If you chose to recrystallise by variation of temperature, you have to find a liquid that dissolves your product much better when heated (or even heated to reflux), than at room temperature. A qualitative test may be performed with a small quantity of substance, for example a tip of a spatula, that you fill in a test tube, followed by a small amount of liquid.
Does your compound dissolve well in the liquid already at room temperature? This is then not a good choice of sole solvent for a recrystallisation, as much of the product will be retained by the liquid. Intentionally I use liquid, because while you likely find a commonly used solvent suitable for recrystallisation, some other chemicals not typical perceived as "solvents" may be useful, too.
Depending on your equipment available, you may extend the accessible temperature range towards lower temperatures by use of an ice bath, or a freezer. Be aware, however, some of the solvents that are liquid at room temperature may solidify then, too (glacial acetic acid, for example).
Often, you face a situation like "ethanol (as one solvent) has a too high solvent capacity for X, and water (as an other solvent) has a too low solvent capacity -- regardless if warmed, or heated to reflux". Then it is time to use a mixture of mutually miscible liquids as solvent of recrystallisation, to tune the solvent properties. This is why, for example, aqueous ethanol (ethanol with an intentionally added proportion of water) is used as solvent of recrystallisation, too.
There are some seemingly obvious choices of solvents of recrystallisation, like methanol and ethanol for not so polar organic compounds, and other somewhat surprisingly well working examples (acetic acid for less polar compounds); hence, recrystallisation offers a spectrum of science, craftsmanship, art, and obviously some luck, too. Note that recrystallisation aiming for purification may be well a different one as aiming for nice crystals (for the purpose of single crystal diffraction analysis, for example).
If you made your choice, retain a small quantity of non-recrystallised compound which may be helpful later. Mix your solid (or tar-like) product with the solvent / solvent mixture at room temperature and heat it up; if you want to heat up to reflux, consider a reflux condenser. At high temperature, all visible solid should be dissolved. If this is not the case, there is to little of solvent present and carefully add more. After each addition of some new solvent, wait a little bit until the temperature of your mixture is the same as prior to this addition. (Caveat: If your product to recrystallise is a mixture, one component may dissolve better, than the other. If so, consider a fractional recrystallisation or other means of purification first.) Then allow the solution a slow cooling to room temperature, for example switch out the heater (and stirrer, too) of the oil bath yet keep the round flask in the oil bath.
Scratching the inner walls of the flask with a glass rod may be dangerous as you accidentally may pierce the wall (and consequently loose product to the oil bath). A long Pasteur pipette is a better choice, too much pressure applied will break it, but not the flask. But in general, it is one good method to generate crystal seeds for nice crystals. (The optimal temperature for crystal seeding need not be the same as the optimal temperature for growing larger crystals.)
If possible (vide supra), complete the crystallisation by cooling below room temperature (ice bath) as this often lowers even more the solvent capacity. If accessible, collect the crystals at this / near this low temperature, too. If you want to rinse your crystals with a "bad solvent" (i.e. with low solvent capacity), equally cool it in advance by an ice bath. If you crop your crystals by vacuum suction, prior cooling the equipment (Buchner filter, suction flask) by an ice bath / in a freezer may be helpful, too.
As a final note (here), do not throw away the mother liquor / filtrate of your recrystallisation too early. It may contain a considerable amount of product. Concentrate it (for example, at the rotary evaporator), and repeat the recrystallisation for a second crop. Do not merge crops of subsequent crystallisation until an analysis reveals them as of sufficient similarity, like comparable purity.