Wireframe isn't a form of modeling-- you can set it to wireframe view, which means it won't show the shaded polygons, just the edges and vertices. The method I'm using is "box modeling" where I start out with a geometric shape (a sphere actually, in this case) and I move and cut up the polygons and edges into the rough shape of a face, when use nurbs to smooth it out for rendering. The other popular method is called "plane-modeling" where you start out with a single, small plane and extrude the edges out in different directions and literally construct the face one polygon at a time. It's MUCH more time consuming, but some say easier, and with better results. Lastly, and less used is called "spline modeling" where you outline the profile of the head with a spline, extrude that out a bit, then select an edge at the crown of the head and extrude that out one polygon at a time to for half of the frontal outline of the head. Then you use bridge to connect polygons of the profile to the polygons of the frontal outline over key locations such as the jaw, cheek bones, and eye brows. Then you fill in the spaces between the cage-like template you just made and move around the polygons to curve with the face reference you should have visible in your viewport, then mirror the results which cuts back on time since you only have to construct half the face-- but that leaves unsightly/sharp seems now the middle of the face because the polygons, when mirrored, don't always meet up smoothly
I wouldn't start again, just carry on building on what you've got. I would say possibly make it "thicker" if you know what I mean? Are you using a wire frame to make this?
.