Build vs Buy decision framework for non-core capabilities

Our team keeps oscillating between building custom solutions and buying off-the-shelf for capabilities adjacent to our core product (auth, observability, billing). Last quarter we built a feature-flag system that a $50/mo tool would've covered. The argument was 'we need full control' but 6 months later it's barely customized. Looking for a structured decision framework: what criteria have you used to draw the line between build and buy, and how do you handle the sunk-cost trap once the build decision is made?