It's true that some startups just randomly try other ideas when the first one doesn't work - and some of those incorrectly call that flailing a "pivot." However, that isn't the actual concept behind pivoting.
Pivots are a thing because early stage startups are often pursuing an opportunity space they are betting is just emerging. Because of this there's often little data to go on when they start and they're making educated guesses about a bunch of things from product features, primary customer, go to market and business model. The concept is that a startup should very quickly develop an MVP which enables engaging with actual customers. That engagement often reveals which initial assumptions were incorrect. Hopefully they were incorrect not because they made a dumb mistake but because the correct answer was unknowable without testing.
That testing also sometimes reveals completely new customer and/or market information about a bigger opportunity in that area leading to a pivot. This is akin to sinking temporary test wells before building an oil derrick and pipeline. Even the best geology analysis can only give a probability of oil being found in a given location. Sometimes sinking test wells reveals the oil is actually a quarter mile to the East. That isn't really a pivot but more like product/market tuning. A surprising number of times, the process of sinking those test wells to discover oil leads to discovering something different but just as good like natural gas or something even better like gold. That's a canonical pivot. Experienced founders and VCs can pretty easily tell the difference between flailing and a pivot because the pivot is motivated by actual real-world customer/market response encountered while validating the first idea.
An example might be a startup whose first guess at the right product and customer was "An expense tracking program for sales people." After building a prototype or MVP and putting it the hands of some actual salespeople, they discover that, while nice to have, it's not compelling enough to buy. However, during their field visits to observe customers using the MVP they learn that field service engineers have a different but related problem in tracking parts inventory which is a major pain point.
In Steve Blank's legendary startup class at Stanford he teaches that initially a startup should be thought of as "a temporary experiment to discover a viable business." Once product/market fit and go to market are sufficiently validated with an MVP, then the process of building the real product and company begins.
Pivots are a thing because early stage startups are often pursuing an opportunity space they are betting is just emerging. Because of this there's often little data to go on when they start and they're making educated guesses about a bunch of things from product features, primary customer, go to market and business model. The concept is that a startup should very quickly develop an MVP which enables engaging with actual customers. That engagement often reveals which initial assumptions were incorrect. Hopefully they were incorrect not because they made a dumb mistake but because the correct answer was unknowable without testing.
That testing also sometimes reveals completely new customer and/or market information about a bigger opportunity in that area leading to a pivot. This is akin to sinking temporary test wells before building an oil derrick and pipeline. Even the best geology analysis can only give a probability of oil being found in a given location. Sometimes sinking test wells reveals the oil is actually a quarter mile to the East. That isn't really a pivot but more like product/market tuning. A surprising number of times, the process of sinking those test wells to discover oil leads to discovering something different but just as good like natural gas or something even better like gold. That's a canonical pivot. Experienced founders and VCs can pretty easily tell the difference between flailing and a pivot because the pivot is motivated by actual real-world customer/market response encountered while validating the first idea.
An example might be a startup whose first guess at the right product and customer was "An expense tracking program for sales people." After building a prototype or MVP and putting it the hands of some actual salespeople, they discover that, while nice to have, it's not compelling enough to buy. However, during their field visits to observe customers using the MVP they learn that field service engineers have a different but related problem in tracking parts inventory which is a major pain point.
In Steve Blank's legendary startup class at Stanford he teaches that initially a startup should be thought of as "a temporary experiment to discover a viable business." Once product/market fit and go to market are sufficiently validated with an MVP, then the process of building the real product and company begins.