If a long-only strategy returns 14% with 18% volatility while its benchmark returns 11% with 16%, the raw sharpe-ratio flatters it by counting market exposure as skill. The information ratio looks only at the 3% of excess return and the volatility of the difference, the tracking error.
Worked example: excess return 3%, tracking error 5%, information ratio 0.60. That is respectable for an active equity manager and would be unimpressive for a market-neutral strategy where all return is by construction active.
Choosing the benchmark is the whole argument. Measured against cash almost anything looks good in a bull market; measured against the obvious passive alternative, most strategies do not justify their turnover.
Related: sharpe-ratio, alpha, beta-estimation, performance-attribution