I am sorry to break the news to you that there is not a PROVEN "rule of thumb" for leakage losses.
If you are really making "estimates", there is likely to be more error in your estimates from uncertainty of the nominal turbine efficiency than from accounting for leakage.
Depending on steam conditions, rating (implying seal diameters), manufacturer, nominal type (reaction or impulse), etc., the HP end leakage could be 0.25% - 5% of the main flow.
The best rules of thumb will be based on measurements or calculations that make some accounting for the actual construction details of the seals on your machine. Performance quoted by any manufacturer will account for the design-basis leakage flows.
The typical equation for these leakage flows is (simply) an orifice type of equation where:
FLOW = C1*C2*AREA * (UPSTREAM PRESSURE/UPSTR.SPEC.VOL.)^0.5
From the equation above, you can see the "the pressure at which the (leakage) occurs" IS RELEVANT.
The constants are functions of geometry and pressure ratio respectively. I've separated them in this discussion because the magnitude of the "geometry flow coefficient" can vary by a factor of 2 or 3 for labyrinth seals alone, depending upon the seals' construction details. The spread is greater once one includes factors such as wear and/or damage.
Actual physical areas (the annulus areas) can also vary, depending upon the turbine design.
The relationship of leakage to lost power also depends on where the leakage occurs. The HP shaft end leakage will typically have done useful work on the first stage before leaking out. If any of the leakage is re-entered at a downstream stage, this will reduce the lost power.
Interstage leakage - over the bucket tips and under the stationary diaphragms - is, of course, different.
