If you want an inexpensive build for now, and spread out the cost of upgrades over a year or two it's not a bad idea. I believe that it can be cheaper than buying new if you build on either end of the price spectrum. Either low end on the pricier parts, or buy a top level group set. Because high end bikes add a premium for style and cachet over parts cost, the full group of components is a smaller portion of the bike's price.
When I built mine, generic aluminum frame with generic components, it was just a few hundred dollars at first. Including tools. I already had some wheels and other bits like seat post and saddle, and it was very basic, 1x7 Sora basic. By the time I improved it, it cost about the same as the equivalent you'd see online.
It's very easy if you're mechanically inclined. CF builds have a few extra considerations, mainly applying correct torque but none of it is really difficult. The big gotcha is having incompatible specs on the various pieces. If you're like I was and don't at first really know all the eccentric details like pull ratios vs brakes and derailleur capacities and so on, research and double-check the research.
*edited1: BTW I went more upscale on parts like the headset and bottom bracket, because those remain on the bike for a long time before having to mess with them. Handlebars or stem or seat post, who cares. I think we need to balance being realistic and being discriminating.
*edited 2: We can have different reasons for wanting to build, but it is a good question because sometimes the reason in not realistic. In my case, it was pragmatic and design philosophy. My "road bike" was a starter bike that was universally reviled (at least a little justified) and I wanted a true road bike and didn't want to lay out much for it. Secondarily I didn't want to cheap out in the places where most of these entry-level bikes do, but rather in components that I'd replace eventually anyway. Other people have explained completely different reasons such as wanting to learn the mechanics of their bike, which seems to me to be perfectly reasonable. Expecting to cut our cost in half in the long run though, would be one of the unrealistic ones.
Last edited by wphamilton; 09-11-20 at 09:15 AM.