The report may show differences between the number of visitors assigned to each variation. For example, you have defined a 50/50 traffic modulation, and your report shows a different split.
This is probably due to the quantity of visitors who are assigned to the test being too low. We arbitrarily assign a visitor who arrives on your site to a variation of your test, but this arbitrary assignment only works on large sample sizes.
Just as when you play heads or tails, it is possible that out of ten tosses, heads comes up seven times and tails three times. Toss the same coin one hundred times and you may see sixty heads and forty tails. The more tosses, the more you will tend towards a 50/50 distribution. As an example, we may see a difference of up to 12.4% if we toss the coin a thousand times, 4.32% if we toss it ten thousand times, and 1.1% if we toss it one hundred thousand times. The difference is only 0.37% if we play a million times.