The biggest mistake we made building Dashblock
June 13th, 2022
Dashblock started focusing on replacing CSS selectors with machine learning to make scraping resilient to website updates. Our first product was a chrome extension called Datap, which got us into Y Combinator.
During the batch, we talked to hundreds of prospects & users, launched a desktop app on Product Hunt and started to receive a huge, constant flow of inbound prospects that wanted to automate websites.
We were eager for growth, so we onboarded everybody to build a generic tool solving any web automation need, which resulted in a lot of issues:
- Marketing: unclear value prop leading to poor conversion rates ;
- Sales: highly custom and not replicable sales process ;
- User experience: huge knowledge gap creating lack of autonomy on users-side, which limits/avoids (up)selling.
- Teamwork: difficulty in focusing on product development vs. customizing it for every customer ;
- Tech: implementation of very different feature requests, resulting in a complex and hard-to-maintain codebase ;
- Growth: linear, ultimately limited for one of the reasons above.
It is worth mentioning that YC continuously pushed us to focus on a specific use case. However, listening was hard because it required the courage to stop growing the way we did, choose a specific subset of users (+ abandon others), and iterate only with them before finally replicating sales to grow sustainably.
Today, I see two paths (not mutually incompatible btw) that could have benefited Dashblock development:
- Devtool positioning: target developers with a strong bottom-up go-to-market (e.g., open-source) and make it as pleasant as possible for them to automate their needs with Dashblock. Sell enterprises licenses, support, and/or a SaaS version ;
- Tackling a use case: target a group of users with a specific need you solve through your product (e.g., think Plaid for gathering banking data). Look for the most efficient selling methods (outbound, inbound, channel, etc.).
We will never know if either of those plans would have allowed Dashblock to be alive by default, but statistics tell chances would have been higher!