Fivetran's moat is data delivery
Diving deeper into
Conor McCarter, co-founder of Prequel, on Fivetran's existential risk
there are just use cases where the analytics suites and the in-app dashboards just won't work no matter how good the embedded tool is
Analyzed 5 sources
Reviewing context
The real moat is not the dashboard, it is getting raw product data into the customer’s own stack, where it can be combined with everything else the business knows. A SaaS app can show built in charts for its own data, but it breaks down when a team wants to join payments to ad spend, support tickets, CRM records, and finance models, then run a custom query in SQL or dbt using its own definitions.
-
Embedded analytics is good at the common questions a vendor can standardize, like usage, campaign results, or support volumes. It is weak at one off company logic, where each customer defines revenue, attribution, churn, or account health differently and needs warehouse level control.
-
This is why the modern data stack unbundled. Teams buy Fivetran to move data, dbt to reshape it, and BI tools to explore it, because the people who maintain pipelines are often different from the people defining business metrics and answering ad hoc questions.
-
For SaaS vendors, native warehouse exports still matter even if in app reporting is strong. They raise retention and can become a paid SKU, often priced as part of enterprise plans or as 10% to 30% of ACV, because customers still ask for the full dataset when dashboards hit their limits.
The next step is not replacing warehouses with better embedded BI. It is making warehouse access table stakes for serious SaaS products, then layering cleaner metrics, faster syncs, and better observability on top so customers can keep using their own tools while vendors still own the data delivery experience.