Building a media kit for a faceless AI channel

A standard creator media kit opens with a face and a story, because both are doing a job: the face says a real person stands behind this, and the story says they will be reasonable to work with. A faceless channel has neither, and every template you find online assumes you do. The result is that faceless operators either send a kit with an awkward hole where the photo goes, or send nothing and pitch in a cold email with three view counts in it.

Neither works well. The fix is not to fake a personality onto the kit. It is to replace what the headshot was doing with evidence a faceless channel can produce better than a personality channel can — an audience you can describe precisely, and a retention curve that shows people actually watch.

What the headshot was actually doing

Strip the sentiment out and a creator photo answers three sponsor questions at once: is there a real person here, will their vibe fit our brand, and will this be a professional engagement. A faceless channel has to answer the same three questions with different evidence.

Sponsor question	Personality channel answers with	Faceless channel answers with
Is this real?	A face and a name	Audience composition pulled from native analytics
Will it fit our brand?	Their on-camera persona	A documented, repeatable format spec
Will this be professional?	Prior brand collabs, tone of the email	Turnaround, revision policy, disclosure policy in writing

Note that the second row is where faceless channels are genuinely stronger. A brand betting on a personality is betting on a person who might post something off-brand next Tuesday. A brand betting on a format is betting on a template — one they can see, and one that will look the same in six weeks. Sell the format as the asset, because it is.

The general five-section kit — cover, showcase, audience, ways to work together, past work — is still the right skeleton, and building a media kit with AI covers that build. What follows is what changes when there is nobody to put on the cover.

The one-pager, section by section

Keep it to one page. A faceless kit that runs five pages is padding, and padding is what a sponsor reads as inexperience.

Header. Channel name, niche in six words, platforms with follower counts, contact. No tagline, no mission statement.
The format spec. The single most important block, and the one almost nobody includes. Three lines describing the actual format: length, structure, voice, cadence. "9-minute documentary-style breakdowns of aviation incidents. Scripted narration, archival-style visuals, published Tuesdays." A sponsor reading that knows exactly what they are buying.
Where the integration goes. Name your slots and their timestamps. "Pre-roll at 0:00–0:20, mid-roll at approximately 40% runtime, end card." Include the retention figure at each slot position. This converts a vague "sponsorship" into an inventory item with a known audience size.
Audience composition. Covered in detail below. Screenshot, not typed numbers.
Retention curve. Also below. Also a screenshot.
Three stills and one link. Frames from your best-performing pieces with one line of context each, plus a link to a full episode. Not a highlight reel — sponsors want to see the normal episode, since that is what they are buying.
Terms in one line each. Turnaround, revision rounds, usage window, exclusivity stance. Pricing goes in the conversation, not the kit; how to set it is its own discipline, worked through in pricing brand deals as an AI-assisted creator.
Disclosure policy. One sentence stating that AI production is disclosed per platform and that sponsored content carries the required material-connection disclosure. This pre-empts a question that would otherwise arrive on the call.

Metric one: audience composition

Follower count tells a sponsor how big the room is. Composition tells them whether their buyer is in it, and that is the number that actually moves a deal — a 12,000-subscriber channel whose audience is 70% one country and one age band is easier to sell than a 400,000-subscriber channel scattered across forty markets.

Include, as a screenshot of the native analytics panel rather than retyped figures:

Geography, top three markets with percentages. If 80% is one country, say so — a sponsor who ships to two countries needs to know that before they pay for everywhere else.
Age and gender split, whatever the platform reports.
Returning versus new viewers. A high returning share is the closest thing a faceless channel has to a parasocial argument, and it is a stronger one because it is measured rather than claimed.
Subscribed versus unsubscribed views, if your platform breaks it out. It tells a sponsor whether they are buying your audience or the algorithm's.

Date the snapshot and use a consistent 90-day window. A screenshot with a visible date range reads as verifiable; a table of typed percentages reads as a claim, and sponsors have been burned enough to know the difference. Pulling the same window across several platforms is what checking your social performance is for, and what belongs in the reporting habit is covered in creative analytics.

Metric two: the retention curve

The retention curve is the metric that does the most work in a faceless kit, because it answers the objection a sponsor has not said out loud: if there is no personality, why does anyone stay?

A curve that flattens after the first thirty seconds and holds is proof of format quality no copy can substitute for. It also makes your mid-roll slot priceable — a sponsor sees the percentage of viewers still present at the timestamp their spot occupies, which is a far better basis for a rate than a view count.

What to include and how to read it:

The full curve for a typical episode, not your best one. Sponsors discount best-case screenshots automatically.
The value at each integration point, called out. "62% still watching at the mid-roll" is the sentence that prices the slot.
Average view duration and completion rate as raw numbers underneath the curve.
The first-30-seconds shape, which is where format problems show. If yours drops hard there, fix the format before you send the kit; the diagnosis process is in audience retention analysis for creators.

Two platform notes before you build a kit around watch-based numbers. On YouTube, monetisation runs on 1,000 subscribers plus 4,000 qualified watch hours in 12 months, or 1,000 subscribers plus 10 million qualified Shorts views in 90 days — and Shorts-feed watch hours do not count toward the 4,000-hour path, so a Shorts-heavy channel can look larger than its long-form watch-time evidence supports. On Meta's Content Monetization Program, shares and comments are weighted above likes, and the view-counting thresholds are reported rather than published as one official table — around three seconds on a Reel and around thirty seconds for long-form, which is the tier that carries ad breaks. If a sponsor asks, quote the surface as well as the number.

One thing to leave out entirely: your RPM. The circulating per-niche ranges — roughly $12–$22 per thousand views at the finance end, roughly $5–$9 at the motivation and story end — are self-reported by creators, with no audited public dataset behind them. They are useful for setting your own floor. They are not evidence about your audience, and putting your ad revenue in a sponsor's hands invites them to price your slot against it.

Building it, and keeping it current

The production side is small. Generate a cover graphic and section headers in your channel's visual style, pull three stills from published episodes, and screenshot the two analytics panels. If the channel runs through a faceless video generator workflow, the visual language is already consistent — the part that usually takes designers longest.

The harder discipline is currency. A kit with six-month-old numbers is worse than no kit, because a sponsor who checks your recent uploads against a stale snapshot now has reason to doubt everything else on the page. Rebuild the two screenshots monthly on a fixed date, keep the format spec and terms static, and version the file so you know which one you sent to whom. The operating model behind all of it is covered in building a faceless YouTube business.

FAQ

Should I invent a persona name and give it a face?

Only if the channel already has one and it is consistent. A narrator persona with a fixed name and voice is legitimate branding and can occupy the cover slot. Inventing a photorealistic "founder" who does not exist to fill the headshot hole is a different thing: it is a claim about a real person, it will not survive a sponsor's basic diligence, and the moment it is discovered it becomes the only fact about your channel anyone remembers.

What do I put in the "past brand work" section with no prior deals?

A spec integration. Build a 30-second segment in your own format for a product in the sponsor's category — not their brand, their category — and link it. It costs one episode's production effort, it demonstrates the format and the integration in the same artefact, and it answers "what would this look like for us" better than a list of past logos would. It also puts you in a different conversation from every creator sending the same generic kit.

Do sponsors care that the content is AI-produced?

They care about specific consequences of it: brand safety, disclosure obligations, and whether the audience is real. Naming your production method plainly in the kit, alongside your disclosure policy, defuses the vaguer version of the concern before it forms. Vagueness is what makes a brand's legal review slow; a written policy is what makes it fast.

One page really is enough?

For the first contact, yes. The point of the kit is to earn a call, and one page of specific, verifiable evidence does that better than five pages of context nobody asked for. Keep a longer deck for the second conversation, where questions about usage rights, exclusivity windows and multi-episode packages actually come up.