1 - Business Domain: The most important thing to consider is how each unit of measure (a numerical field on your form) is communicated between users in the manner specific to their respective industry. Numerical values can often end up being used like jargon within specific industries. A good example is the use of the term basis points in the banking industry.
2 - Implied Dimensions: Consider any other dimensions that are typically implied within the unit of measure. A commonly implied dimension is time, e.g. monthly expenses, annual expenses, quarterly expenses are all dollars but very different units of measure. The form should communicate clearly to the user what implied dimension you are expecting (e.g. monthly, annual, etc).
3 - A Good Typical Value: Assist the user by placing (in a subtle manner), a typical value to the right of the field. A subtle manner could be to have the typical value in light grey rather than black. Using a typical value provides helpful feedback to the user with regards to:
- The expected form of the numerical input, e.g. 70 or 0.7 to represent 70%.
- What level of precision is supported, e.g. 80.67 to represent 80.67% informs the user that the input field supports precision to 2 decimal places.
- A typical value lets the user know if their numerical value is in the right range (or ballpark). For example, using $7,000 as a typical monthly income should catch the eye of a user who enters $50,000 when mistaking the field for annual income.
4 - Correctness: Add some business intelligence behind the form that prompts the user if any input fields seem unusual. Often a user entered numerical value can pass validation but is not appropriate when considered within the context of the unit of measure. For example entering $100,000 within the monthly income is most likely not correct although it might be and therefore you need to prompt the user a second time.