1. I have not visited the insurance agent's website.
2. He is not in my contact list. (And I've not given FB access to my contact list).
3. I've never been to his office.
I have visited the website of the national insurance company that he represents, so if he was my local agent I could see them inferring a connection. However, he is not my local agent. He's from the city I lived in two moves ago--a city I lived in before FB existed.
As far as I know, I have no friends that use him as an insurance agent. I did switch to that insurance company on the recommendation of friends, but that was three moves ago, when I lived in a completely different state. Those friends and I both moved to this state at about the same time, but to different cities, and we switched to different agents when we moved.
It was rather creepy.
1. The individual visited the doctors website, which has a Facebook tracking pixel. This pixel is associated with the doctors page.
2. Contact lists. The user has given FB access to their contacts and the doctor is there.
3. The patient used the doctors free wifi. The user browsed a website which has tracking information that leaks to a third party some identifiable information along with the IP. The third party knows that IP belongs to the doctors office, so they know you are connected. This data is sold back to facebook via one of their DMP (data management platform) partners.
I could go on. Lots of ways leak information without it being purposeful. Ad blockers also don't necessarily solve this problem as tracking can be done server side as well.