Recycling some content that I originally posted in another context:
It took me forever to build out my tree until I got my first connection with the wider family tree. And I see a bunch of people in the forum who are clearly frustrated that they haven't been able to link up their own families yet. Plus, I see plenty of evidence (since I've been working on orphaned profiles) of people who entered what they knew, still couldn't make a connection to the wider family tree, got frustrated, and quit.
So I got curious as to just what percentage of the people in our "target" date range (1 AD to 2003, excluding those under 13) are on WikiTree. Here's what I found out:
First, I needed an estimate of the people in the target range. I found a page on the Population Reference Bureau site called "How Many People Have Ever Lived On Earth?" From that, I extracted an estimate of 60,439,585,668 people who have been born since 1 AD. As of a few minutes ago, there were 11,200,619 profiles on WikiTree. That means that, currently, there are WikiTree profiles for approximately 0.02% (0.185319255%, as long as I didn't forget to carry a 2 or anything...) of the people in the target range. Or, to put it another way, there is one WikiTree profile for every 5,000+ people in the target range (5,396).
Now, granted, we don't see unconnected trees with thousands of people, but I assume that that is partly due to the fact that the vast majority of profiles date from the last 200 years, which happens to be where most newbies are going to be putting in information for parents, grandparents, and so on. But even so, the odds of finding a match already on WikiTree are pretty small, and I assume that most people, like me, have to get their tree up to a couple of hundred people before they can make a connection. No wonder they get frustrated.
I still haven't managed to find a connection by working backwards, even though that's what everybody tells me to do. Nor did starting from past notables with the same last name and working forward give me any connections (even though it was kind of interesting).
What works for me is working "sideways" within what I call the "golden age for genealogists": from 1841 (when the first censuses were made -- in Canada and the UK, anyway) to between 1903 and 1938 (when I start to bump into privacy limits on public records). I have found it amazingly easy to chase all kinds of rabbits and track down 2-4 generations in the families of spouses of cousins and so on. Every connection that I've made to the wider tree (4 so far) has been through lateral connections like that.
(Granted, it would probably be a lot easier for me to work backwards if my ancestors had been rich and famous, but as I look into the details of the places they lived, I find comments like "[Neighbourhood] was notorious as one of the most crowded and poverty-stricken parts of [City]" with disturbing regularity...)
So my advice is to just keep chugging away. I have also found that it helps to take a break from my own family every once in a while, and spend some time working on a challenge, or adding somebody notable (or just plain interesting) who hasn't been entered in WikiTree yet, or building a free space page for an emigrant/immigrant ship, or something else that I find interesting, but that I can do for fun, rather than because I have any specific goal or deadline.