You will never see a competitor's analytics. You don't need them.
What you can see is every decision they made — what they published, how often, in what format, with what production budget, and crucially what they stopped doing. Decisions made under real constraints are more informative than metrics, because a team that keeps making something is telling you it works for them, and a team that quietly kills a format is telling you it didn't.
Here's how to read that systematically instead of impressionistically.
What public data actually contains
Available on most platforms without any special access: publish dates and times, titles, thumbnails, descriptions, tags where exposed, duration, format, view counts, public engagement counts, playlist and series structure, and the full back catalogue.
Not available: watch time, retention curves, traffic sources, demographics, revenue, what they spent, or what they were trying to achieve.
That's tier 1 evidence — real, verifiable, and outside-only. Enough to reconstruct strategy, not enough to judge outcomes. Keep that line clear in whatever you write up.
Pass 1 · Cadence and commitment
Start with the calendar, because it's the cheapest signal and it's about resources.
Plot publish dates for the last twelve months. You're looking for:
Rate, and whether it changed. A jump from two a month to eight a month means someone approved a budget. A collapse means someone lost an argument, or a person left.
Regularity. Fixed-day publishing indicates a production system. Irregular clusters indicate campaign-driven or opportunistic work. The first is harder to compete with and easier to predict.
Gaps. A two-month silence followed by a format change is one of the most legible patterns in public data: that's a strategy reset, and whatever came after it is what they now believe.
Pass 2 · Format inventory
Categorise every piece into a small set of formats — six to ten buckets, defined by you and applied consistently. Talking head, tutorial, customer story, product demo, event coverage, whatever fits your category.
Then plot format mix over time.
The trend is the finding. A format going from 10% of output to 40% is a bet being increased. A format that appears eight times and then never again was tried and rejected — and knowing what a well-resourced competitor abandoned is genuinely valuable, because it's an experiment someone else paid for.
Pass 3 · Production value
Estimate roughly: single camera or multi, scripted or spoken, studio or location, motion graphics or none, licensed music or library, on-screen talent who's clearly staff or clearly hired.
You're not appraising quality. You're estimating cost per piece, because cost per piece times cadence gives you their approximate commitment — and that tells you which fights are winnable.
A competitor making four expensive pieces a quarter is beatable on volume and responsiveness. One making twenty cheap pieces a week is beatable on depth and craft. You can't beat both at once and shouldn't try.
Pass 4 · Topic map
List the subject of every piece, then cluster.
What you're looking for:
- Their core — the three or four topics that recur most. This is what they want to be known for.
- Their edges — topics tried once or twice. Experiments in progress.
- Their absences — the obvious subjects in your category they never touch.
That last one is where the format gap lives, and it needs interpreting rather than exploiting on sight. An absence is either an opportunity or a warning: they may have tried it before your window, or found it doesn't convert, or be avoiding it for legal or brand reasons. Ask which before you charge in.
Pass 5 · The outlier read
Now, and only now, look at numbers.
Within each competitor's catalogue, find the pieces that beat their own median by a wide margin. Not the highest-view pieces overall — the ones unusual relative to that account's own distribution, which is the same outlier logic you'd apply to yourself.
Then ask what those pieces share. If a competitor's five biggest outliers are all one format on one topic, you've found something they may not have noticed themselves.
Also check for the piece that's an outlier because it was promoted. A sudden spike on an otherwise unremarkable video, especially one with a hard call to action, is often paid distribution rather than organic resonance. Public data can't confirm it, so flag it as uncertain rather than treating it as a content lesson.
Pass 6 · The self-description read
Read their titles, descriptions and thumbnails as a set rather than individually.
Titles reveal keyword strategy and who they think is searching. Descriptions reveal the conversion path — what they link to, and whether the link is the same every time. Thumbnails reveal a house style, and whether one exists at all.
Consistency here is a proxy for organisational maturity. A team with a locked thumbnail system and templated descriptions has a content operation. A team without one is winging it, which is a competitive fact worth knowing.
Putting it together
Six passes produce a one-page reconstruction. Something like:
That last line is the deliverable. Five paragraphs of observation, one sentence you can act on.
What not to conclude
Don't infer revenue. Views are not sales. A competitor with better content numbers may be losing money.
Don't infer intent from absence alone. See pass 4.
Don't copy the outliers directly. Their audience isn't yours, and a format that works for a brand with an established presenter may not survive being lifted without one. Treat it as a hypothesis to pressure-test, not a recipe.
Don't do this once. A reconstruction is a snapshot of a moving target. The value compounds when it's a standing watchlist rather than an annual deck.
Where Acumin fits
Competitor watchlists run the mechanical part of passes 1, 2 and 5 continuously from public data — cadence, format mix, and outliers scored against each competitor's own distribution — and surface changes rather than requiring you to go looking.
The interpretation is still yours, and it should be. What the tooling removes is the reason people do this exercise once a year instead of continuously: it takes a full day by hand and is stale within a month.
Everything it reads here is public and labelled as tier 1. No private data, no scraped analytics, no inferred revenue.
How to use this today
Pick your single closest competitor. Open their channel, sort by oldest, and log the last twenty pieces in a spreadsheet with four columns: date, format, topic, views.
Twenty rows is enough to see the cadence and the format shift. It takes about forty minutes and it will tell you more than the last competitive deck anyone made.
Related: How to research your competitors on YouTube is the platform-specific mechanics. Finding the format gap is what to do with the absences you find in pass 4.