By Lesli Rose · April 30, 2026 · 7 min read
This operator has what a lot of creators chase for a decade: 22,000+ Substack subscribers, weekly Friday cadence since 2023, a 70K LinkedIn following, and four operating businesses underneath the newsletter -- a recovery resource hub he founded over 15 years ago, a marketing agency he runs today, a partnership in a household-name copywriting brand, and an operator community. The writing is good. The audience is engaged. The businesses are real.
When I looked under the hood, the newsletter URL was reading to AI as "another Substack publication." Not the operator. Not the empire. Not the agency, not the recovery brand, not the copywriting partnership. Just one of millions of Substack publications, with default-platform metadata and zero structured data connecting the URL to the four businesses underneath it.
The audit recommendation is unusual for me: do not move off Substack. Build a thin owned-brand domain alongside it. Substack stays exactly where it is. The newsletter never blinks. The cross-link strategy makes the entity graph legible to AI without disrupting the working distribution surface.
The publication has zero application/ld+json blocks in the raw HTML. Zero microdata. No Person schema for the operator. No Organization schema for the umbrella business. No FAQPage. No NewsMediaOrganization.
This is not laziness. Substack does not allow root-level JSON-LD injection at the publication level. Per-post Article-flavored OG tags exist as a floor, but the entity layer that should tie the operator to four operating businesses simply has nowhere to live on Substack itself.
For a creator-only operator, this is a survivable gap. For an operator with four businesses underneath the newsletter, this is the largest single missing piece of infrastructure. AI cannot connect a Substack publication URL to a recovery non-profit, a marketing agency, a copywriting brand, and an operator community without an entity hub that names all of them in structured form.
The publication’s title tag is "[Operator Name] | Substack." The meta description is the auto-generated "Click to read [Operator Name], a Substack publication with tens of thousands of subscribers." Same string is reused as og:title and og:description.
That is platform talking, not operator. It is also the same string serving search snippets and social previews -- two different jobs that should be done by two different strings. Meta title is for search engines; OG title is for humans scrolling LinkedIn. They should never be identical.
I searched four queries an audience-relevant person would type in 2026: "best business newsletter on Substack," "best newsletter for entrepreneurs 2026," "best operator newsletters," and direct comparison to a name-brand competitor.
Zero of four returned this operator. The category was dominated by the same five names every roundup features: Lenny Rachitsky, Justin Welsh, Codie Sanchez, Sahil Bloom, Morning Brew. Each of those creators owns a domain, ships full schema, and has been featured in tier-1 publications. The audited operator has 22K subscribers but is absent from every "best of" listicle on page 1-2.
The brand-name query ("[founder] of [old company]") returned cleanly with positive sentiment. People who already know the operator find them. People who should find them have no path.
22,000 subscribers. Zero money pages. The operator runs a paid community and a marketing agency, both invisible from this URL. Every Friday essay sends people back to a homepage that asks them to subscribe -- but they are already subscribed. There is no second action for an existing reader who would happily pay $300/year for the community or refer the agency to a marketing director friend.
This is the most expensive gap in the funnel. It is also the gap a Substack-only architecture cannot fix on its own, because Substack is a publishing engine, not a commercial site builder.
LinkedIn says the marketing agency. Substack says "Substack publication." The copywriting brand lists the operator as an author. The recovery brand lists them as the owner. None of the four surfaces names all four businesses. AI builds an incoherent entity from this.
The fix is one canonical positioning sentence, pasted identically into every bio. Same name, same descriptor, same four businesses, same link target. Inconsistency is not an SEO offense -- it is an entity-recognition offense. AI cannot return one coherent person if the platforms it scrapes describe four different ones.
Build a 12-page Next.js site at the operator’s personal domain (or a media-brand umbrella). Static MDX, $5-20/month to host. The site contains:
Every Substack essay links once to its pillar on the owned-brand domain. The pillar pages link forward to the essays. The newsletter never moves. The voice never changes. The Substack discovery surface keeps doing what it does. The owned-brand domain just gives the four-business empire a home AI can map to.
This is the Galloway model. One entity hub. Multiple distribution surfaces.
The newsletter is good. Don’t change it. The cadence is real. The voice is operator-specific in a way that competitors who teach about being an operator (without being one) cannot replicate. The topic mix -- AI personal branding, healthcare investing, lead generation, recovery -- is differentiated and fresh. The 22K-subscriber list is a working asset.
None of this audit is about the writing. It is about giving the writing a permanent home that compounds.
If you are an operator running multiple businesses with a Substack-first publishing strategy, the entity-graph gap probably looks identical to this one. Substack is excellent for distribution and weak for entity infrastructure. The fix is structural, not creative -- and the recommendation is layer, not migrate.
Substack publications can be cited, but their structural ceiling is lower than a domain you control. For a creator who is also one Substack among millions, default-allow on AI crawlers is workable. For an operator with multiple businesses underneath the newsletter, the entity graph that should connect them all has nowhere to live on Substack itself.
No. The recommendation is alongside, not instead of. Substack handles distribution, email, and discovery. The owned-brand domain handles the entity layer (schema, sub-organizations, llms.txt, money pages). The two cross-link. The newsletter never moves.
AI builds entity graphs from structured data + cross-platform consistency + third-party citations. Operator-creators usually have one signal-rich URL per business but no unifying entity hub. AI fragments the person across surfaces. A single owned-brand domain with Person + Organization schema collapses the fragments into one entity AI can recommend with confidence.
Scott Galloway has profgalloway.com as a unified entity hub that ties together Section, No Mercy/No Malice on Substack, Pivot, Prof G Pod, NYU Stern, his books, and his board seats. AI returns him as a single coherent entity. Most creator-operators have the multiple distribution surfaces but not the unifying hub. Building the hub is what makes AI eligible to return them as an entity rather than as a publication URL.