is closer to the suite. In practice, the line is blurring. Many companies now keep a warehouse or a lakehouse at the centre, and connect a few separate tools around it.