The difference between 75% efficiency and 48% efficiency can be 5 degrees.
While this may not sound like a tragedy, it can be. According to Greg Noonan, author of "New Brewing Lager Beer" For a light beer the target Scarification temperature is 150 degrees, and for a malty full bodied beer the target is 157 degrees. That's only a seven degree difference from light to full bodied.
The last three mashes I brewed I targeted 158 degrees to enhance the body of the beer. Because my thermometer was reading high by 5 degrees the actual mash temperature was 153 degrees which is the target for a typical ale. That's not the thick malty style I was going for. Had the target been toward the low end, to create a highly ferment able wort, it is likely that there would not have been complete conversion. This would leave starch in the wort as well as create a low initial gravity and a low efficiency.
So what can you do about it?1) Buy a better thermometer. Pens style thermometer seem to be more accurate than the oven style out of the box in my experiance. The Tru Temp that my wife purchased is very accurate. If you want to be certain then purchasing a thermometer that has calibration data will let you know how far it is off. Many brewers swear by the Thermapen. It is pricy, so I haven't tried one, but this does have calibration data. +/-2 degrees F. Also, looking for something with a Type K or J thermal couple is probably a good place to start.
2) Compensate for the thermometers inaccuracy. The two easiest known temperatures to measure are boiling and freezing. Mash temperatures are closer to boiling. So, if you boil water, and live near sea level, you should measure 212 degrees. If you know how many degrees off your thermometer is you can correct the value that is read.
3) Calibrate your thermometer. This is most easily done with an oven style thermometer than can be unplugged. If you're an electrical engineering geek like me, this is a fun little project. Measure the resistance of the thermal probe at freezing and boiling. record what the thermometer reads under these conditions. Calculate the required trimming potentiometers to add in parallel with the thermal probe and in series. Trim the large parallel resistor to make ice water read 32 degrees, then trim the small series resistor to make boiling water read 212. Check again at 32 and repeat if needed. It looks like for my oven thermometer a 10Meg trim pot in parallel and a 1k trim pot in series will do the trick.