Competitor monitoring that gives you something usable
Competitor monitoring works when their posts and the posts about them are measured identically. Read the complaints and watch the direction, not the position.

Listening to your own brand tells you how you are doing. Competitor monitoring tells you what to do about it, and most of it is public and uncollected.
Most of the second is available and almost nobody collects it.
What you can see about a competitor
Everything public: their posts, the engagement those earned, the impressions the platform reported, and every post mentioning them written by someone else. That last category is where the intelligence is. Their own account is marketing. The posts about them are evidence.
Tell us who to watch and we will set the comparison up, within the Twitter compliance rules.
Measure everyone identically or do not bother
The single rule that makes a competitive benchmark usable: same period, same operator structure, same decision about reposts, same source. A comparison where one brand's figure came from an API and another's came from an estimate is a chart with no meaning. It will still get presented, and it will still be wrong.
Apply the same exclusions to every brand too, even when one of them collides with something the others do not. Note the asymmetry rather than fixing it silently.
The complaints are the product research
Filter a competitor's mentions to replies and read what people ask them for. Recurring complaints are a list of gaps someone else is failing to fill. It costs nothing, it is already written down, and it tells you more than a survey because nobody was asked to be polite.
Sort by impressions and the complaints that spread come first.
Share of voice by reach, not just volume
Divide your impressions by the category total, and separately divide your post count by the category total. A competitor can be mentioned half as often as you and still be seen twice as much, because one large account posted about them. Volume measures how many people spoke; reach measures how many were reached.
Reporting only one of the two is how benchmarks end up flattering whoever posts most.
Watch the direction, not the position
A single month's ranking is noise. Plot it monthly and read the slope. A competitor climbing steadily for four months is a more important finding than whoever happens to be ahead today, and it is the kind of thing that only appears with a consistent method over time.
Keeping it defensible
State the competitor set and the rule behind it. The three closest by size, everyone in a named category, or whoever the brief specified. An unexplained set is the first thing questioned in the meeting, and the fastest way to lose the room. We hand over the rows so the comparison can be audited brand by brand. Tell us the category and we will build it.