So you would rather pay $600 million per launch with 95% chance of success vs $200 million per launch with 93% chance of success?

There is no such thing a 100% chance of success, everyone has a chance of failure. I would not put ESA more than a few % more reliable. Math says your satellite would need to be really expensive for the price difference to be worth it.

Depending on the cost of failure, however for virtually every mission the 600M$ is a winner this is a no-brainer.

That "little" difference has huge huge economic repercussions, cost of insurance, ease of financing, and many other things. Not to mention that a satellite going boom can kill an operator if it doesn't have a backup. This isn't an insurance case where you get your money and try again the week after, building a satellite is a huge undertaking it takes years to build, years to negotiate the contracts for and you have to do it from scratch all over again if it goes boom.

The satellite build costs $200m.

Launch 1 costs $600m: total cost, $800m.

Launch 2 costs $200m: you can build 2 spare satellites for the same cost (assuming you don't pay for the launches that go boom; and even if you do, that still gives you one spare satellite).

Production doesn't work like this all the time, not everyone is the US government.

Building 2 would cost more than double, or it will take twice as long.

There are huge lead times involved, subcontractors with other commitments and other bottlenecks in the pipeline. Not to mention that even if you do manage to build 2 in the same time window at only double the cost you still have the issue with the launch window, a launch accident would most likely ground the provider for a very long time which means you have to find another launch provider.

Spacecom had a launch window from spacex, spacex has a queue of 9-10 more launches already set, there will now be a huge delay and possible cancelation even if Spacecom had another unit they couldn't just launch it next month.

The flip side is that estimating the reliability of a launch vehicle is an inexact science at best given the paucity of actual launches to provide data points (frankly, there was no real way of assessing the reliability of the Falcon 9 other than to note that launch failures or partial failures tend to occur more in the first couple of years of launches of a new rocket type is backed by a reasonable amount of evidence across the history of other launch vehicle programmes)

Given SpaceX's long manifest of scheduled missions, it seems like the people who actually pay for this stuff don't always agree with you. Going with the more expensive option is apparently not a "no-brainer."

Someone already answered about the price, regarding the success rate, there is a much bigger difference than this:

Space X has had 29 launches, 2 failures and 1 partial failure (F9-004) so a success rate of 89.6%.

Ariane on the other hand hasn't had a single failure or partial failure since 2002 so if you compare since Space X started its services it has a success rate of 100%. If you take the whole history of Ariane it has indeed a success rate of 95.2% or 95.4% for the Ariane 5 version but the fact that Ariane has flown 72 consecutive missions without failure (2002 - Present) compared to 14 for Space X (2012 - 2015) says a lot about their reliability.

Where did you take your $600 million and $200 million cost per launch? Ariane costs are around $95 million, but for bigger payloads than what the Falcon 9 can handle.