A little chemical detective work gives a plausible answer.

Say the metal is cobalt and it's oxidation state is +2.  Then, with the cobalt calcined to the oxide, you have two possible acid-base reactions:

$\ce{CoO + 2 NaBO2 => Co(BO2)2 + Na2O}$

$\ce{CoO + B2O3 => Co(BO2)2}$

Both reactions look good, but upon further review the first one gives something we recognize as a powerful base, more likely to be a reactant than a product.  Moreover,  a quick check [here](https://en.m.wikipedia.org/wiki/Boron_trioxide) reveals that boric oxide has a low enough melting point to liquefy in the flame, favoring the second reaction kinetic-ally (hyphen put in because device autocorrect is acting badly).

So the second reaction above is really the only plausible one.