Well, what I made wasn't trial-and-error, it was (a) measure signal strenth on smartphone in distant room (b) bend aluminum to what looks right and stuck it on wall behind antenna (c) go back to distant room and measure signal again, one bar more, hmm, not bad, can't be bothered fiddling about with it !
And on that subject: once upon a time it was discovered far too late, that a technician employed in a satellite cleanroom had accidently struck one of the axial thrusters on a shiny new spacecraft, knocking it skew-wise with a hammer while working on the transport baseplate. After launch, attitude manoeuvres went just fine. At the Board of Inquiry he explained he hadn't bothered to report the accident as, (and I quote) "I just tapped it back into place until it looked right again".
That's the way to do it !