Search depth: video that works for years

Search depth: video that works for years

A video platform is built as a search engine rather than a feed: a clip is found by a query a year after publication. Hence different subjects, different titles and different economics — one piece of material works for months. Below is what follows for content and how to choose subjects.

A video platform is built as a search engine rather than a feed: a clip is found by a query a year after publication. Hence different subjects, different titles and different economics — one piece of material works for months. Below is what follows for content and how to choose subjects.

The clip got two hundred views and died

That is what almost everyone who tried video and gave up thinks. Recorded it, posted it, it picked up a little in a week, then silence. The conclusion: it did not take off, we should have done short formats.

The conclusion is wrong, and here is why. On social platforms material lives hours. Here it lives years.

A clip shot today can gain most of its views eight months later, when people start finding it by query. Judging it after a week is like judging a planted tree after a week.

That changes everything: what to film, how to measure and why a business needs it at all.

A video platform is a search engine

The root cause of most mistakes: people treat it as a social network.

On a social network material is shown to those who follow you or resemble them. Here the person enters a query himself — he has already formulated the problem and is looking for an answer.

Three consequences that invert the approach.

The subject is chosen by query rather than by interest. Not "what I would like to talk about" but "what people search for in words". Nobody searches for "My thoughts on the future of the profession". "How much does a landing page cost" gets searched every day.

The title matters more than the content. Not because content does not matter but because nobody reaches it if the title did not match the query.

The audience is colder than anywhere. The person does not know you and did not come to you — he came for an answer. Trust has to be earned in the first thirty seconds.

The misconception: "you need a lot of clips". You do not. You need a few that hit real queries. Twenty clips about nothing lose to three standing in the results.

What to film for YouTube search

Three types that work in search.

An answer to a specific question. The most reliable type. A person is looking for the solution to a problem and you give it in full. "What to answer a client asking about price", "How much does a landing page cost", "What to do if you are asked for a test task".

An analysis of a real case. A format almost nobody uses: from first contact to result, with the figures and with what went wrong. People watch to the end because they want to know how it turns out.

An analysis of a mistake. It works better than average and is almost unoccupied. It is also the fastest way to earn trust from a cold viewer.

What not to film: news and release round-ups. Aggregators produce those industrially and they date within a month — that is, they contradict the very nature of a platform where material should live for years.

How a clip that gets watched is built

The first thirty seconds decide everything. You cannot introduce yourself and talk about yourself here. Here you name the problem in the viewer's words and promise a specific result: "We will work through what to say when a client asks the price outright — I will show you a real exchange."

One question, one clip. The temptation to pack everything in at once kills the views: the person was looking for a specific answer and got a lecture.

Proof inside. A screen, a document, figures, a real example. Talking to camera without demonstration loses to an analysis with something shown.

And a platform rule people forget. Video platforms cut monetisation and reach for coarse language. A register that passes in a messenger costs money here. No swearing at all.

What to measure on YouTube

Not views in the first week. They mean nothing on a search platform.

Retention. Where people leave. A mass drop in the first minute means the title did not match the content. A drop in the middle means there is padding there.

Traffic source. Whether people came from search or from recommendations. Search is what all of this is for: it does not depend on algorithmic luck and does not run dry.

Click-throughs. How many people followed the link. That is the only metric connected to money.

The horizon for judging is six months. Earlier is not possible. A page reaches the results in two or three months, and that is a normal timescale.

Title and thumbnail

The two things most of the decision to watch rests on.

A title is written in the words of the query. Not the way it looks good but the way a person states the problem. "How much does a landing page cost" rather than "Pricing in web development". The test is simple: could you type this into a search box yourself?

The thumbnail repeats the promise rather than decorating. Three or four words large, readable on a small screen. Not the whole title and not an abstract picture.

Title and content must match. A mismatch gives the worst possible outcome: people click in and leave immediately, and the platform stops showing the clip at all.

And one expensive mistake: optimising the title for a clip that has already been shot. The right order is the reverse — the query first, then the script for it.

Why a business needs search depth

Three reasons, and none of them is fame.

Search traffic does not run out. A clip that reaches the results brings people in for years with no further investment. It is an asset rather than a publication.

Video explains what text cannot. Showing how something is done on screen is stronger than any description. For complex products, or ones people distrust, that is the decisive argument.

One clip feeds every other platform. A recorded video turns into a short clip, a carousel, a story series and a post. That is, this platform is not extra work but a source of raw material for the conveyor.

The last point changes the economics: filming becomes cheaper than not filming.

How many clips are needed and how often

The answer is not obvious: not many and regularly.

Five to seven clips that hit real queries work better than fifty random ones. A search platform does not reward volume — it rewards matching a query and being watched through.

Rhythm matters more than quantity. One clip a week, or even a fortnight, but consistently. A burst of ten videos in a month followed by silence gives a worse result than a steady pace.

And the horizon. For the first three months you are working blind: the page is not in the results yet, views are random and there is nothing to draw conclusions from. That is the most common moment for people to quit — and the worst one.

What the business gets

By the methodology I use, this platform is part of the social packaging stage — 4–5 days for all platforms. Part of full positioning: 3–4 weeks, 25–35 hours, around $2 000 for the package by the Freelance 2.0 course method. Honestly: that is the price of the service by our methodology, not market statistics.

Do it yourself or commission it

Doing it yourself. The mechanics above are the "Video" block of “Freelance 2.0”: choosing the channel concept, planning videos and script generators separately for expert and personal formats. All work from the knowledge base — the same one feeding the other platforms.

The course costs $9 and is included in Basic. Registration is free: three days of the Basic plan switch on with one button, no card needed. Basic includes 200+ AI assistants across 14 categories and the "Sales" course.

Commissioning it. The service exists. The sign of it being done wrongly: if a contractor promises a result within a month, he does not understand how a search platform works, and you will be paying for an expectation that will not be met.

Do one thing today: write down the five questions clients ask you most often, in the words they use. Those are ready titles for your first five clips, and they already match queries because they come from live speech.

What this looks like on a real project

Below is the same work on a live client: a dental studio run by two doctors, a husband and wife. The work was delivered in eleven sections, the seventh being video as search depth.

Why this example is here. It shows that positioning is not an abstraction but a service with a clear composition. And it sells to any business with an audience and a product: a clinic, a school, an agency, a shop. The mechanics are the same; the content changes.

An analysis of a deal like that from first message to payment is included in the “Sales” course (the “First order” series) on the Basic plan.

The platform's job

Here people do not scroll — they look for an answer and are willing to watch at length. Someone searches "why does my jaw click", finds a long analysis and goes into the bot.

The main principle: editing, not filming

One shoot, one day, all 29 subjects live: solo, as a pair, in the clinic.

It then unfolds into two formats: a long cut of 3–10 minutes with chapters, timecodes and a search title, and short clips of up to 60 seconds — the same hook clip leading to the long one.

The cost of adaptation = editing the cut + the short + title + thumbnail. There are no reshoots.

Four differences from other platforms

A search title is a query rather than a provocative hook. "Why does my jaw click in the mornings", not "you will not believe what happened".

Chapters and timecodes — the viewer jumps to the part he needs and retention rises.

Thumbnails with faces — the doctors' faces plus three or four words. For cases, the result is shown.

The description leads into the bot — hook, chapters, code word and a link to the same bot. No separate funnel is needed.

Five playlists from the same pillars

The content plan's pillars renamed as a shopfront for the viewer: "The cause everyone missed" — the flagship · "Genuinely not frightening" · "Two doctors behind the scenes" · "The unvarnished truth" · "Honestly about aesthetics".

The rhythm: around two long pieces a week plus two or three shorts. Of the 29 subjects, 21 go long, 3 go long with a teaser, 1 vertical only, 2 as written notes with no video.

Five prompts for video

Freelance 2.0, the "YouTube" block, Basic plan:

Video without filming. Cerebrum is a chat where the task is described in plain words: the right Gen AI models are picked for you, paid from the balance on actual use, the price visible before launch. The Full plan also opens the AI SKILLS connector for Claude: describe the task in Claude and it generates frames, video and voice and edits a clip with subtitles, graphics and sounds.

Assistants, Basic plan: Universal AI scriptwriter, Video editing expert.

Skills for video and a channel

The main point the first item confirms. Video works for years, but only if it is properly set up: title, description, tags, playlist. A clip without that will not be found however good it is.

What to pick up on the platform for this task

What to read next

How to find trends — how to look for subjects by query.

Editing expert videos — how to produce volume without a studio.

The content factory: a nine-week plan — where the content comes from.

How to turn a blog into a source of clients — the whole chain of eleven stages.

AI video generation in 2026 — the full guide to video models.

Also on AI SKILLS