What metrics should you track for each competitor?
Four: price and its direction, hiring volume, shipping cadence, and repeated complaints. Everything else is a number you will collect faithfully and never once use.
Most competitor metrics exist because they are easy to collect, not because they are useful. That is a different criterion than the one you want.
Four are worth your time.
Price, and its direction
Not what they charge. Which way it is moving, and how often.
A single price is a fact. A price that has dropped three times in eighteen months is a story, and it says the growth is not coming and somebody is getting nervous. A price that has risen twice with no new features is a company testing what the market will absorb.
So log every change with a date. The log is the metric. The current number is just the latest row.
Hiring volume by function
Count open roles, grouped by function, over time.
This is the most predictive number available, because staffing precedes shipping by months. A company that has gone from one open role to nine has decided something. A company that had twelve and now has zero has frozen, and a freeze is usually not voluntary.
The function matters as much as the count. Nine engineering roles and no sales roles is a build. Nine sales roles and no engineering roles is a push. Those are completely different futures and the number alone will not tell you which.
Shipping cadence
How often does anything actually change in their product.
Count changelog entries, or docs updates, or new pages appearing. Monthly. The absolute number does not matter much, because different companies count releases differently. The trend does.
A competitor whose shipping has quietly halved over two quarters is a competitor in trouble, or a competitor doing a rebuild that will produce nothing for six months. Both are windows.
A competitor whose shipping has doubled has money and urgency, and you should expect to see the result within a couple of quarters.
Complaint patterns
Not review score. Review score is nearly useless, because it aggregates everything into a number that hides the only interesting part.
What you want is the count of the recurring complaint. How many people this quarter said the same thing about their onboarding, their support, their billing.
If that number is climbing, they have a structural problem getting worse, and structural problems getting worse are the ones that eventually move customers.
The test for anything else
Before adding a metric, do this out loud.
Say what you would do if it doubled. Then say what you would do if it halved.
If both answers are the same, or if both answers are nothing, the metric is decoration. It exists to make the spreadsheet look serious.
Their follower count went up. What do you do? Nothing. You feel slightly worse and then you carry on. That is not a metric. That is a mood, and you can get one of those for free without a dashboard.