I'm not a professional typographer that does type design for a living, so in my case, it usually started after designing some letters for a logo and decided to draw the rest of the character set.

My priority is to make the glyphs look good at display size and well-spaced using it's normal sidebearings, with kerning pairs as minimum as possible. I have never designed a text face. :)

I usually have a brief from the client, which may involve some technical constraints and a more or less tightly identified target usage. But the brief seldom specifies the actual style of the typeface, so within the constraints of the brief the first thing I start thinking about is the overall look and feel of the type, what a block of text will look like.

In my limited experience:

I usually start from one or two characters that have been sketched in my notebook or on a scrap of paper (or sometimes a shape that emerged playing in Illustrator. So, I have this one character, and I start building around it. As I make more characters, I might realize that the weight of the stems has to change. Nothing is set from the beginning and the glyph that started it all might not even make it into the finished design.

I do try to pin down the details as quickly as possible, so as not to have to redraw glyphs too often.

@BeauW & mk2

Really? You are embarking on a months-long, if not years-long endeavor and this is all you need to start?
Wow. I'm staggered!

:o) Well, my thought is, "Is this really necessary?"

-=®=-

@terminaldesign

Yes, in my case. Like I said, I'm not a professional typographer. I make type just for my personal pleasure. I don't turn every logotype I've designed into a typeface, of course.

I'm not talking about creating a wide family of text face overhere, just another silly looking typeface you could find scattered on the internet. Being staggered is really unnecessary. :)

@russelm

The answer is 'yes' if you're asking to curious peoples like me. :)

"You are embarking on a months-long, if not years-long endeavor and this is all you need to start?"
:-) I think I'm of the opposite sort; I tend to over-think concepts and have trouble actually getting started. That said, I do think it's a good thing to have at least a semi-solid concept before starting out – at least basic things like intended use / target point size, serif size/shape (if applicable), basic proportions, and something like the "feel" the whole thing should have.
But I'm just starting out designing type, so this process is still very much under construction (and will likely remain so for the years to come ;-) ).

What I think about before I begin creating a typeface often winds up being very different than what I end up with. I start with a trope or gimmick (which can include, and is not limited to, contrast, ductus, serif shape, &c.) but as I carry it through it just doesn't work the way I thought it would, which leads me in new directions I hadn't considered previously.

If you really want to look at my process, I've sort-of blogged my works-in-progress here in the critique forums.
I keep a list of links to them here.

Well, so far most of my fonts have been more of a weekend or two scope.

I've been working on one whole typeface (something appropriate for hobbits, is my current opinion) that has pan-European glyph coverage, regular, bold and italic fonts and it all stated with a capital 'H' I doodled on the metro one day.
I mean, you have to start somewhere, and for me- designing type without a brief, it starts in my notebooks. I don't think I have anything significant to add to the world of text fonts yet, so for me- the more ambitious project I've undertaken- which may last a year, has mostly been an exorcise undertaken to learn all the parts that need to go into a complete typeface and to try and work out an efficient workflow (hint: not the way I've been working)

I often start with a concept that has two or more themes.
What would a typeface look like (and how could I get it to work), if it supported both "idea A" and "idea B"?
These ideas can be aesthetic, technological, cultural, market-driven, serendipity, whatever.
Making the font(s) is a process of working things out.