There are several ways to create Bose-Einstein condensates or systems that behave that way, there are ultracold atomic gases, solid state quasiparticles, and even photon condensates.
Since you are obviously interested in ultracold atomic gases, I am going to cite Experimental methods of ultracold atomic physics by Kurn and Thywissen:
The material must be constructed prior to each measurement, starting from a new, hot atomic vapor. The typical experimental sequence is
(1) using laser cooling and trapping to gather atoms from the vapor into a magneto-optical trap, with temperatures on the order
of $100 \mu K$,
(2) trapping the atoms in a conservative potential, e.g. a magnetic (Sec. 2.2) or optical-dipole (Sec. 2.3) trap,
(3) evaporatively cooling the atoms by gradually lowering the depth of the conservative trap and letting atypically high energy atoms escape the trap, reaching temperatures in the quantum-degenerate regime (usually sub-$\mu K$) and
(4) putting the final touches on the material by turning on the system Hamiltonian (interaction strength, lattice type, spin admixture, etc.) that one wants to examine.
Once finally produced, the lifetime of the ultracold material is short, and so the entire experimental sequence for probing the material is quick, typically much shorter than the time it took to prepare the material in the first place. At the end of each measurement, the sample is discarded. This cycle is repeated at the cycle time of a few
seconds to a few minutes, depending on the speed of the accumulation and cooling stages. In examining a graph of data from a cold atom experiment (For example, Figs. 3 or 4), one should value the fact that each point on the graph represents one or several repetitions of a make-probe-discard experimental run. In light of this protocol, it is somewhat stunning to hear of cold-atom experiments that require hundreds or thousands of "shots", each reproducing a gas under almost identical conditions, to obtain the high precision
required to reveal new phenomena or test recent theories.