A major U.S. financial institution-call it HugeBank-was justly proud of the innovative online services it offered its millions of customers. It finally got around to analyzing its virtual customer base. The findings stunned management. Roughly 90 percent of the bank's online customers logged on barely once a month. By contrast, an obsessive-compulsive 10 percent logged on at least once a day. HugeBank was both embarrassed and annoyed that its pathological dependence upon averages had led it to create a "typical customer," concealing the extent of this disparity.

But here's the steel-toed kicker: that laggard 90 percent accounted for barely 30 percent of the bank's deposits; the hot-wired 10 percent accounted for almost 70 percent of the money. In other words, HugeBank's most valuable customers by far were those who embraced the online service.

Those stark statistics obliterated what had once been a broad and aggressive innovation agenda filled with Internet-based strategies. The bank discovered that "Internet innovation" had completely different meanings-and profit potentials-for each of these customer segments. The bank's belief that a single new-product development group could create cost-effective innovations for all the company's customers provoked a huge internal debate that continues to this day.

With apologies to Harvard Business School's Clayton Christensen, this revelation represents the innovator's real dilemma: innovation for whom? It's not obvious which of these customer segments merits what kinds of innovation. Should the bank focus its technology initiatives on better service for its nebbishy 90 percent or for its most opulent top tenth? A solid business case could be made that the bank would be better off investing the barest minimum to support the 90 percent. Then again, the bank had best be careful about how innovative it should be with the most active customers. They may want less online innovation, not more. Indeed, they may not want innovation per se: they may simply want better service.