This is difficult to answer as it is very general, but I’ll try.
Firstly, and most importantly, make sure the student knows exactly what they are aiming to do/prove/measure.
Secondly, make certain (on reflection, this may be number one too!) that they can measure something and record the data.
Thirdly, be (very) cautious of investigations that require the student to construct or develop equipment/techniques (just an additional complexity they do NOT need.)
Finally, make sure they understand the theory (and check their understanding, students are masters at concealing misconceptions) before they start. Just knowing what to plot (think discharging a capacitor investigation if they can’t do logs to plot the graph and find a value for RC) and how they will get actual data on day one is vital.
Overall it is important that they follow their interest, that they want to do it, but students often require (sensitive?/clear?) guidance to be successful.