As shown in the article, that first recommendation leads directly to the next step(s). In the ideal case (which actually happens sometimes in the real world!), you end up with something like this:

    if (bad1) fail();
    if (bad2) fail();
    if (bad3) fail();
    
    step1();
    step2();
    step3();
    step4();
    return SUCCESS;

To me, this is a very nice way to structure things when you can. You get a checklist of things that might be wrong, and then comes the wonderfully naive straightforward implementation.