Most of Fivetran's customers are small to medium businesses. I sat down with Tina Wang, Director of Product at Fivetran, and the numbers surprised me. Fivetran started as a data ingestion tool and has grown into a full data platform with transformations, reverse ETL, and managed data lakes, backed by 750+ connectors. All of these features were built with the needs of small teams in mind. Here is the pattern Tina described. Companies start with CSV exports, spreadsheets, or querying production databases directly. It works until it does not. Data engineers end up spending most of their time patching fragile pipelines instead of building anything new. Tina called it death by a thousand cuts. The result is unreliable data and slow decisions. AI is making this worse. As AI agents and non-technical teams start pulling their own insights, bad data can create confident wrong answers at scale. Garbage in, garbage out now moves faster. On cost, Fivetran's argument is simple. The sticker price might be higher than DIY or open source. But total cost of ownership drops once you account for engineering time and the cost of downtime when pipelines break. Pricing is consumption based, tied to data volume synced, and they recently lowered prices for the SMB segment specifically. They also give you 14 days free, and the clock only starts after your first sync completes. The part I found most important: Fivetran leans into open formats like Iceberg and Delta Lake for their managed data lakes. No lock-in. You can bring new tools or AI capabilities later without a painful migration. The infrastructure conversation used to be about moving data. Now it is about whether your data can keep up with how fast AI wants to use it. Use Fivetran. #data #ai #fivetran #etl #theravitshow
Most of Fivetran's customers are small to medium businesses. I sat down with Tina Wang, Director of Product at Fivetran, and the numbers surprised me.
Fivetran started as a data ingestion tool and has grown into a full data platform with transformations, reverse ETL, and managed data lakes, backed by 750+ connectors. All of these features were built with the needs of small teams in mind.
Here is the pattern Tina described. Companies start with CSV exports, spreadsheets, or querying production databases directly. It works until it does not. Data engineers end up spending most of their time patching fragile pipelines instead of building anything new. Tina called it death by a thousand cuts. The result is unreliable data and slow decisions.
AI is making this worse. As AI agents and non-technical teams start pulling their own insights, bad data can create confident wrong answers at scale. Garbage in, garbage out now moves faster.
On cost, Fivetran's argument is simple. The sticker price might be higher than DIY or open source. But total cost of ownership drops once you account for engineering time and the cost of downtime when pipelines break. Pricing is consumption based, tied to data volume synced, and they recently lowered prices for the SMB segment specifically.
They also give you 14 days free, and the clock only starts after your first sync completes.
The part I found most important: Fivetran leans into open formats like Iceberg and Delta Lake for their managed data lakes. No lock-in. You can bring new tools or AI capabilities later without a painful migration.
The infrastructure conversation used to be about moving data. Now it is about whether your data can keep up with how fast AI wants to use it. Use Fivetran.
#data #ai #fivetran #etl #theravitshow