With vibe coding these days, you can just ready, fire, aim if it's simple enough. If it takes a month to do research, but a weekend to build a prototype, just build the prototype and show them to the customer.
Even putting aside the wisdom or lack thereof of "vibe coding", this is still wrong. A weekend of AI generated code will still be better if it is informed by a few hours of prior conversation with customers.
And really, that is the answer to OPs question - you should keep your research in line with your coding efforts. Small coding == small research. Big coding == big research.
Something I found that helps me is asking the question - if in retrospect the outcome of my plan will not be as positive as I have planned - will I regret not researching more right now.
But, this came with a lot of experience. In the beginning it's hard to imagine how and in what ways the outcome can be not positive and how it connects to the research today. And, I feel my experience is still lacking around this.
Research should not be done in surveys. The old way was at least a powerpoint presentation, vertical slice, or a wishlist. However, people have gone wise to wishlists. Prototypes are the new PPT and faster to set up.
Customer interviews tend to exclude a lot of things too. Do people want a browser without a million tabs open? Sure. Do people want a payment gateway that can be set up in an hour? That sounds like a scam, but why not? Do people want a game about a plumber jumping on tortoises who shoots fireballs whenever they eat a flower? Probably not.
I do believe people have a good idea of what they want to bring to the world and at some point it's just easier to show what they had in mind.
It depends.
If you need to ask this question, then ship small to your audience first.
With vibe coding these days, you can just ready, fire, aim if it's simple enough. If it takes a month to do research, but a weekend to build a prototype, just build the prototype and show them to the customer.
Even putting aside the wisdom or lack thereof of "vibe coding", this is still wrong. A weekend of AI generated code will still be better if it is informed by a few hours of prior conversation with customers.
And really, that is the answer to OPs question - you should keep your research in line with your coding efforts. Small coding == small research. Big coding == big research.
Something I found that helps me is asking the question - if in retrospect the outcome of my plan will not be as positive as I have planned - will I regret not researching more right now.
But, this came with a lot of experience. In the beginning it's hard to imagine how and in what ways the outcome can be not positive and how it connects to the research today. And, I feel my experience is still lacking around this.
Research should not be done in surveys. The old way was at least a powerpoint presentation, vertical slice, or a wishlist. However, people have gone wise to wishlists. Prototypes are the new PPT and faster to set up.
Customer interviews tend to exclude a lot of things too. Do people want a browser without a million tabs open? Sure. Do people want a payment gateway that can be set up in an hour? That sounds like a scam, but why not? Do people want a game about a plumber jumping on tortoises who shoots fireballs whenever they eat a flower? Probably not.
I do believe people have a good idea of what they want to bring to the world and at some point it's just easier to show what they had in mind.