Well, let's be honest. People on these boards can appreciate the technology behind 3D touch, the A9 chip, and camera, but the masses really don't care. I can't think of a single person that was like "oooh, I need 3D touch." The masses need something more flashy and tangible. Though, that is just my opinion.
3D touch, the A9, and the camera add to the experience, but they don't substantially improve/change it like a larger screen or a new design. I anticipated lackluster response, but what do I know.
I don't think the technology was that compelling this time either. 3D Touch is cool, but it isn't transformative. Look at how many people never use right/control-click on their computers. The A9 chip? Faster and better sure, but does it so dramatically improve device performance that it becomes a "must have"? No. Better camera, sure, but the previous one was "good enough."
I'm still using a 5s. My first iPhone was the original. I upgraded from the original to the 4 and from the 4 to the 5s. Sure, each skipped model offered some compelling features, but in the end I didn't find any of them so compelling that I needed to upgrade. And I think one could fairly argue that the 4s and the 5, for example, offered many more compelling new features over the previous model than the 6s does over the 6.
There's a small contingent of people who always need the newest device, just because. But for "the masses", I think most people upgrade for practical reasons, things like the battery not holding a long enough charge or the latest iOS rendering the device too sluggish. Or a broken screen. All of my friends and business colleagues have iPhones. I only know one person who upgrades annually. The rest just wait until something pushes them to upgrade.
I think Apple would benefit from slowing down the iPhone release schedule. Skip the S models altogether and release a new flagship every 18-24 months. There would be more excitement around each release and the new features would be more compelling. The S model is little more than a minor upgrade of the previous model, yet analysts and journalists, etc. all expect it to perform like a non-S model release.