Alex O.

Virality starts with knowing who the content is for, described through situations and behaviours rather than broad demographics.

• Three questions come first: who the real user is, what problem they already talk about, and what competitors are missing.
• The answers come from places users already write, including Reddit, TikTok search, and App Store reviews, and from studying top-performing videos in the category.
• When a video describes the problem the way users describe it themselves, engagement rises and the content converts.

Instead of writing for "students", Alex O. points to medical college students preparing their thesis; instead of "creators", short-form video editors trying to speed up daily workflows; instead of "founders", first-time SaaS founders trying to validate a product before fundraising. The repeated phrase is that a video should sound like the user's internal monologue.

Read original

No one can know in advance which hook, angle, or format will work, so the reliable path is to publish enough to get feedback, read clearer signals, and scale the format that starts outperforming the rest. Teams that try to find the winner before they start usually stall; teams that publish enough find it.

One team spends weeks debating hooks, angles, and formats before posting anything; another publishes 30 videos in the same period. Alex O. asks which learns faster: the publishing team.

Read original

Search the niche’s main keywords on TikTok, filter for the most-liked videos of recent months, and save those that clearly performed, with 500,000 views or more as a reasonable signal.

• Study them, rather than just collecting them: look at the hook in the first few seconds, where the creator switches scenes or adds a pattern interrupt, and how the pacing works.
• The same structures appear again and again, offering patterns to build on.

A saved video that cleared 500,000 views is not copied. Instead, it is annotated for its opening hook, scene changes, and pacing, then used as a structural reference for a new angle.

Read original

The test is blunt: remove the app from the sentence and see whether the hook still makes sense.

• If it does, cut it.
• The survivors become modular scripts, with the product load-bearing rather than tacked on at the end.
• The scripts go out as creator briefs to real people, generally in the 2,000 to 10,000 follower range with one or two past videos over 100,000 views, not to avatars.
• The same video is pushed across 50 or more brand-owned accounts, each with a different hook.
• A daily read decides what to kill, keep, or put paid spend behind.
• The rule: publish widely, but filter cheaply first.

A hook that still works after you delete the app is cut before anyone films. One that collapses without the app makes it into a modular script, then goes to a small creator roster and across dozens of brand-owned accounts with different openings.

Read original

A folder setup gives the assistant persistent memory of the brand, the creators, and every angle already tested.

• Six workflows run on a fixed weekly cadence: mining customer language, generating hooks, briefing creators, rotating angles, seeding comments, and identifying winning creative to put paid spend behind.
• A VOICE.md file is meant to keep generated scripts sounding like a real person who found something, not a brand announcing a product.
• The one-variable rule limits each brief cycle to a single change, and a Monday diagnostic names exactly one thing to fix.
• Thresholds trigger actions, such as starting comment seeding when a video crosses 10,000 views.

The post says the system flags a saturating angle, identifies a Spark Ad candidate, and triggers comment seeding at 10,000 views. A Monday diagnostic gives one fix, not a list.

Read original

This is the shortest statement in the set of the step Alex O. says teams skip.

The post argues that most teams jump straight to making videos without establishing “content–market fit.”

• Alex O. lays out three questions to answer first: who the real user is, described through situations and behaviours rather than demographics (not “students” but medical college students preparing a thesis; not “founders” but first-time SaaS founders validating before fundraising); what problem users already talk about, found by looking at Reddit, TikTok search and App Store reviews; and what competitors are missing, found by searching the category on TikTok for angles nobody covers or frustrations nobody addresses.
• The claim is that content converts when a video sounds like the user's internal monologue.
• The post ends with an offer to book a call about identifying content angles.

For brands planning a campaign

This is the intake step for an app campaign.

• If a client has an app but no defined user segment, run these three questions before briefing any creator.
• The answers become the language for hooks and scripts.

It is the cheapest entry point in this set and the method most directly reusable by anyone starting from zero.

Alex O. recommends building an “idea bank” instead of generating ideas from nothing.

• The method: search the main niche keywords on TikTok, filter for the most-liked videos of the last few months, and save videos that clearly worked — roughly 500,000 views or more as the signal.
• The point is not collecting but studying: the hook in the first few seconds, where the creator switches scenes or adds a pattern interrupt, and the pacing.
• They say the same structures recur, and the job is to take what already works and improve it.

For brands planning a campaign

Suggests briefing researchers with a shortlist of proven competitor structures rather than holding a blank-page ideation session. For an agency running several app clients, the bank is a per-client asset that can shorten every later production cycle.

If you read one thing to understand how Alex O. thinks about content marketing, this is it — and it takes a minute.

The post contrasts two teams: one debating hooks, angles and formats for weeks, the other publishing 30 videos in the same period.

• Alex O. frames content marketing as a discovery process: no one knows in advance which concept will work; volume produces feedback, feedback produces clearer signals, and eventually one format outperforms the rest and gets scaled.
• The stated failure mode is trying to find the winner before starting.

For brands planning a campaign

Sets the budget expectation for a UGC engagement: pay for a test-and-scale cycle with a publishing cadence, not a single set of approved creative. Useful when a client asks for a fixed deliverable count without a test window.

This is the closest thing here to a description of the work Alex O. would do for a client.

Alex O. describes turning a UGC process into a set of Claude “skills” that run end to end, in eight steps: mining App Store reviews, Reddit threads and competitor comment sections for phrases users repeat; writing 20 hooks per angle; scoring every hook before filming by deleting the app from the line and cutting anything that still makes sense; turning survivors into modular scripts where the product is load-bearing rather than tacked on at the end; sending each script as a creator brief to real people with 2,000–10,000 followers and one or two videos past 100,000 views; pushing the same video across 50+ brand-owned accounts, each with a different hook; running a daily automated read that flags anything past 5,000 views with 7%+ engagement for killing, keeping or paid promotion; and logging winning hooks, scripts and angles in a “Hook Vault” so the next batch starts from proven lines. They claim install costs as low as $0.08 and offer free access to the packaged system.

For brands planning a campaign

The most directly relevant piece for an agency campaign: it names the roster criteria, per-video thresholds and reuse mechanism that a paid system engagement would involve. The performance figures are the creator's own claims and are not verified in the posts; ask for supporting evidence before relying on them.

Read it after the UGC factory post to see what the day-to-day setup looks like.

The post describes running Claude Cowork as an app's “UGC content backend” rather than a script writer, with a folder setup that gives Claude persistent memory of the brand, creators and tested angles.

• Six automated workflows run on a fixed weekly cadence: customer language mining, hook generation, creator briefing, angle rotation, comment seeding and Spark Ad identification.
• A VOICE.md file is meant to keep generated scripts sounding like a person rather than a brand announcement; the “one-variable rule” holds that changing only one thing per brief cycle separates compounding campaigns from plateauing ones.
• The setup flags saturating angles, identifies Spark Ad candidates, and triggers comment seeding when a video crosses 10,000 views, with a Monday diagnostic naming exactly one thing to fix.
• Alex O. cites Tea App (6M installs), Turbo AI (500K users in five months) and WhatColors ($320K in organic revenue in four months), and gates access behind commenting “COWORK” and connecting on LinkedIn.

For brands planning a campaign

The one-variable rule and Monday single-fix diagnostic are process items an agency retainer could adopt without the tooling. Access requires commenting and connecting, so ask for a walkthrough of the setup.

The fix is to answer three questions first: who the real user is, described through situations and behaviours rather than demographics; what problem those users already talk about; and what the category's top-performing videos leave uncovered.

• Find answers in places users already write — Reddit, TikTok search and App Store reviews — and by studying the highest-performing videos in the app's category.
• When a video describes the problem in users' own words, engagement rises and the content converts.

Narrow the user to a specific situation, such as medical students finishing a thesis or first-time SaaS founders validating before fundraising. Then capture the exact phrasing users repeat and target angles competitors have not covered.

For brands planning a campaign

Works as a short pre-brief for app clients with a product but no defined audience, giving the agency an evidence-based way to choose angles before spending on production.

Examples: Source 1

Search the niche's main keywords on TikTok, filter for the most-liked videos of recent months, and keep those that clearly performed, using 500,000 views or more as a reasonable signal.

• Study them rather than just collecting them: note the hook in the first seconds, where the creator switches scenes or inserts a pattern interrupt, and the pacing.
• The same structures recur, providing a starting point to improve on rather than copy.

Filter by keyword and recency, shortlist by view count, then annotate each saved video for its hook, interruptions and pacing.

For brands planning a campaign

A low-cost first research pass that produces angles and format references grounded in what the category already rewards, useful before commissioning new creative.

Examples: Source 1

Content marketing is a discovery process — nobody can know in advance which concept will work — so the reliable way to find a winner is to publish enough to get feedback, read the signals and scale the format that starts outperforming the rest. Teams that try to find the winner before they start usually stall; teams that publish enough content discover it.

For brands planning a campaign

Supports a test-and-scale budget model for app clients: produce more low-cost assets and get faster reads, rather than relying on a few heavily planned videos to carry all the risk.

Examples: Source 1

Score every hook before filming and cut the weak ones.

• The screening test is blunt: remove the app or product from the sentence and see whether the hook still makes sense.
• If it does, cut it.
• The survivors become modular scripts in which the product is load-bearing, not tacked on at the end.

Generate 20 hooks per angle, apply the delete-the-product test and keep only hooks that stop making sense without the app.

For brands planning a campaign

Gives the agency a fast, repeatable in-house screen to keep organic and paid production focused on concepts that depend on the client's app.

Examples: Source 1

The same video goes out across 50 or more brand-owned accounts, each with a different hook.

• Each morning, the system reads the numbers and says what to kill, keep or put paid spend behind, flagging anything past 5,000 views with 7% or higher engagement.
• Winning hooks, scripts and angles go into a 'Hook Vault' so the next batch starts with proven lines.
• Alex O. says this can produce install costs as low as $0.08, versus paying to produce a handful of untested ads.

Fixed view and engagement thresholds for paid decisions, plus a running library of proven hooks, scripts and angles.

The follower ranges, performance thresholds and $0.08 install cost are Alex O.'s own description. The app results cited elsewhere — 6M installs for one app, 500,000 users in five months for another, and $320,000 in organic revenue for a third — are self-reported. Ask for supporting data before using them to set client expectations.

For brands planning a campaign

Maps directly onto an agency's need for a system it could buy or replicate for app clients: creator rostering, distribution across owned accounts, daily optimisation and a shared record of what has worked.

Examples: Source 1

The system has six automated workflows on a fixed weekly cadence: mining customer language, generating hooks, briefing creators, rotating angles, seeding comments and identifying winning creative for paid spend.

• A voice file keeps generated scripts sounding like a person who found something rather than a brand announcing a product, while the one-variable rule limits each brief cycle to a single change.
• Thresholds trigger specific actions — for example, comment seeding starts when a video crosses 10,000 views — and a Monday diagnostic names exactly one thing to fix.

One-variable rule per brief cycle, a fixed weekly cadence of six workflows, and one diagnostic item each week rather than a list of fixes.

The playbook is distributed through a gated offer, asking readers to comment a keyword and connect to receive it, and the three app results Alex O. attributes to this setup are self-reported rather than verified here.

For brands planning a campaign

Relevant to an agency seeking an internal production system: this operating model defines triggers and separates research, briefing and optimisation. Test the tooling and thresholds on the agency's own clients before promising results.

Examples: Source 1

Ask what Alex’s relationship to the app is: client work, employment, or another arrangement.

View example

Ask how Alex relates to the business.

View example

Ask what Alex’s relationship to the app is.

View example

The post presents the product as theirs but names no company behind it or publication date.

View example

The posts describe use of the product, not a partnership or sponsorship with Anthropic.

View example

While Alex O. filters hooks before filming, Soriano starts after an ad has run and breaks performance down by stage: a weak hook rate means the opening three seconds failed; a weak hold rate means the middle drags; a good hold rate with low clicks points to the offer; and clicks without conversions point to the landing page.

• His re-edit list turns one existing UGC clip into five testable ads by swapping the opener, leading with the problem, product or proof, and tightening the cut.
• Both treat volume as something earned through selection, not simply by shooting more.

Soriano focuses on paid-ad diagnostics and recutting footage a brand already owns, mainly for DTC advertisers on Meta and TikTok.

• Alex O. spends more time before production, defining audiences and the language they use to describe problems.
• Their pipeline targets app installs rather than retail purchases.
• Soriano assesses performance from an advertiser’s media-buying perspective; Alex O. focuses on content and creator briefing.

Read original · Original 2

He argues that TikTok Shop starts the buying journey before any search, with creators showing products in real situations and comments and follow-up videos addressing objections.

• He identifies three types of content that convert: problem-solution, product demonstration and creator review.
• Problem-solution content captures attention, demonstrations build understanding, and creator trust drives the purchase.
• That closely matches Alex O.’s emphasis on using the audience’s own language for the problem and working with real creators rather than avatars.

Akbar focuses on TikTok Shop as a commerce channel and on how shoppers move from no intent to checkout, speaking to ecommerce and DTC brands selling physical products.

• Alex O. covers app installs and the production pipeline behind creator content, including the hook filter, multi-account distribution and the tools that run it.
• Akbar explains what kind of content sells; Alex O. explains how to produce and screen it at volume.

Read original · Original 2

He argues that agencies hire for portfolio polish when conversion-focused creative requires different roles.

• He describes four: someone who mines customer reviews and support tickets for exact phrasing; someone who turns that language into testable angle hypotheses; someone who executes formats quickly; and someone who tracks hook stop rates and angle saturation.
• His contrast between a research lab and a content factory, and his question about whether a team is testing beliefs or just making new creatives, echoes Alex O.’s distinction between publishing enough to learn and buying untested ads.

Taylor writes about structuring and hiring a creative team.

• His detailed case study covers a paid-social account spending large sums on Meta for a DTC brand.
• Alex O. focuses on the content itself, from audience definition through creative direction and daily optimisation, in a mobile-app context.
• Taylor’s performance numbers describe agency output and hit rate; Alex O.’s claims concern a specific UGC pipeline and its reported install costs.

Read original · Original 2

Soriano's re-edit list turns one raw UGC clip into five testable ads without a reshoot: swap the opener and bring the strongest line from the middle to the front; lead with the problem, the product or a proof point; then adjust the pace and rewrite the on-screen captions so the first line viewers read matches the new hook.

• Soriano also offers a straightforward diagnosis for an underperforming ad: read hook rate, hold rate, click-through and conversion, then fix only the layer where viewers drop off.
• For an agency with a short test window and clients who have already filmed, that is cheaper and faster than briefing a new creator roster.

The reviewed Soriano posts focus on DTC apparel and short-form paid social, not app install campaigns.

• They do not cover app store review mining, content-market fit for an app or install-cost targets, which are central to the Alex O. system.
• Soriano's method multiplies tests using footage the client has already paid for; it does not build the standing pipeline of angles, creator briefs and multi-account distribution an agency needs across several app clients month after month.
• Choose Soriano for a one-off re-cut sprint on existing assets; choose Alex O. when the client needs a repeatable production system.

Examples: Source 1 · Source 2

Akbar treats TikTok Shop as demand creation, not demand capture: a creator shows a product in a real situation, comments answer doubts, another video builds trust and a live session closes the sale.

• The buying journey can start with someone who had no intent five minutes earlier.
• The three content types Akbar names—problem-solution, product demonstration and creator review—give buyers a simple way to sort content by what it does for the viewer.
• For an agency whose client sells a physical product and has live selling and product reviews available, Akbar's framing maps directly onto the assets on hand.

The Akbar posts contain no app-install data, hook scoring, pre-filming tests or daily paid-scaling decision rules.

• Akbar's described mechanics depend on product listings, reviews, comments and live selling, none of which exist for a mobile app.
• Akbar explains why social commerce content works and how to categorise it; they do not describe a production chain from angle mining through creator briefs to explicit kill, keep and scale thresholds.
• Use Akbar when the deliverable is physical product sales through TikTok Shop; use them when the deliverable is app installs and in-app outcomes.

Examples: Source 1 · Source 2

Taylor argues that agencies hire for portfolio quality and end up with beautiful ads that do not convert.

• They rename the roles accordingly: a customer language translator mines reviews, Reddit and support tickets for the exact words buyers use; an angle strategist turns those words into belief-shift hypotheses; a format executor is hired for speed and versatility; and a performance analyst tracks hook stop rates, creative lifespan and angle saturation.
• Taylor also walks through diagnosing a high-spend DTC account where one angle had been iterated to death, no record of past tests existed and the founder had been the creative strategist, so the system left when he stepped back.
• That diagnosis is useful when the agency's real problem is internal process and institutional memory.

Taylor's posts describe agency team structure and a DTC supplement account, not app campaigns, and end with a call booking rather than a delivered system.

• Taylor's reported numbers—30 to 40 percent lower volume with two to three times the hit rate—are self-reported and should be checked against the agency's own accounts.
• Taylor explains how to build and staff a team that compounds learning; they do not provide the hooks, briefs, distribution accounts and daily thresholds that Alex O. describes as a running system.
• Choose Taylor when the agency is restructuring its own function; choose them when it wants a defined UGC production system it can run across app clients.

Examples: Source 1 · Source 2