Why AI recommends who it recommends
When someone asks ChatGPT or Perplexity "who is the best people management coach" or "how do I delegate effectively as a business owner," the engine does not roll dice. It runs a live web search, reads the pages it trusts most, and assembles an answer from named, well-evidenced sources. The names that come out are not the best coaches in the world. They are the coaches whose evidence the model can find, parse, and trust.
Mads's audit makes the distinction sharp. When we asked the AI by name, it described him accurately: "a management coach and consultant known for expertise in effective management, productivity and business leadership." So the model already knows who he is. But across 11 category queries that no buyer would phrase with his name in them, he was cited 0 times. The AI knows him but never recommends him. That is a surfacing gap, not a knowledge gap, and it is the most fixable kind there is.
Four levers decide who gets surfaced. Every recommendation an engine makes is some combination of these.
Entity strength
The model has to be confident it knows who you are before it will put your name in an answer. That confidence comes from explicit, machine-readable identity: Person and Organization schema, a sameAs trail tying your site to your socials and podcast, and one consistent positioning everywhere. A site with no homepage schema and a 500-erroring sitemap reads, to a model, as an entity it cannot pin down. It hedges by naming someone it can.
Third-party validation on the platforms AI trusts
Models weight what other sources say about you far more than what your own site says. In Mads's audit, the source domains feeding the category answers were dominated by YouTube, which appeared roughly 7 times, alongside UK coaching and training sites. madssingers.com appeared in zero category source lists. The engine builds its answer from where the conversation actually lives, and right now that is not on his domain.
Answer-shaped content
Models quote the page that most directly answers the exact question. A definitive "how to delegate as a business owner" guide, a clean "best frameworks for people management" comparison, a straight "what is DISC and how do you use it" explainer. Mads ranks for 1,833 US keywords, but every one of the top 200 points to informational /management/* theory, not to a page shaped like an answer to a buying question. He has the traffic and none of the answer pages.
Being the named authority for a lane
Models love a clear owner of a narrow topic. "Marshall Goldsmith for executive coaching." "Simon Sinek for leadership." When a query is specific, the engine reaches for the specific name. The category answers Mads loses go to a roster of authors (Goldsmith, Sinek, Brene Brown, Liz Wiseman, John C. Maxwell) and generic DISC vendors (Thomas International, Everything DISC, Discflow). Nobody owns "DISC coach for entrepreneurs." That lane is open.
Step by stepThe 6-step playbook
This is the exact sequence OO runs to move a business from described to recommended. It is ordered. Each step makes the next one work harder, and the early steps are the cheap, high-leverage ones.
Entity foundation
Give the model an identity it cannot mistake. Add Person and Organization schema site-wide, not just on /about where it sits today. Wire a full sameAs trail from the homepage to his socials, his podcast, and his media features so the engines stitch one entity together. And fix the sitemap.xml HTTP 500 first, because a broken sitemap actively hampers crawl and AI discovery. This is the single highest-leverage technical fix on the site.
Open now Homepage has zero schema. Consulting and coaching pages have none either.
Evidence pages
For every money query he is absent for, build one definitive page that exists to answer it. "Management coach for entrepreneurs." "Executive coaching for delegation." "DISC training for business owners." These are the commercial pages the site does not have today. Right now 0 of the top 200 US keywords point to a commercial page, so even his strong traffic never reaches a page a buyer or a model would cite. Each evidence page carries its own schema, real testimonials, and a clear answer in the first paragraph.
Answer-shaped content
Reshape and add content in the formats AI quotes: "best frameworks for X" lists, head-to-head comparisons, and definitive how-to guides ("how to delegate without losing control"). These formats win because they map one-to-one onto how people phrase questions to an engine. Mads's existing /management/* theory library is a real asset here. It becomes the supporting evidence behind the answer pages rather than the destination itself.
Platform presence
Show up where the engines already look. The audit found YouTube was the single most-cited source domain (~7 appearances) feeding the answers, plus DISC and coaching directories. So the plays are concrete: build out the YouTube presence around the exact category questions, and get listed and active on the DISC and coaching directories AI is pulling from. The goal is to be present on the pages the model already reads, not to hope it discovers a new one.
Third-party citations
Earn mentions on sites the model already trusts: guest posts, podcast appearances, and directory listings under the lanes he wants to own. This compounds Lever 2 directly. It also doubles as a recovery win, since his backlink profile has 505 broken links out of 1,208 (~42% dead) across 285 referring domains. Reclaiming dead links and adding fresh, relevant ones rebuilds the authority signal the engines read.
Measure
Re-run the exact AI probe monthly: the same 11 category queries across ChatGPT and Perplexity, web-search-backed, scored on whether Mads is cited and which sources won. The metric that matters is surfacing lift: moving from 0 of 11 category citations toward owning the queries one at a time. Because the model already knows him, the first wins tend to come fast once the entity and evidence layers land.
Why the order matters
Steps 1 and 2 are cheap and load-bearing. A confident entity plus a real answer page is what makes every later signal payload land on something the model can cite. Doing platforms and citations (4 and 5) before the entity and evidence layers exist is pushing traffic and trust at pages that cannot convert a model's attention into a recommendation. Foundation first, then evidence, then volume, then proof, then measurement.
The openingMads's unfair advantage
The DISC lane is wide open. When the AI answers DISC questions, it cites only generic vendors: Thomas International, Everything DISC, Discflow, DISCGB. It names tools, not people. There is no recommended DISC coach in any of those answers. That is a named-authority vacancy (Lever 4) sitting in the open.
Mads is built to fill it. He already has the IP, a roster of around 12 testimonials, and the kind of authority that AI weights heavily: he has coached people at Shell and Coca-Cola and worked with hundreds of entrepreneurs. The pieces the engines reward are already in his story. They are just not yet wrapped in the machine-readable form that lets a model reach for his name.
He is one entity pass plus a content pass away from being the answer to "who is the best DISC coach for business owners." The category is uncontested, the underlying authority is real, and the gap is the most fixable kind: described, not recommended.
The IP
Effective People Management, delegation, and a DISC-led methodology already taught and packaged. The substance the answer pages need exists.
The proof
Shell and Coca-Cola in the credential line, hundreds of entrepreneurs coached, around 12 testimonials. Trust signals models read directly.
The lane
No named DISC coach in any AI answer. The owner slot for "DISC coach for entrepreneurs" is unclaimed and waiting.
This is the playbook OO runs, the same one behind every page in this orb. Mads is welcome to it whether or not OO ever touches the site. The mechanism is the gift here. If any step is useful to hand to whoever works on the site next, take it and run. No pressure, no catch.