An Application that Needed no Help

It has been said that what people really want is telepathic computers, Well, while that is something we cannot do (yet), we can and did manage to develop one application that was so efficient, intuitive and user-friendly that it needed no online help buttons. Here is how it happened.

The application - Eric - was designed for use by research chemists to submit sample compounds for chemical analysis. The usual efficiencies were built into the user interface, such as:

• Logical workflow
• Default pre-filled data
• Carefully worded labels and instructions
• Automatic error checking
• etc.

The server was also sized and tuned to ensure acceptable operation speeds. However, the key feature was that almost every item on the screen was live (or in programming terminology, almost everything was a control). Consequently, the key training statement was:

Whatever you want to do, moving the cursor to the relevant place & clicking on the mouse will usually make it happen.

If a user was having difficulty with something, all they needed to do was click on the relevant item, and an input window or new screen would appear that usually made it clear what to do. They could also 'cancel' out of the operation if it was not what they expected.

The validation of this approach came during end-user testing. A test panel of real users was selected to cover a wide range computing ability - from "awkward customer' to 'whiz kid'. The panel was split into three groups that were trained as follows:

• Given one-to-one training and a comprehensive user guide.
• Given the user guide only.
• Given nothing, but just told how to launch the application, and asked to see how they got on.

A formal post-trial survey revealed that all three groups got on equally well, showing that the approach worked. In the production system, a summary user guide was all that was needed to be provided by way of training.

The original pilot application was liberally salted with Help buttons, but there was no actual online help behind them. Online help is labour-intensive to write, so we only wanted to write help where it was really needed, rather than guess and waste time. This was explained to the test panel, and they were asked to make a note of the places where they felt they needed help.

At the end of the trial, only three cases were reported where help was needed, and in every case this could be provided using a one-line instruction on the appropriate screen. The Help buttons therefore seemed to be unnecessary. This was put to the test panel. They agreed, and consequently all the Help buttons were removed in the final version.