What would be the reasons for not using UTM parameters?

Note: by [UTM parameters](https://en.wikipedia.org/wiki/UTM_parameters "UTM parameters - Wikipedia") I mean any kind of parameters is used to track website traffic. One can have their own set of parameters and I still count them as using. Or they can use a link shortener to hide the long, ugly and awkward tail.

Occasionally I see a marketing/seeding link that doesn't have UTM parameters, just the bare link. I think some of the reasons are:

- *They don't know about UTM parameters.* Perhaps they are just a regular writer on their domain and not knowing about any sort of web analytics

- *They don't define their success on how impact their websites are on the internet*

- *The benefit of having web statistics isn't worth the effort.* Perhaps the effort of adding the parameters manually (even via link builders) outweighs the benefit

- *Their main sharing platforms have built-in traffic tracking methods.* If they don't run a lots of channels they could just be using a more singular channel pixel or conversions API to provide measurement. If you only have one source of traffic then UTM parameters may be overkill

However, there are cases that I can't assume any reason for not doing so. For example, while the link in [this answer](https://webmasters.stackexchange.com/a/141639/52079) is genuine, I presume that: - The author knows about UTM parameters well - Their success is defined on how impact their websites are on the internet - It's likely that they have their own tool to generate the parameters effortlessly - They have multiple source of traffic

Thus I find no reason for them to not use UTM parameters.

It's possible that the author's company has some methods of analysis that don't really rely on traffic data, like [attribution modeling (MTA) and media mix modeling (MMM)](https://www.youtube.com/watch?v=d5We-HR9_uM "Attribution modeling (MTA) vs media mix modeling (MMM). Whats the difference? # YouTube"). They use probabilistic approach to correlate changes in impressions across channels with changes in revenue. This seems to be only possible for established companies, where they have a dedicated team for expensive and sophisticated data analysis. I understand that this reduces the need of using the parameters, but isn't that the more data the better for them? I don't see why using these models make the benefit become completely futile.

So what would be the reason to not doing so?