← mikelitman.me Fractional Head of AI
Fractional Head of AI

Most brands don't have a reach problem. They have a publishing problem.

How I locate where a brand sits, and the 90-day arc that moves it up one rung. Strategically, creatively, hands-on, and fast.


A brand either broadcasts, or it publishes. My job is to move it up the ladder.

What follows is the method, in public. The named work it produced is a teardown of one real brand, kept private and shared one to one.

§ 01The ladder

Every brand sits on one of three rungs. Most never name which. The whole engagement is locating the rung, then moving the brand up exactly one.

01 / BROADCAST
Pushes out
Announcements, on the brand's schedule. Measured in volume and views. The audience receives; it does not return.
02 / PUBLISH
Is read
Recurring formats with a voice. Measured in whether people come back. The audience returns to a cadence, not to an announcement.
03 / OWN
Holds a position
A thesis the category associates with one name. Measured in share of voice on an idea, not volume of posts.

The ladder is the portable part. It does not care what the product is, how big the following is, or which platform is in fashion. It only asks: when you stop posting, does anyone notice? Broadcasters disappear the moment the calendar goes quiet. Publishers are missed.

Reach is rented. A position is owned.

§ 02What publishing sounds like

I would rather show the standard than describe it. So here is a publication I built and run on the middle rung, every single morning: a daily culture brief distilled from 150-plus sources, published at 8:30 with an audio edition, 133 editions in and counting.

Showcase · daily brief + audio edition
The Pattern: before it’s obvious
Daily culture intelligence across fashion, design, technology, brands, music, and art. A recurring format with a voice, on a schedule an audience can return to. The method and the product are the same artefact.
Read this morning’s edition →

This is the bar I hold client work to. Not a deck about what AI could do, but the thing itself, shipped, in a register people actually read. Most people can write about AI. Fewer have built with it. That overlap is the entire job.

A publication that names the shift before it is obvious is not a pitch. It is the proof.

§ 03The diagnosis: See, Cut, Grow

The audit sorts everything a brand publishes by attention mechanics, not vanity metrics. Three calls fall out of it.

01 / SEE
What works, and why
Name the structural cause of a win, so it repeats on purpose instead of recurring by luck. A win you cannot explain is a win you cannot keep.
02 / CUT
What does not
A subtraction discipline. Most brands have a volume problem wearing the costume of a reach problem. The fix is fewer, better, not more.
03 / GROW
The arbitrage
The assets earning far below the distribution they could command. The best content a brand owns is usually reaching its smallest audience.

Vanity metrics flatter broadcasters. They count what was pushed out, never whether anyone stayed. Attention mechanics ask the harder question: which structures earn return visits, and which just fill the feed? The audit answers it asset by asset.

Views are a vanity metric. Attention is a structural one.

§ 04The transformation: Kill, Fill, Build

The plan is a deliberately sequenced 90-day arc. The order is the point, and the first phase is the one almost everyone skips.

PHASE 01 / KILL
Subtract first
You cannot install a cadence on a clogged calendar. Mostly removal, plus hard constraints. Stop the work that is busy, not heard.
PHASE 02 / FILL
Install cadence
Replace one-off broadcasts with recurring, ownable formats. The deliverable is predictability. People return to a schedule.
PHASE 03 / BUILD
Claim a thesis
Graduate from publishing to owning. Stop adding posts and start compounding a position the category files under one name.

Generalised, the arc is subtract, then systematise, then own. Skip the subtraction and the new cadence lands on top of the old noise and suffocates. Skip the cadence and the thesis has nothing to stand on. The sequence is not a preference; it is load-bearing.

Audiences return to a schedule, not to announcements.

§ 05The matched pair

The audit and the plan are not two documents. They are one engine. Every finding in the diagnosis maps to exactly one action in the transformation.

Diagnose (the audit)Transform (the plan)
SEE what worksBUILD it into the owned thesis
CUT what does notKILL it, and protect the gap it leaves
GROW what is underusedFILL it into the cadence

This is what turns a one-off teardown into a repeatable method. The diagnosis is never abstract; it is already the plan, written in a different tense.

None of this is magic, and the labels are not the point. See, Cut, Grow and Kill, Fill, Build are scaffolding for a single discipline: decide what actually earns return attention, then build only that, in the right order. The value is the sequencing and the evidence behind each call, not the acronyms.

The audit and the plan are the same document, read twice.

A brand that publishes on a schedule beats a brand that announces on a whim.
The whole thesis, in one line

§ 06Three laws underneath it

Three laws sit underneath all of it. They are why the ladder works.

  1. Evidence beats claims. Demos beat descriptions. Show the thing working and the audience stays; describe it and they leave. When the product can do the talking, let it.
  2. Participation beats announcement. Programs beat products. An audience given a way in shows up again. An audience handed a press release does not.
  3. Human beats logo. A founder or a personal voice tends to outperform the brand account on identical content, often by a wide margin. People follow people. They tolerate logos.

People follow people. They tolerate logos.

§ 07The build record

The method above is half the job. The other half is that I build. Since 15 January, working solo with AI agents doing the heavy lifting, I have shipped 27 products, and the meter that proves it runs in public.

59.6B
tokens since 15 January
27
products shipped solo
1,438
novels’ worth of output
119×
return on the tool spend: $142,622 of API-equivalent work on $1,200 of subscription

None of these numbers are taken on trust. They render daily from a public dashboard with its own provenance strip, and the whole six-month run is written up as a long read: what compounded, what broke, and the rules that came out of it.

And the range is not all meters and dashboards. Labs is the playful end of the same method: six small local-first tools, from a poker-night ledger to sneaker cost-per-wear maths, each a single HTML file, shipped in one afternoon.

Anyone can advise on AI. The meter shows the building, daily, in public.