Training results can be measured on several levels. You can collect trainee perceptions about training, observe how trainees perform while in training, measure whether knowledge/skills learned in training are transferred to the job, and determine whether training makes a difference in your organization's overall performance. You can also quantify whether the benefits of training outweigh the costs (through cost benefits analysis). These levels of evaluation each yield a valuable, unique data set, but each also has associated costs. 
 

It is relatively simple and inexpensive to collect trainee perceptions about training. "Smile sheets" (end-of-training critiques) are very commonly used and, although this type of evaluation yields soft rather than hard data, knowing how trainees felt about their training can be a moderately useful barometer of training quality. Obtaining trainee feedback is also critical to continuous process improvement efforts, so this level of evaluation should always be performed. 
 

Evaluating how trainees perform while in training is also straightforward and is typically done through some form of testing. Testing can yield hard data about what a trainee has learned, but caution must reign. Valid tests are not easy to design, and invalid tests produce invalid data. Also, many managers/supervisors often overlook the fact that while testing can verify employee performance during training, test results are not necessarily reliable predictors of on-the-job performance. On-the-job performance is complex and is typically affected by many factors. 

Critiques and testing are by far the most commonly used evaluation tools. You can obtain more robust data, but higher levels of evaluation can be difficult and costly to perform. Quantifying training impact on job performance or on business results requires that you have pre-training baseline data and the ability to control other work environment variables (so that training effects can be isolated). This is no small feat. For this reason, many organizations choose to evaluate post-training results only in informal, observational ways. 

You determine how much data you need to quantify the success of your training initiatives. If you need hard data, seek expert advice and invest in higher-level evaluation methods. If hard data is not required, stick to lower levels of evaluation...but do evaluate.