The Competitor Video Analysis Framework for Brands
A repeatable competitor video analysis framework: inventory, cadence, hook taxonomy, engagement ratios, and turning findings into production briefs.
Every quarter, someone on a brand team screenshots a competitor's viral video into Slack with the message "we should do something like this." That is not competitor analysis. That is competitor envy, and it produces exactly one deliverable: a worse copy of a video that already worked for someone else, shipped a month late.
Real competitor video analysis is boring in the way useful things are boring. It is an inventory, a set of ratios, and a written answer to one question: what do their numbers prove about our shared audience that we can act on differently? Done properly it takes about three hours per competitor per quarter, plus a 20-minute weekly touch, and it will reshape your content calendar more than any brainstorm. Here is the framework I run, step by step.
Step 1: Pick the right five accounts
Most teams analyze the wrong competitors — the ones sales complains about rather than the ones competing for the same attention. Build a set of five:
- Two direct competitors (same product, same buyer).
- Two audience competitors (different product, same viewer — for a sleep supplement brand, that might be a mattress company and a meditation app).
- One aspirational account from an adjacent niche whose formats your niche has not adopted yet. This is where your actual edge comes from, because everything your direct competitors do is visible to everyone.
Revisit the set quarterly. Accounts that stopped posting get swapped out; new entrants that grew fast get swapped in — fast growers are more informative than large incumbents, because their tactics are what worked recently.
Step 2: Inventory the last 90 days
For each account, pull the last 90 days of video posts into a sheet. Per video, record: date, platform, format type, duration, hook type (first 2 seconds), topic, on-screen text yes/no, and the visible engagement counts (views where shown, likes, comments, shares if exposed). This is an hour of manual work per account, and resist the urge to skip it — the sheet is the asset. Everything after this step is just arithmetic on it.
Then compute the ratios that actually mean something:
| Metric | Formula | What it tells you |
|---|---|---|
| Cadence | posts / week | Their production capacity and commitment |
| Consistency of returns | median views vs mean views | Mean >> median = hit-driven; close = systematic |
| Engagement rate | (likes+comments) / views | Audience quality, comparable across account sizes |
| Comment-to-like ratio | comments / likes | Conversation-provoking vs passive content |
| Format concentration | % of posts in top format | Whether they found a formula or are still searching |
| Hit pattern | what top 10% share | The brief for what to test yourself |
The single most revealing number is median-vs-mean. An account with 2M mean views and 40K median views is buying lottery tickets; an account with 300K mean and 220K median has a machine. You want to study the machine, even when the lottery account is bigger.
Step 3: Decompose their outliers, not their averages
Take each competitor's top 10 percent of videos and break them down structurally: the hook mechanic, the pacing (count the cuts in the first 10 seconds), the audio choice, the caption style, the CTA, and the comment section's actual content — comments tell you why it worked in the audience's own words, and they are the most underused data source in competitive research.
Do the same for their bottom 10 percent. Failures are as informative as hits: if their polished studio content consistently underperforms their phone-shot content, you just learned something about the audience's taste that applies to you too. This decomposition step is the same skill as breaking down trends — the trend reverse-engineering approach applies directly, pointed at a competitor instead of a trend.
Step 4: Map the white space
Lay your inventory sheets side by side and look for what is missing. White space comes in four flavors:
- Format gaps — formats working in your aspirational account's niche that no direct competitor uses. Highest-upside, lowest-risk finding.
- Topic gaps — questions appearing repeatedly in competitors' comments that nobody has made a video answering. Free content briefs, pre-validated by demand.
- Cadence gaps — days, platforms, or durations the whole set neglects. If everyone posts 30-second Reels, a weekly 3-minute YouTube piece owns unclaimed depth.
- Quality gaps — a format everyone does badly. If every competitor's product demos are boring, an excellent demo format is a moat.
The output of this step should be written: five to ten one-line hypotheses, e.g. "No one in the set answers sizing questions on video despite ~40 comments/month asking; a weekly sizing-QA short should convert high-intent viewers." Hypotheses, not observations — each one must be testable with a post.
Step 5: Turn findings into production, fast
Analysis that does not change next week's calendar is a hobby. Each surviving hypothesis becomes a brief with a format, a hook, and a success metric, and goes into production immediately. This is where the economics have shifted: testing a competitor-inspired format used to mean a shoot, which meant most hypotheses died in the backlog. Generating the test with an AI video generator means a hypothesis can become three format variants in an afternoon, and multi-scene formats map onto existing workflows with your product swapped in. Run each hypothesis for three to five posts before judging — single-post verdicts are noise.
File the losers in your team's reference library alongside the winners; a structured video swipe file is the natural home for both your competitor decompositions and your own test results, and it stops the next quarter's analysis from starting at zero.
The weekly touch and the quarterly rebuild
The full framework runs quarterly. Weekly, spend 20 minutes: scan the five accounts, log new posts into the sheet, and flag anything anomalous — a format change, a sudden hit, a cadence spike (often the tell that a competitor hired an agency or adopted AI production). Pair this with your broader trend analysis loop; competitor signals and trend signals feed the same calendar, and the fit filters are nearly identical.
One warning, from experience: competitor analysis has a gravitational pull toward convergence. If you only ever test what worked for competitors, the whole niche's content collapses into one grey average. The framework's job is to find validated divergence — the white space their numbers reveal — not to photocopy their greatest hits. Keep the aspirational account in the set precisely to pull you out of the niche's orbit.
FAQ
How many competitors should I analyze?
Five is the sweet spot: two direct, two audience-overlap, one aspirational from an adjacent niche. Fewer misses the pattern; more turns the quarterly inventory into a full-time job and the extra accounts rarely change the conclusions.
How often should competitor video analysis run?
A full inventory-and-ratios pass quarterly, plus a 20-minute weekly scan to log new posts and catch anomalies. Quarterly is frequent enough to track strategy shifts and infrequent enough that the 90-day windows contain real sample sizes.
What if competitors' view counts are hidden?
Use engagement counts as proxies — likes and comments are visible almost everywhere, and comment-to-like ratios plus posting cadence still expose most of the strategic picture. Median-vs-mean patterns hold on likes nearly as well as on views.
Is it okay to copy a competitor's video format?
Formats are not ownable, and everyone builds on everyone. The line I draw: adopt the structure, never the script, the specific creative, or the brand voice. Better still, adopt formats from adjacent niches your direct competitors have not touched — same benefit, zero convergence.
How do I analyze competitors who barely post video?
That absence is the finding. A competitor set with weak video presence means the audience's video demand in your niche is unserved — which makes video your cheapest available land grab, and makes the aspirational account in your set the more important reference.
Run the inventory this week — five accounts, one sheet, ten hypotheses. Then close the loop: turn the best hypothesis into three test videos with the AI video generator before the analysis goes cold. Free credits daily.