A lighter page is worth the lost functionality when the feature you cut is non-essential and its speed or experience cost is genuinely hurting rankings or users, and it is not worth it when the functionality is core to the page’s value or its conversions. The pivot is non-essential-and-costly versus core-to-value. And before you cut anything core, the right step is to optimize it, because a lighter page only helps when you are shedding weight that was not earning its keep.
Cutting is the right call when a feature costs more than it returns. Heavy widgets, third-party embeds, autoplaying media, chat tools, sliders, and similar add-ons often load significant scripts while contributing little to why the reader is on the page. When something like that is dragging load time, hurting Core Web Vitals, or cluttering the experience without serving the content’s purpose, removing or deferring it makes the page faster and better without losing anything the visitor actually needed. Non-essential plus costly is the clear signal to drop it.
Cutting is the wrong call when the functionality is the value. An interactive calculator on a tool page, a working configurator, a booking flow, a key demo, or any feature central to what the reader came to do or to how the page converts cannot be stripped for speed without making the page fail at its job. A blazing-fast page that no longer does the thing people visited for is not a win, it is a faster way to lose them. When a feature is core to value or conversion, the heaviness is a problem to solve, not a reason to remove it.
The practical move is to cut the non-essential features whose cost outweighs their value and optimize the core ones rather than dropping them. Go through what the page loads, separate the elements that genuinely serve the reader or drive conversions from the ones that merely add weight, and remove or defer the latter. For the features you must keep, make them lighter through compression, lazy-loading, and trimming the scripts behind them. Optimize before you sacrifice, and only sacrifice what the page did not need.