You help your Prime avoid making a bad investment in a result that may look good on paper (or his mind's eye), but that doesn't have much chance of improving productivity and end-user morale.

You learn from the best informed people (those who do the work) what they need, increasing the likelihood that your Prime will get what s/he's trying to invest in (better profit numbers and/or lower costs through greater productivity).

You know that point at which the ROI (Return On Investment) formula goes negative (as in "You're investing more in the project than you can expect to get out in worthwhile results.") and, consequently, that point at which you should either (a) get much more creative or (b) abandon the project for something with greater promise.

You can give your stakeholders a much better estimate of project return than they could get without the end-user information you've gathered. (Note: Bottom line return estimating is a tough enterprise, but project managers are in a good position to gather that kind of information beforehand and, with a disciplined approach toward comparing estimates with actuals, project managers do become much more accurate ROI estimators.)

You can avoid disrupting end-user "busy times" when the introduction of your project result would reduce productivity and induce resistance.

You can design use training for your project results that builds from what you discover your end users know to a point where your result can be confidently and productively used. (You lose big points by going over familiar ground and/or confusing end users by starting your training at a point that's over their heads).

You choose 'typical' beta testers from the end user population who can point out the problems that their 'typical' work mates will encounter rather than masking problems by choosing the eager overachievers who make your result look easier and better than it really will be.