Filter custom deals by asking: will this integration generalize to 5+ future customers?
Custom deals and integrations can either become the blueprint for your next customer tier or hamstring your roadmap for years. The key question: does this custom work solve a problem multiple future customers will face, or is it a one-off that only benefits this deal? For Assembled, Robinhood wanted custom enterprise features. Ryan wrote a long doc to convince the team it was worth it because those features would generalize to the next tier of multi-100k customers—and they did. But a big airline wanted Microsoft Dynamics and Outlook integrations. Ryan declined because their core customers used Zendesk and Slack—that work wouldn't generalize. The filter: only say yes if the custom work stretches you toward your next customer segment. If it pulls you sideways into a different market, it's a trap. Custom work should be an investment in your future product, not a consulting project.
When to use
When evaluating large custom deals or integration requests. When customers say 'we'll pay extra if you build X.' When deciding between a big deal in an adjacent market vs smaller deals in your core. When your team debates whether custom work will hamstring the roadmap. When transitioning from 10 to 50+ customers and professionalizing your deal evaluation.
Don't do this
Taking every custom deal because revenue is revenue. Building for whoever pays the most without asking if it generalizes. Saying yes to custom work in markets you don't plan to serve. Not documenting the generalization logic (Assembled wrote a long doc to convince themselves). Thinking all custom work is bad—some custom work is the right strategic bet. Letting sales drive product strategy without founder/product filter.
1 Founder Who Did This
Turned down a big airline deal requiring Microsoft Dynamics and Outlook integration because it wouldn't generalize to 5+ future customers