A discovery phase is the paid research step that happens before anyone opens a design tool. It's where a studio works out what the actual problem is, who the site or product needs to serve, and what "done well" means for this business, before a single screen gets designed. To a founder who wants to see design work start on day one, it can look like a delay with an invoice attached. It's closer to the opposite. It's the part that keeps the rest of the budget from being spent solving the wrong problem.
What happens during it is concrete, not ceremonial. A real discovery includes conversations with the people who run the business about what has and hasn't worked before, a look at whatever data already exists, a read on what competitors are doing, and enough contact with actual users or customers to know who the work is really for. Studies of how discoveries run in practice find the average one uses close to five distinct methods stitched together, not one kickoff call and a mood board. The point of all of it is a written problem statement everyone agrees on, not a folder of notes nobody reads again.
Skipping this step is usually framed as the frugal choice. The evidence says otherwise. In an industry survey of UX practitioners, projects that included a discovery phase were reported successful 83 percent of the time, against 52 percent for projects that skipped it. And the honest reason most discoveries get cut isn't the budget: time pressure is the leading reason teams skip it, cited far more often than lack of money. Skipping discovery is a bet on being right without checking, and rebuilding a site that quietly solved the wrong problem costs more than the discovery fee would have.
Set against the size of a full project, discovery is not where the money goes. Design and build projects of this kind average in the tens of thousands of dollars over several months of work. A discovery phase is a front loaded slice of that: typically a small team's time over a matter of days to a couple of weeks, not the bulk of the invoice. What that fee buys is certainty before the expensive part, the actual build, gets underway. Not extra polish charged on top of the same work.
The way to tell a real discovery from a padded line item is to ask what it produces. A real one ends with something you can point to: a written problem statement, notes from actual conversations with real stakeholders or users, and a design brief that is different, and better, for having gone through it. A discovery that ends in a slide deck of stock inspiration images and nothing else is not research. It's a delay wearing a research phase's name.
Before paying for one, ask what document or decision comes out the other side, who gets talked to, and how it will change the brief a designer works from. A founder who wants to skip straight to screens is optimizing against the wrong risk: the cost of a wrong guess, not the cost of a few days spent making sure the guess is right.