Telling your executive VP such investigations can be done quickly and inexpensively.

Not having corporate counsel sign-off on the equivalent of an in-house search warrant before searching network accounts and cubes or offices for evidence (both electronic and nonelectronic).

Conducting searches without a solid understanding of where employees do, and do not, have an expectation of privacy.

Misidentifying your policy violator and interviewing the wrong employee (during which you may imply they are a pervert or at the very least dishonest).

Allowing IT staff to "assist" with the technical aspects of the investigation. (Remember what happens when the fox gets to guard the hen house?)

Allowing investigations and analysis of E-mail and Internet activity to be used for witch hunts.

Not realizing that forensic standards for law enforcement and forensic standards for corporate investigators are significantly different.

Treating every investigation like it will be going to federal prosecution.

Not using an eyewitness or pinhole camera to tie your policy violator to the keyboard in question at the time of the original incident or when the incident reoccurs.

Failure to personally interview the policy violator, victim, complainant, witnesses, and peers in the incident under investigation.

Allowing human resources to participate in the technical investigation before the employee interview. (Can you say leak?)

Failure to follow a reasonable "chain of custody" procedure when handling evidence.

Not being able to describe/define the process used to discover and acquire evidence to senior management in terms they can understand.

Improper storage of evidence (not under monitored lock and key).

Allowing unauthorized employees to examine the computer or evidence discovered such that allegations of evidence tampering can be made.

Not understanding the types and locations of potential logs containing evidence that are produced by security controls within your infrastructure.

Not understanding how easy it is to spoof MAC and IP addresses.

Analyzing the original evidence. (Use duplicate copies for this whenever possible.)

Not verifying who had access to computers where evidence has been discovered.

Failing to perform a complete virus/Trojan check on the evidence prior to analysis (avoiding the "someone else caused it" argument).

Not verifying that the timestamps of computers involved are accurate, making event correlation difficult to impossible.

Being unable to pay attention to boring, minute details (the ones that often end up cracking the case).

Deviating from accepted procedures while handling or examining evidence.

Not documenting ongoing discovery and analysis activities in a detailed log.

Being unable to get a signed, handwritten confession from your policy violator.