Ask HN: Roast my B2B SaaS idea? Productionising data warehouses
Hey all, excited to pitch you Garam, my B2B SaaS start-up idea. Looking forward to some roasting here :)
Product
What is it: an out-of-the-box data warehouse. In other words, a single “enterprise-glue” platform that contains ETL/ELT, a data warehouse and a semantic layer specifically tailored to B2B SaaS companies (eg., includes templated code for MRR, ARR, subscriptions, sales pipelines, etc).
Use case: Garam simplifies the data journey for start-ups. At the moment, start-ups need to pay for an expensive data team and a fragmented data stack. It takes ages to build a data warehouse. Worst of all, the data modelling only starts then, and the chaotic nature of this work results in inaccuracies.
Most of the code written by data engineers is similar across business models, and thus can and should be productionised. This templated semantic layer is the USP of Garam.
Who would want it: B2B SaaS start-ups that have scaled to the point that they cannot rely on spreadsheets for their data infrastructure anymore.
Product analysis
*Competitors:*The fact that the direct competitors have launched only recently suggests two things to me: the problem shows validation and the timing could not be better for me to launch. Happy to be challenged on this.
A few recently-launched start-ups offer something similar (eg., pliable.co, agiledata.io, arch.dev). However, they focus on no-code interfaces or they cater to data consultancies. To me, these are solutions that will not scale for start-ups, the same way that a no-code website builder hits its limits very soon.
Traditional data stack tools which are too technical, need to be “glued together” and offer no modelling (cube.dev, DBT, etc.)
Data consultancies (their offer does not scale and is a lot more expensive).
Key difference: two things: enterprise glue and templated metrics. In addition, this semantic layer comes with built-in data quality checks and automated documentation (once again, in 99% of companies order IDs should not be duplicate or NULL. This is business logic that can be templated; data engineers should not replicate this work across companies).
Customer conversion strategy: probably SaaS events, startup accelerators, etc. This is something to work out once the MVP is ready. Advice welcome.
Stage: designing the MVP, very early stages.
Why me?: 7+ years of experience as a head of analytics (incl. leading +20 data engineers and data analysts) at both start-ups and scale-ups. I have led several data modelling initiatives, built full-fledged data warehouses from scratch, etc. These experiences have helped me realise that there is a gap in the market, and I want to contribute to productionising this process.