Yes, a custom typeface is a good idea

Custom typefaces often get a bad wrap as the fetishized, self-indulgent projects of “visual designers,” but platforms risk ignoring real metric-moving opportunities by failing to invest in typography.

In 2012, eccentric documentary-maker, Errol Morris, ran a test with the New York Times. The test asked readers to verify whether or not a series of statements were true. Each test-taker was shown the test in a different typeface in an attempt to see whether the believability of simple statements, such as “Gold has an atomic number of 79,” was affected by font choice. The results showed that traditional serifs, like Baskerville, made statements seem more true, while child-like fonts, like Comic Sans, were seen as less trustworthy.

Statements written in serifs seemed more believable to respondents than sans-serif or stylized typefaces.

In 2015, I ran a similar, larger experiment at Yieldmo, an enormous ad network, where I was interested in seeing if changing ad typefaces would make people trust (and click) on mobile ads more often. The results leaned in support of the 2012 NY Times experiment: typefaces influence user behavior.