There is not a "correct way" for Marathon, like there is not a "correct way" for any other event.
What the most part of you, and, sorry, the most part of coaches, don't understand, is that the most important factor in training is the "individualization". Every person is different from other persons, and every athlete, running the same distance, is different from the other, so the first step is to individuate the basic qualities of everybody, and some time these qualities are not evident, and our task is to discover them.
You seem to forget that Eliud Kipchoge was World Champion of 5000m in 2003, and had PB of 3'33" - 7'27" and 12'46", and his specific qualities were very different from the one of the most part of runners, especially now that many don't have a previous career on track lasting years, but go to marathon directly.
Look at this very simple calculation : between an athlete able running 5000m in 13', and another able running at his best in 13'30", there are 30" of difference for every 5 km, that means a final difference in marathon of 4'10". So, it's clear there are two possibilities only :
a) The athlete is, very simply, WEAKER, and never can reach the same level, or
b) The athlete, who clearly is weaker under the point of view of speed, has more talent in the opposite direction, being more resistant to the specific marathon speed.
The case of Eliud is the first, the case of Sondre is the second.
Sondre NEVER can become faster than Eliud, and working a lot for increasing speed means to make him LESS SLOW, not faster. So, where we can develop his talent ? Making him able to run with continuity long fast runs, including many sectors at marathon pace, or faster, trying to exalt the ability to run for long time at a high percentage of his Threshold.
This is exactly the reason because Karoki was not able to react, and at the end arrived 3 minutes behind. Because, like the most part of Kenyans, he runs very fast the tests, but very slow the recovery, and, after trying a break with few km around 2'50" - 2'53" per km, he needs to slow down, something Sondre didn't do.
Probably, if I was the coach of Eliud, I'd change very little from the training he always uses, thinking of an athlete who had 11 years of continue activity on track at top level. Maybe I could have the idea to ask him some long run faster than what he normally does, maybe not : for deciding, you need to be always with the athlete, analyzing his situation, and deciding what he needs, every time.
However, is a fact that Eliud is not in the norm, like Haile or Kenenisa. For example, in my plans for Kenenisa I didn't follow the same principles I used with Sondre, and in my plans fo Geoffrey Kirui I don't use the same system used by Eliud or by Sondre, but something still different.
Here there is not the competition for who is the best coach : never we can decide, since every coach depends on the results of the athletes he coaches, in the general vision, and these results depend (not only, but in great part) on the talent of the athletes themselves.
So, we can have athletes that are the best, till when another athlete is not able to beat them and/or their records, and coaches who are able to understand the best solution for everybody.