You did the competitor teardown. It was genuinely useful. It's now four months old, two of those channels have changed strategy, one has stopped posting, and nobody has opened the document since week three.
Research decays. The fix isn't doing the teardown more often. It's converting it into something that runs continuously and interrupts you only when something has actually changed.
That's a watchlist. It's a different artefact from a teardown, with different design rules.
The two are not the same thing
A teardown is deep and one-off: inventory, formats, outliers, a decision. Covered in how to research your competitors on YouTube.
A watchlist is shallow and continuous: a small set, a few numbers, checked on a schedule, surfacing changes.
Do the teardown first. The watchlist is what keeps it alive. And you can't design a good watchlist without having done the deep pass, because you won't know what normal looks like.
Choosing the set
Smaller than your teardown cohort. A watchlist you check is worth more than a comprehensive one you ignore, and the failure mode is always the same: too many entries, too much noise, quietly abandoned.
Five to eight. Under five and you'll miss category-wide moves. Over eight and there's something happening every week, which trains you to ignore it.
Three types earn a place:
The pace-setter. Whoever in your category is genuinely setting the agenda. Not the biggest. The one others copy.
Your closest comparable. Same rough scale, same audience. This is the one whose numbers are actually meaningful next to yours.
One attention competitor. Different product, same audience. Usually the source of the most surprising signals, because they aren't bound by your category's conventions.
Everyone else can be re-checked at the next teardown.
What to track
Four things. Resist adding more; every extra metric dilutes the signal.
Cadence. Uploads per month, and the direction. A channel going from eight a month to two is telling you something about their internals no individual video will.
Median views. The median, not the mean, over a rolling window. The trend matters far more than the level.
[Engagement per view](/glossary/engagement-per-view). The fairest cross-channel comparison available on public data, because it strips out the advantage of simply being bigger.
[Outliers](/glossary/outlier). Any post that substantially beat that channel's own norm. This is the highest-value item on the list: a competitor's breakout is a natural experiment run at their expense, sitting in public.
What should actually interrupt you
The hard part of a watchlist isn't collecting data. It's deciding what deserves your attention, because a feed of everything is a feed of nothing.
Four events genuinely worth knowing about:
A breakout. Something beat their median by a wide margin. Go and watch the first thirty seconds, read the comments, and ask whether the thing that worked transfers.
A cadence shift. A sustained change in posting rate (up or down) usually precedes a visible strategy change by a month or two.
A new format. Something appearing in their output that wasn't there before. Either they've found something, or they're testing. Both are worth knowing.
A sustained trend change. Median moving consistently over several weeks. Not one good week. The direction holding.
Everything else is noise. A single post doing slightly better than usual is the normal behaviour of a system with a long tail, not a message.
How often
Weekly is right for most categories. Fast-moving consumer categories might justify more; B2B rarely does.
The pattern that works: a weekly skim, a monthly read, a quarterly teardown. The skim is two minutes and answers "did anything break out?". The monthly read looks at trend. The quarterly teardown re-examines the set itself, including whether the right channels are still on it.
Put the skim on a recurring slot. A watchlist without a scheduled moment to look at it decays exactly like the teardown did.
Reading it without fooling yourself
Three failure modes worth naming, because they're common and they're expensive.
Treating their wins as your instructions. You can see what they published and how it was received. You cannot see their retention, their traffic sources, their conversion, or their costs. Their most-viewed video may be the one their CFO wants cancelled. Public data is tier 1 and has a hard ceiling.
Copying the format without the context. A format that works for a channel with a decade of audience trust may not work cold. What transfers between channels is usually structure (how the hook works, how the argument is built), far more than subject or style.
Reacting to every move. A watchlist should change your behaviour a handful of times a year. If it's changing your plan monthly, you've built a machine for chasing, and your own outliers are a better guide than anyone else's.
Where Acumin fits
Watchlists and Competitor Snapshot do the collection on a schedule: the same public data, the same four numbers, with outliers surfaced rather than left for you to find. The Morning Brief on your Overview is the interrupt layer: it's what tells you a competitor broke out this week, instead of you discovering it a quarter later.
Share of voice is computed across the set, which is where the trend gets its meaning: the level is fairly arbitrary, the direction isn't.
The judgement stays yours. Nothing here decides whether a competitor's breakout is relevant to you.
How to use this tomorrow
Take your teardown cohort and cut it to six. For each survivor, write the one sentence: what would I do differently if this changed?
The ones where you can't answer come off. That deletion is the whole exercise: a watchlist earns its value by being short enough to actually read.
Related: How to research your competitors on YouTube is the deep pass this keeps alive. Outliers, not virality is how to read a competitor's breakout once you've been told about it.