The main argument given during discussions on the Internet concerning changes in body composition is that it can be achieved thanks to aerobic exercises, “burners”, strength workouts or sophisticated reduction diets. There's nothing wrong in basing on such information, however, I would like to pay attention to massive distortion connected with the measurement of body fat with the BIA method. Bioelectrical impedance analysis (BIA) is based on different resistance of tissues in the body.
Unfortunately, it is not the best measurement method (but it's cheap, fast and non-invasive). Once, I tested a machine for £2,000 when I visited its official distributor. At that time, my body fat level wasn't higher than 10% (measurement of pockets of fat + visual assessment). The machine showed almost 18% of body fat! If you believe such measurements uncritically, you can definitely notice “amazing changes in body composition” after a month. And then you cannot believe the articles which suggest that such fat reductions are impossible to achieve even using advanced pharmacology. One of the reasons of mendacity may be using body composition analyzers and scales using BIA technologies. The cheapest devices can be bought for really small amount of money (bathroom scales, e.g. TANITA UM-018). Unfortunately, measurement taken with such machine is worthless. The seller offers accuracy of 0.1% of body fat. That's very interesting – a world class precision for a device costing less than £30... Whereas the equipment for almost £10,000 (!) GE LUNAR PRODIGY (DXA/DEXA technology) may have the measurement error up to 4% (the matter of software and patient's specificity)! Against the marketing lies, a device for less than £200 using BIA technology may be treated only as an expensive toy. Bathroom scales are useless – the measurements involve gigantic error.
What about “the professional body composition analyzers” for £2,000 – £4,000? (you can find simpler, second-hand devices much cheaper on some auction services). The machines measuring body fat found at the dietitians' or clinics are usually based on the BIA techniques. These are usually produced by TANITA, SECA or JAWON. According to the distributors' declarations, the TANITA BC 420 SMA analyser has measurement error of 0.1%. It sounds great. Unfortunately, marketing slogans are nothing else but bantering clients. In reality, the measurement error with BIA method may range between 3-5%, whereas, the analysis with DXA method – dual-energy X-ray absorptiometry – has the repeatability of about 2%. And what is the measurement error with the DXA method? Well, for a group it's about 1-2%, but... the individual measurement error may even range between 4-5%! [1, 2, 3]
And what about the direct confrontation of BIA method with the other – much more accurate?
“The experiment included 145 women at the age of 22 – 40 (averagely 31.5 ± 5 years) with proper BMI (21.8 ± 1.7 kg/m2). The percentage of body fat, as well as fat and lean body mass were measured with body composition analyzer TANITA BC 420 SMA (BIA) and densitometer LUNAR Prodigy (DXA).
The average percentage of fat content was:
32.05 ± 5.2 % measured with DXA method
26.05 ± 5.1 % (BIA; p < 0.02).
The bioelectrical impedance, comparing to DXA, had good specificity (96%), but low sensitivity (35%) in identifying women with the percentage of body fat over 30%”. 
Comment: as you can see, the BIA method may lower the level of body fat even by a dozen of percent! But that's not all! The BIA method may perfectly and malignantly improve reality. Why?
“Comparing to DXA, in the BIA measurement, the average percentage of fat was significantly lower (by 18.6%), similarly to the total fat mass (by 14.5%). Whereas, lean body mass in BIA was higher by 13.2% than in DXA.” 
So, not only does a woman find out that she has much less body fat than in reality, but also the machine suggests that her muscles have grown like after using steroids (by 13.2% more LBM). In reality, the ladies had 18.5 kg of fat, and the BIA method only showed 15.82 ± 4.5 kg. If it comes to the body components burning calories, including muscles, bones, inner organs and connective tissue + necessary fat: in reality, the DXA measurement: 38.72 ± 4.2, BIA method showed 43.85 ± 3.2 kg.
The matter of body composition and kinds of tissues were widely described here: http://potreningu.pl/articles/937/sklad-ciala-a-mozliwosci-wysilkowe
Is it all? No, it's not!
In the next study from 2013 , 97 women were measured concerning the amount of body fat using DXA measurement, pocket of fat and the BIA eight-point analysis. The measurements took place before and after 21 weeks of training.
Unfortunately, the BIA method understated the level of body fat by 6.5%, it showed less fat by 3.42 kg, and it significantly overstated “lean body mass” (muscles and other components burning fat) by 3.18%! For the method of measuring the pocket of body fat, there was the underestimation, but minimal – only by 1.69%.
It's not worth using the BIA method on cheap scales. You should remember that even the results achieved on some equipment for £6,000 may be unreliable.
Bathroom scales do not do the “body fat measurement”. The results are not the reference point to anything. You'd better use estimation on a website.
Significantly big mendacity can be noticed among sports people involved in strength disciplines, who have low amount of body fat.
Base on a mirror, centimeter and pocket of fat measurements. Use the BIA measurement only as a complement.