There are quite a few "rules of thumb" that people use about third-party grading. None of these are hard-and-fast rules. They are just ideas that people have come to accept as a type of collective wisdom. They can be helpful, though.

Value. Many people set an arbitrary coin value where grading may make financial sense. A lot of people draw the line around $150-$200 minimum value. Others of us set the limit higher. (I draw the line around $1500, except for heavily counterfeited dates.)

Authenticity. Many people have coins graded where authenticity either is doubtful or where having the
TPG guarantee could matter. An example would be a G/VG 1909-S VDB cent. Another example is gold coins that trade largely at bullion value, such as US modern commemorative half eagles. If bulk submissions can be done, having the slab greatly helps sales.

Collectible varieties. Sometimes having a particular variety recognized makes a coin more valuable or increases the ability to sell a coin. An example might be a 1917 double die cent.

Bulk submissions, where the per-coin price is low. An example might be purchasing a couple BU rolls of a pre-war
Lincoln Cent. Take 1937-D as an example. It is common in mid-UNC grades. but the bump to MS-67 RD is over $100 per coin, and a bump to MS-68 RD is around $6,000+. It only takes a couple MS-67 RD coins to make the bulk grading worthwhile, and one MS-68 RD is a walkoff grand slam home run.

Gambling on a higher grade. For many coins, just a one-point bump in grade can be worth hundreds of dollars. That's also true for improving a designation, such as RB to RD for a cent or getting a PL or DMPL designation for a
Morgan dollar.

It costs nothing to post coins on here and see what others think of the grade. Doing that first can help inform the decision whether to send the coin to a
TPG, or in some cases which
TPG might be better for a given coin.
Just some random thoughts.