Making MVP tests useful

Everyone knows you should have an MVP to validate your idea. However, most builders find that the well-intentioned test is full of noise. So, when should you trust your MVP results? - At the survey stage? - At the mockup stage? - At the barely functional app stage? I found my answer with: https://twitter.com/founderYonz/status/1619761186949693440 "Build a slice across, instead of one layer at a time." Instead of focusing on trying to spend the smallest amount of effort in the wrong direction, treat all the stages as useful but inconclusive info until you build a slice across the entire product pyramid. Maybe we should call the slice minimum loveable product. Share your thoughts below.