We stopped buying SaaS at 1752vc and started building it instead
by•
We are a venture firm, not a software company. But when we need a tool now, the first question is not what to buy, it is whether we can just build it ourselves. A few subscriptions we used to renew without thinking are gone.
Curious what others are seeing. Have you cancelled anything because you built the replacement, and did it hold up?
33 views
Replies
As I leave a comment, I'm building us something and canceling a subscription, lol
Don't forget the cost of maintaining and supporting the app.
@cecil_pang Very true! So far still cheaper
we build a custom GSC/GA4/Google ADs pipeline using mcp connections through Windsor.AI. it feeds into our marketing and SEO data. not completely custom but tools like crawlraven.com/ still dont allow for longer than. a90 day pull, their mcp server is. not operational yet we were tired of jumping between interfaces to get similar data.
Self Build is definitely easier than before and we will continue to build our our dashboards. mailchimp, attio and automatic googlesheet update web hooks are in place already. for tracker building etc.
I think the first question a venture firm asks is whether it can make money by backing the company's growth. But have you actually compared the engineering cost of building, maintaining, and continuously improving it with the cost of buying an existing product?
If we're talking about polished data visualizations, like those built in Lovable, sure. But if we're talking about mission-critical infrastructure with KPIs such as deliverability and click-through rates that directly affect revenue, I'm much more skeptical.
Can you name any widely used self-built replacements for Google Calendar, project management tools, or financial management software? I experiment with AI extensively, but it's still far from being a jack of all trades.
@lucasjpols - We’re seeing the same shift. Building v1 has become incredibly cheap; owning v17 where it gets expensive. Our rule is to build when the workflow is core and unusually specific, buy when it’s standard. Otherwise, every cancelled subscription risks becoming a permanent side project 😅
I am curious about that tradeoff at 1752vc. Do you factor engineering time and long-term ownership into the build-vs-buy decision?