What it is:
How it works/Example:
To create a whisper number, somebody (usually a website, investment bank, brokerage firm, or other institution) polls investors and analysts about what they really think a particular company's per share are going to be for the next quarter and/or year. Obtaining the results -- the whisper number -- is usually a matter of privilege: preferred clients, paying website subscribers, and other connected people are often the only ones who have access to the information.
Whisper numbers often differ from consensus estimates, which are generally company-guided, analyst-driven, published forecasts. Thomson First Call, Zacks, and Reuters are prominent providers of consensus estimates. Whereas consensus estimates are typically math-driven, whisper numbers often are the result of "gut feel" or less quantitative information. They are a big part of the Wall Street rumor mill.
Why it matters:
Whisper numbers are most useful when they differ from the consensus estimates, because they can sometimes predict earnings surprises. Earnings surprises in turn often lead to quick changes in the stock's price--when actual earnings exceed expected earnings, the stock price jumps; when actual earnings "miss" expected earnings, the stock price falls. It is possible for a company to beat consensus, for example, but miss the whisper number.
If the whisper numbers prove more accurate than the consensus, they begin to carry a tremendous amount of weight; for companies that don't have a lot of formal analyst coverage, whisper numbers can be even more influential.