Fault in new cartesian sign conventions

If we analye, while proving mirror,lens or lens makers formula we make use of sign coventions but when we solve numericals we again make use of sign conventions why? This creates problem for new cartesians sign conventions