CAUTION: The following post (or a page linked to) includes or discusses loading data not covered by currently published sources of tested data for this cartridge (QuickLOAD or Gordon's Reloading Tool data is not professionally tested). USE AT YOUR OWN RISK. Neither the writer, The Firing Line, nor the staff of TFL assumes any liability for any damage or injury resulting from the use of this information.
I use the rule of thumb to take 5% off the powder charge and work back up. This is because Charles Petty did some testing in Handloader in 2006 with the 223 Remington in which he got additional velocity equivalent to what about a 5% increase in powder charge would have produced. That's the most difference I've ever seen a primer make. There is no rule that applies perfectly all the time, though. When German Salazar's Rifleman's Journal was still up, he had some large primer tests in 30-06 that produces, IIRC, less than 20 fps average difference, so kind of in the noise. There are even some situations in which magnum primers decrease velocity depending on the ignition characteristics of the powder involved, whether or not the space was small enough for the primer to start unseating the bullet before the powder burn really got underway, etc.
One thing I've done before is look up a load in Hodgdon's data and divided the difference between the minimum and maximum load velocity by the difference in the minimum and maximum powder charge. This gives you fps/grain in that load region. I multiply that by 0.05 to get the number of fps I expect to lose if I don't change the primer. I can then download 5% using the magnum primer and if the first few shots produce that projected reduction, then I say it is making no significant difference to velocity and move the load back up to where it was before. But if it is not producing that much reduction, I divide the difference in the velocity I am getting and the velocity the old load got by that fps/grain number to get how much to increase the powder charge to get back to that original velocity and increase the charge only by that much.
That may seem complicated, but it saves a lot of test shots. Ten with the reduced charge gives me a mean velocity that is usable for that purpose and that's it. It is more precise if you have a few of your old primers to load up to get your velocity comparison under the same conditions.