There are many varying definitions of a minimum viable product. For an e-commerce startup, minimum viable product might be as simple as a basic landing page, just enough to test if people are interested in buying a specific product online (in fact, that was what Zappos did to find out if people were ready to buy shoes online). In other cases, minimum viable product can become much more complex, e.g. the original version of iPhone is considered by some to be a minimum viable product as well.
So what is a minimum viable product when it comes to web applications? Is it possible to test your assumptions without putting out some code? Sometimes yes, but hey, can you imagine how assumptions behind Twitter, LinkedIn or Facebook could have been validated without launching an actual product? I cannot.
See what the first Twitter home page looked like in 2006: http://www.mediabistro.com/alltwitter/first-twitter-homepage-design_b13740. It started as an app for publishing SMSs users shared with their followers, hence the limit of 140 symbols in one tweet. Could it have gotten to where it is today by only talking to potential customers and not actually developing functionality (no matter how basic it was)?
The first edition of LinkedIn was launched without many of the features we’ve gotten used to today. There was no way to endorse connections for skills, no way to recommend connections or see a list of recommended connections, and one could only connect to an other person knowing the other person’s email address. I’m not even mentioning groups, jobs or other advanced functionality. Everything was developed after the initial assumption was validated. Check out this screenshot of the first release: http://hitenism.com/linkedin-mvp/.
Facebook’s MVP was very similar to that of LinkedIn: No apps, no news feed, no timeline, no videos, no status updates, and no pages. Just a bunch of interlinked user profiles. All additional functionality was added gradually as the project grew: https://blog.shareaholic.com/happy-facebook-ipo-day-10-screenshots-of-the-old-facebook-designs/.
No focus groups will give you the quality of information you can get from observing real users’ interactions with your app. Building a product with basic functionality is essential for learning and measuring the achievement of progress.