I am not speaking with any real knowledge or fact - just conjecture.
In my 2017 MKZ 3.0T, I often use the regular Nav and then set the same route on my phone in Apple Maps or Waze. I have not used Google Maps on my phone.
Depending on the route, the mapping systems will often set slightly different routes resulting in different times. If there is a major highway blockage, the Nav system will ask you how you want to handle it as you get closer, while Apple and Waze may just route you to a different route - perhaps using local streets that don't have traffic data and might be equally slow. Even if the routes are the same, the timing is different 95% of the time.
Along a long route, it kind of seems like the Nav system takes accidents or blockages anywhere along the route (even 30 or 40 miles away) into account (and timing) immediately, while Apple Maps and Waze don't take it into account until you get closer. Taking it into account when you get closer is probably the better way since the accident or blockage may not even be there once you get there.
Once started on the road, the Nav system time often drops a little almost immediately. And once on the road, the Apple Maps or Waze time typically rises a little. It seems like the time is constantly recalculated as things happen.
I have noticed that the Nav system will typically always have the longer time. It seems like the more stoplights involved, the more difference. I wonder if the Nav programmers are using a more conservative or worst case scenario at how long you might be at a stoplight. For example, about 1 mile from my house I have to make a left turn at a light onto a major highway. If I just miss that left turn, because it is a major intersection, it is about 3 minutes until I can turn left again. So the question is, did the programmers for the different mapping programs allow an average of 30, 60, 90, 120 or more seconds for that turn. Multiple that by a lot of turns and you can get a big difference. Being a programmer in a prior lifetime (not for mapping though), I tend to think about that kind of stuff.
Waze is rather new and I have only tested it over the last few months. I don't like to use it in the real world because of the way it reroutes people onto residential streets, Especially now that school is open again, Waze is rerouting rush hour people off of clogged major highways and onto residential streets - where it is interfering with school bus routes, etc. I have used Waze just to see how it is different from Apple Maps or the Nav system. I have used the Apple Maps timing tests with my 2013 MKZ Nav (no longer own this) and my 2015 MKC (still own this and has Sync 2). Sync 3 in the 2017 MKZ is generally more accurate in timing than the Sync 2 in the 2015 MKC.
As I said, just conjecture. I don't think that it is something that can be "fixed", just noted and allowed for.