Competitor Content Analysis: A Director-Level Method, Not a Bookmark Folder
How senior analysts reverse-engineer competitor short-form: hook timing, beat structure, cluster patterns. Anchored to Kendall Hope Tucker (Ramp), Rachel Karten (Link in Bio), Lexie Barnhorn (Notion), Paddy Galloway, Jenny Hoyos, Mitra Mehvar (Buffer), and Metricool 2026.
By Bell Chen, founder. Last updated May 19, 2026.
Kendall Hope Tucker, Ramp's Head of Creative Experimentation, told Marketing Brew on October 22, 2025 (marketingbrew.com) that the Brian's Office campaign was built on a single director-level decision the team made before filming, per Tucker: "Accountants have been using the same software for 30 years, they're not looking for alternatives. So we're like, how do we make that pain feel visceral?" The answer in frame one was a man inside a transparent glass box in Flatiron Plaza surrounded by paper receipts. Per Rachel Karten's Link in Bio breakdown of the campaign (milkkarten.net), the Andy Buckley cameo cut at @ramp.com/video/7561836281752194334 (tiktok.com) carried 181.9K likes, and the campaign cleared roughly 112 million combined views. The reason Tucker's team won at that scale was not that they spotted a trend. It was that they reverse-engineered the visceral-image-in-frame-one pattern from Cluely's Brian's Office series before Ramp shot a single frame, then re-ran that pattern with a higher-budget production. That is what competitor content analysis actually is when it works. It is the discipline of identifying the mechanic underneath a winning post and rebuilding that mechanic inside your own brand. Saving competitor posts to a bookmark folder and getting inspired by them is the opposite of that discipline.
This page documents the analysis method I have used to break down competitor content for three accounts in 2026: my own small B2B product accounts, a friends-of-the-house DTC skincare brand, and several creator-deal evaluations where the cost-to-replicate math depended on getting the format mechanic right. Every claim about hook timing, beat structure, cluster patterns, and director-level analysis in this page is attributed to a named operator (Rachel Karten, Kendall Hope Tucker, Roy Lee, Lexie Barnhorn, Jenny Hoyos, Paddy Galloway, Mitra Mehvar, Juan Pablo Tejela), to a named study (Metricool 2026, Buffer 2026, Socialinsider 2024), or rendered as a clearly disclosed fictional worked breakdown. The methodology runs in a notebook plus a screen grid. The tool is not the work.
What director-level analysis actually means
Most brand teams do not run competitor analysis. They run competitor admiration. They bookmark posts they like, share them in Slack with this is cool, and reproduce surface aesthetics two weeks later. The reproduction is shallow because it copies what is visible and misses what is structural. The visible part is the aesthetic: the color grade, the on-screen text, the trending audio. The structural part is the director's actual decisions: where the hook lands in time, how the beats progress, what the camera does at the transition, when the payoff arrives.
Lexie Barnhorn, who runs social at Notion, made the structural point in CreatorIQ's Earned podcast Episode 98 (creatoriq.com) when she described how the @notionhq workflow-timelapse format (the one that cleared 59.2K likes on the most-shared version) worked. The format was not new. The structural decision was that the first frame showed a stack of half-finished documents, a desk, a half-empty coffee cup, and a closing laptop, in a specific arrangement that read, in under one second, as this is what a workday looks like before the tool. The reveal was not the tool. The reveal was the recognizable mess the viewer already lived inside. Reading a Notion post as cool timelapse, do a timelapse misses the decision. The decision was about the opening frame's recognizability, not about the timelapse technique.
Paddy Galloway, who consults on YouTube strategy for MrBeast, Logan Paul, and Mike Tyson (named in his Colin and Samir interview (colinandsamir.com) and across his published vidIQ profile), put the analyst's posture directly when he described how he watches competitor videos for clients. Galloway watches every video at least three times: once at full speed for the experience, once with the sound off to test the visual carry, and once frame-by-frame on the first three seconds. The three-pass method is the working version of director-level analysis. The first pass tells you whether the post worked at all. The second pass (the mute test) tells you whether the post worked because of the captions and visuals or because of the audio. The third pass tells you whether the hook was earned by an opening frame or by a delayed reveal.
Jenny Hoyos, who runs a YouTube Shorts channel that has cleared 10 million views per video, named the mute-test discipline in her vidIQ profile interview (vidiq.com). Hoyos said the mute test is how she checks her own scripts before publishing, per Hoyos: "If the video is good, the comments will come. If the video isn't good, asking for them won't help." The same test applies to competitor analysis: if a competitor video does not carry meaning at zero volume, the post is leaning on audio (trending sound, voice-over, music) for its work, which is a copyable surface element. If the video does carry meaning at zero volume, the post is leaning on visual structure (composition, expression, on-screen text, motion), which is the part you have to reverse-engineer to replicate.
Mitra Mehvar, who runs social for Buffer, wrote in February 2024 that the Buffer team treats the structural-versus-surface distinction as a measurement principle, per Mehvar: "If a metric doesn't change what we do next, it doesn't belong in the report." Translated to analysis: if a competitor observation does not change a production decision on your next shoot, the observation does not belong in your competitor doc. They use trending audio does not change a decision. Their first frame is a recognizable mess that reads in under one second, and ours is a product shot changes a decision. The competitor doc that survives the next month is the one filled with the second kind of observation.
The Metricool 2026 Social Media Study (metricool.com), built on 39,762,999 posts across 1,059,949 accounts, provides the platform-level benchmarks that make competitor numbers readable in context. Metricool CEO Juan Pablo Tejela summarized the 2026 baseline plainly, per Tejela: "Reels reach is down and algorithmic overcrowding is real." For competitor analysis, Tejela's line matters because it changes what counts as a strong post. A competitor's Reel at 50,000 views in 2023 looked above-average; the same post at 50,000 views in 2026 may be top-decile against a depressed reach baseline. The analysis has to absorb the platform shift before it tries to extract the format lesson.
The five-pass competitor breakdown
Pass one, the full-speed pass
- When / duration
- 5 to 8 minutes per video
- Tools
- a quiet room, the video, a notebook
- Deliverable
- a one-sentence first-impression opinion (this worked because of X)
Watch the video at normal speed, end to end, without notes. Form a one-sentence opinion: this worked because of X. Write the sentence down before any other pass. The first-impression sentence is the working hypothesis that the rest of the analysis tests. Most analysts skip this step and go straight to frame-by-frame, which produces an analysis untethered to whether the post worked at all.
Pass two, the mute pass
- When / duration
- 5 to 10 minutes per video
- Tools
- video player with sound off, captions toggled per the post
- Deliverable
- a separate one-sentence muted opinion to diagnose audio dependency
Watch the video again at normal speed with the sound off. Captions on if the video has them, off if it does not. Form a separate one-sentence opinion: muted, this works because of X, or muted, this does not work. The delta between pass one and pass two is the diagnostic. A video that works at full speed and does not work muted is leaning on audio. A video that works in both passes is leaning on structure. Paddy Galloway's three-pass discipline names this directly: the mute test is the test of whether the format is replicable without the original audio.
Pass three, the first-three-seconds pass
- When / duration
- 8 to 15 minutes per video
- Tools
- frame-by-frame scrubbing (QuickTime, CapCut, or browser dev tools)
- Deliverable
- a frame log naming what is on screen at frame one, 15, 30, 60, and 90
Scrub frame-by-frame through the first three seconds. Note exactly what is on screen at frame one (the cover or first shot), at frame fifteen (roughly 0.5 seconds), at frame thirty (one second), at frame sixty (two seconds), and at frame ninety (three seconds). The first-frame note is the most important. Tucker's Ramp glass-box decision was a frame-one decision, not a script decision. Roy Lee's Cluely Brian decision was a frame-one decision; the recognizable office-comedy visual landed before the first line was spoken. Most strong short-form videos win or lose in the first frame, and the analyst who skips this pass cannot tell you why.
Pass four, the beat-map pass
- When / duration
- 10 to 20 minutes per video
- Tools
- video player with timestamp scrubbing
- Deliverable
- a five-beat map (hook 0-3s, setup 3-8s, tension 8-18s, payoff 18-25s, CTA last 2s) with timestamps
Map the full video onto a beat structure: hook (0 to 3 seconds), setup (3 to 8 seconds), tension (8 to 18 seconds), payoff (18 to 25 seconds), CTA (last 2 seconds). The exact durations vary by platform and by format, but every short-form video has these five beats, and the analyst's job is to identify the timestamp at which each beat starts and ends. The beat-map is the document that becomes a producible script. A hook at 0.8 seconds with a named-number open, setup at 2.5 seconds with a visual cut, tension at 8 seconds with the on-screen text changing, payoff at 17 seconds with the founder face on camera, CTA at 27 seconds asking for a save is a reproducible recipe. Their video is fast-paced is not.
Pass five, the cluster pass
- When / duration
- 5 to 10 minutes per video
- Tools
- the previous 14 video breakdowns, a synthesis spreadsheet
- Deliverable
- a cluster role label (which archetype this video belongs to and which other videos share that archetype)
Compare this video to the other four to fifteen videos in the competitor's strong post set (the posts that beat the account's median by 3x on saves per reach or shares per reach). Identify which structural decisions repeat across the cluster and which are one-offs. The repeating decisions are the format archetype; the one-offs are the noise. Tucker's Ramp campaign is a cluster of dozens of variations on the glass-box visual; the archetype is the visceral-pain-in-frame-one decision, not any single video. Lee's Cluely Brian series is a cluster of episodes; the archetype is the recognizable office-comedy character, not the specific bit. The cluster pass is the pass that distinguishes director-level analysis from single-video admiration.
The output of all five passes for one video is a one-page document with five sections: the first-impression sentence, the mute sentence, the first-three-seconds frame log, the beat-map with timestamps, and the cluster role (which archetype this video belongs to and which posts share that archetype). For fifteen videos across five competitors, the output is fifteen one-page docs plus one synthesis document naming the five to eight repeating archetypes across the set.
What good looks like (a worked competitor breakdown)
The numbers below are realistic but redacted from the shape of breakdowns I have run on friends-of-the-house brand accounts and on creator-deal evaluations in 2026. The brand name, the post titles, and the competitor identities are fictional, calibrated against Metricool's 2026 saves-per-reach benchmarks and against the operating shape of DTC skincare on Instagram and TikTok. Treat this as a worked example, not a case study.
Auditing brand: Vespera Skin (fictional sample DTC skincare, $4M ARR, 22,000 Instagram followers). Competitors in scope: three named DTC skincare brands at similar price band, three videos per competitor selected as 3x outliers on saves per reach. Auditor: Bell, sample analysis.
The nine-video grid produced four cluster readings. Comp A founder routine (sink edition): founder face side angle, works muted, 0.6s named-number opener ("3 products"), founder routine, named-number opener. Comp A ingredient explainer: product close-up, works muted, 1.2s text overlay ("This $14 product"), ingredient explainer, named-price opener. Comp A UGC repost transformation: creator face week-1 vs week-6 split, works muted, 0.3s split-screen reveal, transformation, visual-only opener. Comp B founder routine (kitchen edition): founder face full body, audio-dependent, 1.8s voice-over name drop, founder routine, voice-led opener. Comp B behind-the-scenes warehouse: wide shot packing line, audio-dependent, 2.4s narration, BTS, narration-led. Comp B customer testimonial: customer face direct address, works muted, 0.5s named-number ("I tried 7"), testimonial, named-number opener. Comp C quiet-luxury aesthetic: product on stone slow-motion, works muted, 0.2s visual composition only, aesthetic, no-narration opener. Comp C ingredient explainer (peptides): founder face talking head, audio-dependent, 2.0s claim opens, ingredient explainer, voice-led. Comp C UGC repost: creator face brand-edited, works muted, 0.4s named-number ("$28 toner"), UGC, named-number opener.
The cluster reading: across three competitors and nine 3x-outlier posts, the durable winning pattern is visual-or-named-number opener inside the first second, audio-optional, founder-face decorative. The replicable next-month format for Vespera Skin is the founder-routine or UGC-repost variant of the same pattern, not the audio-led founder routine that Comp B uses (which requires voice-over talent Vespera does not have).
What this changes for Vespera Skin next month: every Reel opens with a named-number in the first second; audio is optional (all scripts read at zero volume); the founder face is decorative, not load-bearing. The metric expected to move is saves per reach, with a kill criterion of 0.40 percent at the May 22 mid-month check. The breakdown took six hours of focused analyst time (Bell), produced 9 one-page docs plus one synthesis, and replaced what would otherwise have been three months of let us try a thing and see in Vespera calendar.
Where competitor analyses break
Failure mode one: the bookmark folder. The team saves competitor posts to a Slack channel or Notion database, accumulates 200 saved posts over three months, never opens the database again, and reproduces a couple of surface aesthetics from memory. The fix is the five-pass rule: a saved post that has not been processed through pass one through pass five inside one week of saving does not count as a competitor observation. The bookmark folder is not analysis. It is a staging area for analysis that mostly never happens.
Failure mode two: copying the surface. The team identifies a competitor pattern (trending audio, color grade, on-screen text style) and reproduces the surface element while missing the structural decision (first-frame recognizability, beat timing, named-number opener). The output reads as a flat copy that performs below the original. The fix is the mute test: if your reproduction does not carry meaning at zero volume but the original does, you have copied the audio layer and missed the visual layer. Rebuild with the visual layer load-bearing.
Failure mode three: analyzing competitors who are not actually winning. The team analyzes a competitor's posts that look impressive (high follower count, visible production budget) without checking whether the specific posts they are analyzing actually performed in the competitor's own data shape. A post at 50,000 views from an account with 500,000 followers is below median; the analyst treating it as a winner is studying a competitor's failures. The fix is to use the 3x-outlier-on-saves-per-reach rule from the 30-day audit playbook: only analyze posts that beat the source account's median by 3x on the primary intent metric. The strong posts are the ones worth reverse-engineering; the average ones are not.
Failure mode four: skipping pass two or pass four. Of the five passes, the first-impression pass and the cluster pass are the ones analysts default to. The mute pass and the beat-map pass are the ones that get skipped because they feel mechanical. The skipped passes are where the audio-versus-structure diagnosis happens (pass two) and where the reproducible recipe gets written (pass four). An analysis that skips either pass produces an opinion document the creative team cannot build against. The fix is to time-box each pass and treat the full sequence as one work block, not five optional steps.
A counter-perspective worth flagging
Several agency analysts I respect have argued in public that the five-pass method is overbuilt for trend-driven niches (fashion, beauty, lifestyle) where the lifecycle of a strong format is two to four weeks, not three months. The honest version of their argument: the time spent on pass three (first-three-seconds frame-by-frame) is wasted when the format will be dead before the analysis ships into a producible script. Their alternative is a 15-minute weekly trend-spot scan that names the current format and ships a script in the same week. Rachel Karten's November 18, 2025 piece (milkkarten.net) named the underlying platform shift, per Karten: "The FYP ate the follower." The trend-spot version of the analysis is the right move for fast-cycle niches where the recommendation surface is rewarding novelty over format-stable production. The five-pass version is the right move for slow-cycle niches (B2B SaaS, finance, education, durable consumer goods) where a strong format archetype lasts six to twelve months and the reverse-engineering work compounds. Both can be true. The choice depends on the niche cycle, not on which methodology is universally correct.
Metrics to extract from each competitor video
Saves per reach is the primary intent metric on Instagram and the closest proxy for cluster-level signal. A 3x-outlier on saves per reach is the threshold for inclusion in the strong post set; a post that does not clear that threshold is not worth analyzing because the synthesis from it will not change next-month production.
Sends per reach is the secondary intent metric and the strongest distribution-multiplying engagement signal on Reels per Adam Mosseri January 8, 2025 video. A post with high sends per reach but low saves per reach is built for in-network distribution rather than reference value; the cluster reading should distinguish the two.
Completion rate (TikTok) and watch time (Reels) anchor the retention diagnosis. A post that retains viewers all the way through is leaning on hook-plus-tension structure; a post with high engagement but low completion is leaning on early visual hook with no back-half payoff. Both are valid; the cluster reading should name which one the competitor is running.
Profile visits per reach isolates the discovery cluster from the engagement cluster. Posts driving profile visits are the discovery-engine archetypes; posts driving saves are the reference-value archetypes. The competitor analysis should separate the two because they fund different production decisions on the brand side.
Follows from the post (when visible) is the rarest metric to extract from public surfaces but the most directly attributable. A post driving follows is the format the competitor is winning on; the cluster pass should give that archetype extra weight in the synthesis.
Where a planning-first tool fits
Most of the breakdown runs in a notebook, a screen-grid, and a stopwatch. The director-level analysis pass is the one place a tool earns its slot, because the manual version (frame-by-frame scrubbing across 15 videos, transcribing beat timestamps by hand) costs roughly five to seven hours per breakdown. Tools that automate the beat-mapping and the first-three-seconds frame log compress the analyst time from a full working day to two to three hours. Superdirector is one option among several (Foreplay, VideoToolkit, and a hand-coded combination of Whisper plus FFmpeg all work for the same step). The cluster judgment in pass five is the work no tool replaces. The tool can produce the per-video beat-map; the analyst has to decide which beats repeat across the cluster and which are noise. That decision is the analysis.
Sample Execution Plans
These example scripts show what this use case looks like once strategy turns into an actual production brief.
Across matched samples, the use case is translated into scripts of about 4 beats, repeatable setups in Darkened bedroom/studio space and Home office desk and Minimalist living room corner, and reference-backed decisions from linusekenstam and prettylittlemarketer.
Script examples
The Glossier Billion-Dollar Blueprint
Glossier turned their everyday customers into an unstoppable sales army, building a billion-dollar empire off their backs.
Discover how Glossier built a billion-dollar empire using community-led affiliate marketing, and how modern founders can replicate it without burning out.
Reference source (curated reference): here’s how Glossier turned their customers into a billion-dollar sales force (and what it actually means for your brand in 2026) 👀💰📣 most brands think affi… by @prettylittlemarketer
The Conversion Truth: Beyond Viral
The real reason your Reels aren't closing deals (It's not the algorithm)...
A high-retention, music-driven hook challenging the myth that viral reach is the primary metric for service-based revenue.
Reference source (curated reference): 1) A confused lead will not buy If a lead cannot immediately place who you are and who you help - they’ll place you in their mind as “helpful,” but not an “ind… by @thesocialbungalow
The $60 Cyber-Studio Stack
My exact $60 AI filmmaking stack
A high-octane visual breakdown of how a $60 AI software stack transforms a solo creator's bedroom into a cinematic, cyberpunk blockbuster.
Reference source (curated reference): Kanye is going viral in China, it took one guy $60 and 3 hours to make this. by @linusekenstam
Production cues
- The examples are intentionally executable: roughly 4 beats and a clear hook up front.
- The production setups repeat around Darkened bedroom/studio space and Home office desk and Minimalist living room corner.
- Each sample keeps a direct link from reference video to script so the workflow remains auditable instead of purely conceptual.
Adaptation notes
- Use the sample hook as a structure reference, then replace the subject matter with your own offer or audience pain.
- Keep the setup light enough to reproduce inside your normal weekly shoot day.
- Treat the linked analysis as the creative reference and the script as the execution layer you customize.
Disclosure by Bell Chen, founder of Superdirector: the brand-profile and competitive analysis features mentioned in this piece are part of the product I build. Methodology and benchmarks here are sourced from the named operators and reports cited inline; treat the tooling note as one input among several.
Frequently asked questions
How is this different from just bookmarking competitor posts I like?
Bookmarking captures surface admiration; the five-pass method captures structural decisions. A bookmark folder of 100 saved competitor posts produces zero next-month production changes if the posts have not been processed through pass one through pass five. Mitra Mehvar Buffer measurement rule (https://buffer.com/resources/social-media-team/), per Mehvar, applies: "If a metric doesn't change what we do next, it doesn't belong in the report." Translated to analysis: an observation that does not change next month production decision does not belong in the competitor doc. The five-pass method generates those observations on purpose.
How many competitors should I analyze?
Five is the working number. Three peer-stage accounts plus one larger benchmark plus one indie at smaller scale than your launch budget. Fewer than three produces too narrow a sample; more than seven dilutes the synthesis across too many surface variations. The five-account rule is the same one I use in the brand launch sprint playbook, because the format-library work is the same work in both cases.
How often should I run a full competitor breakdown?
Monthly is the default cadence for most brands. The cycle: 15 videos across five competitors, six to eight hours of focused analyst time, one synthesis document fed into the next month content calendar. For fast-cycle niches (fashion, beauty trends, AI tooling), the cadence compresses to bi-weekly. For slow-cycle niches (B2B SaaS, finance, durable consumer goods), it extends to quarterly. The cluster signal needs at least three months of synthesis documents to start surfacing format archetypes that compound across breakdowns.
What metrics should I extract from each competitor video?
Saves per reach, sends per reach, completion rate, profile visits per reach, and follows from the post if visible. The save and send metrics are the strongest non-trivial intent signals on Instagram and the closest proxies for cluster-level signal. Comments are a secondary signal because comment-bait inflates the count without inflating the intent. Rachel Karten's March 11, 2024 measurement piece (https://www.milkkarten.net/p/how-to-measure-success-on-social-media), per Karten, named the discipline directly: "Measuring everything is the same as measuring nothing. Pick the two or three numbers that change what you'd do tomorrow." The competitor version of Karten's rule: pick the two metrics that distinguish a 3x outlier from a median post in the same account, and run the analysis only on the 3x outliers.
Is reverse-engineering the same as copying?
No, if the work is done at the structural level. Kendall Hope Tucker's Ramp Brian's Office (https://www.marketingbrew.com/stories/2025/10/22/ramp-viral-livestream-brian-baumgartner) campaign reverse-engineered the office-comedy archetype from Cluely's Brian series; the reproduction was a different character, a different setting, a different production budget, and a different narrative arc. The archetype was borrowed; the surface was original. Copying is when the reproduction shares both the archetype and the surface. Reverse-engineering is when the reproduction shares the archetype and rebuilds the surface to fit the new brand. The five-pass method is built to isolate the archetype from the surface so that the surface can be rebuilt.
What if my competitor posts are mostly junk?
Then the competitor is the wrong reference. The five-account rule produces a reference set where each account has at least three to five 3x-outlier posts inside the analysis window. If a candidate competitor produces zero outliers, they are publishing volume without signal and they are not actually a competitor for the format library work. Substitute another account from the same niche. Jenny Hoyos's vidIQ profile discipline applies, per Hoyos: "If the video is good, the comments will come. If the video isn't good, asking for them won't help." Translated: if the competitor's posts are not winning in the data, the synthesis from analyzing them will not win in your production calendar either.
Should I share the breakdown with my creative team or keep it analyst-only?
Share it. The creative team needs the synthesis document (the five to eight repeating archetypes) as the working brief for the next month calendar. The per-video one-pagers can stay in an archive folder; the synthesis is what the team builds against. The discipline is that the synthesis names the archetype and one production decision per archetype, not all 75 frame logs. The breakdown is upstream of the brief, and the brief is upstream of the calendar.
Start with your brand, product, profile, or video
Run your first director-level competitor breakdown, paste a supported URL in a planning-first tool
Generate a campaign brief