For the past few years, Fortune has been obsessed with the story of transformation. Every sector of the economy is going through a digital-driven shift of some sort, and we’ve chronicled that, from Marriott and Citigroup and BMW to L’Oreal and Microsoft.

Most Fortune 500 incumbents follow a pattern:

Stage 1: Dismiss the disruptive factors that could lead to their demise.

Stage 2: Wait for their stock price to start slipping. Witness Airbnb’s valuation creep above their own market cap and whine as Amazon trades at a P/E ratio of 187X.

Stage 3: Tout their internal innovation and bristle at the suggestion that they’re behind.

Stage 4: Panic and throw some money at vague innovation initiatives. Promote the heck out of those efforts, regardless of how effective they are.

Stage 5: Give up and make a crazy-sounding, headline-grabbing acquisition that looks like a Hail Mary.

Stage 6: Replace the CEO.

I’m most interested in the “make or break” moment that happens around Stage 4. There’s an intense desire from large corporations to associate themselves with startups and tech innovators. And of course, startups desperately want to work with big companies that can bring them legitimacy and potential customers. But those relationships don’t often work as planned.

Sign up for Term Sheet, Fortune’s newsletter on deals and dealmakers.

500 Startups, the accelerator program and venture fund that bills itself as the most active startup investor in the world, has surveyed 100 corporate innovation programs, which range from pilots and partnerships to direct investments and acquisitions. The survey found that — surprise! — most of them aren’t very successful. A full 81% of those surveyed say that fewer than 25% of their startup pilots have resulted in commercial deals.

Why such a low success rate? The study blames corporations for being slow, disorganized, and too conservative. (That blame suggests that the corporations have more to gain from these relationships than the startups do.)

500 Startups found that corporations aren’t working with enough startups — only 9% do more than 50 startup pilots a year — which isn’t enough to yield a meaningful number of successful deals. They need a “portfolio” approach. It’s self-fulfilling with 50 or more pilots a year are generally viewed internally as more successful and get more resources to execute more pilots.

Likewise, corporations move too slowly – 20% of companies take more than six months to do a deal, which may as well be six decades in startup years. Companies that “fast-track” processes like short NDAs, short purchasing agreements, centralized points of contact, and simple inbound application processes are more successful.

Lastly, corporates that struggle don’t have “organizational alignment” to keep startup engagements on track, get them the right resources, and ensure all parties are moving in the same direction.

One thing this study is lacking is a giant chart which maps out which programs move fast, experiment a ton, and produce positive results, and which ones are merely shopping for ideas to steal. (I’ve heard countless horror stories of the latter, especially from the auto industry when I was closely covering the self-driving car startups last year.) This is especially important information for cash and time-strapped startups, as more legacy corporations open up Silicon Valley outposts, launch accelerator programs, and go on “innovation tours” down the peninsula.

Find the full report here.