What I'd change next, from the book.

I read the full SOURC-E manuscript, all twelve chapters and every case study, then held it up against your current deck and talk-track. The webinar teaches the framework accurately. The book is sharper, and it reframes SOURC-E in a way worth stealing.

The book's actual thesis

AI visibility = Offsite × Uniqueness × Relevance × Credibility Structure , measured by Evaluation

The pillars multiply, they don't add. Structure is the denominator: great content divided by broken structure lands near zero, not half a win. Evaluation sits outside the formula as the instrument you read it with. Your slide 13 shows six identical boxes, which quietly says the six are equal and independent. They aren't, and fixing that is the single biggest upgrade on this page.

Make the formula the framework

Replace the flat six-box overview with the formula plus one metaphor, and echo it in the close. The book gives you two that land.

The band

Singer, guitarist, drummer, songwriter are Offsite, Uniqueness, Relevance and Credibility: the talent that multiplies when it plays together. Structure is the sound system. If the mic cuts out, none of that talent reaches the room. Evaluation is the mixing board.

“Fix the sound system, then let the band play.”

The real estate

Your house, the walls and wiring and furniture, is Structure and Relevance: what you build and own. The land underneath and the reputation of the street is Offsite, and you don't build that, you earn it.

“A gorgeous house on a bad block still appraises for nothing.”

Seven changes, ranked by impact

Each is drawn straight from the manuscript. The order is what I'd do first.

Make “information gain” the spine of Uniqueness

What: Rebuild the U pillar around information gain: does the page add something genuinely new, or restate what's already indexed?
Why: The book calls it “the concept that explains everything.” It reframes the AI-content panic: the risk isn't getting caught using AI, it's publishing hollow pages that add nothing and bleed authority. Slop is fast, generic, empty content, whoever wrote it. It's also the memorable, quotable idea the talk lacks.
From the book: The predictability test: “could a competent writer, or an AI, have produced this without doing the work?” Google's own example: “7 Tips for First-Time Homebuyers” (commodity) vs “Why We Waived the Inspection: A Look Inside the Sewer Line” (they have to come to you). Original data gets cited 30 to 40% more.
Do this: Rebuild U around information gain and the predictability test. Consider a callback in the open: it's sticky enough to carry the whole talk.

Upgrade “why this matters” with the compressed-journey stats

What: Add a beat to Part 2: AI has become the new referral engine, and the sale is mostly decided before you're ever in the room.
Why: The current “be the source or be forgotten” slide is good but soft. These numbers are what make a business owner's stomach drop, and they double as your pitch.
From the book: 6sense, ~4,000 B2B buyers: ~60% of the journey is done before they contact a seller; ~4 of 5 shortlist names are locked before they fill a form; the winner comes off the day-one shortlist ~95% of the time; the first vendor contacted wins ~80% of the time.
Do this: Insert one “the shortlist is set before the first call” slide between what's-changed and the proof. It sharpens urgency and pre-sells the offer.

Fix the Evaluation pillar: it's measurement, not page speed

What: Rewrite the E slide. Move page speed, mobile and bounce to Structure, and make E about measuring the channel.
Why: In the book, UX and behavioural signals live under Structure. Evaluation is the measurement dashboard, and it's more useful. “E for page speed” is off-model and a weak note to end the framework on.
From the book: Two numbers that aren't the same: citation share (how often AI quotes you) vs AI-referred traffic (humans who click), and they don't move together. The prompt library: 20 to 50 real buyer questions run monthly. The “direct traffic lie”: AI referrals hide in Direct because the referrer is stripped.
Do this: Rebuild E around citation share vs AI-referred traffic and the prompt library. It lets the call promise “we'll measure your citation share.”

Give each pillar one sharp fact it's missing

What: The pillar slides list generic best-practice. Add one specific, memorable book fact per pillar (full set in the table below).
Why: Specifics are what make you sound like the person who does this daily, not a summary of the theory. Each is a “write that down” moment a generic SEO webinar can't match.
From the book: JavaScript is “the silent killer”; brand mentions are 3× more correlated with AI visibility than backlinks; “AI doesn't quote vibes”; case studies with real numbers beat testimonials.
Do this: Drop one fact into each of the six pillar slides. One sharp fact beats four bullet points.

Add stronger, more current proof than Kadi alone

What: Keep Kadi, but add one or two book case studies with harder, AI-era numbers.
Why: Kadi is a single, mostly traditional-SEO example. The book gives you a portfolio of AI-specific proof that maps straight to the thesis.
From the book: Sylvane: LLM mentions +600%, AI-referral revenue more than doubled month on month, ranked #3 for “dehumidifier”, a term they'd never ranked for. JobAdder: bottom-of-funnel conversions +2,600%. The Hairy Pill: 2,766% ROI from E-E-A-T around one named expert.
Do this: Use Sylvane on the proof slide (it proves the whole formula) and hold The Hairy Pill for the Credibility pillar.

Point the “Claude Skills” giveaway at the real toolkit

What: The bonus names “Top 10 Pages, URC Elevation.” The book ships 17 real, numbered Claude skills.
Why: A far richer, genuine giveaway that ties the webinar to the book and its companion site, the real asset you're building toward.
From the book: Search-Era Self-Audit, SOURC-E Scorecard, Structure Audit, Brand Mention Builder, Original Data Finder, Topical Authority Mapper, Credibility Packet Builder, AI Visibility Dashboard, and more, one per stage.
Do this: Hand out the SOURC-E Scorecard skill plus the companion site, and use it as the reason to collect emails.

Tie the offer to the book's core advice: diagnose the constraint

What: Frame the strategy call as “we run your SOURC-E diagnostic and find your one constraint,” not a generic audit.
Why: The book's closing advice is explicit: don't work the pillars in order, diagnose your actual constraint first. This makes the call feel like the next step of the lesson, not a bolt-on pitch, exactly what the “do we have a deal” open promises.
From the book: “The formula tells you where to look. Start by diagnosing your actual constraint.” Most teams default to content because it's satisfying; the constraint is usually Structure or Offsite.
Do this: Reword the offer slide: on the call we find your single biggest constraint across the six pillars and the three fixes that move it, whether or not you work with us.

The one fact to add to each pillar

Six “write that down” moments, one per pillar slide.

S · Structure	JavaScript is “the silent killer”: AI crawlers (GPTBot, ClaudeBot, PerplexityBot) don't render JS, so your content can be invisible to them while looking perfect to you. The 30-second test: `curl -A "GPTBot" yoursite.com`. Also: ~9 of 10 ChatGPT citations come from pages in Google's top 10, and FAQ schema makes a page ~3× more likely to appear in AI Overviews.
O · Offsite	“Brand mentions are the new backlinks”: in a 75,000-brand study, mentions were 3× more correlated with AI visibility than backlinks. Branded search demand is the signal that can't be faked. AI reads Reddit heavily (the consensus layer, worth $60M/$70M in licensing to Google and OpenAI) yet cites it directly only ~2% of the time.
U · Uniqueness	Information gain: could a competent writer or an AI have produced this without doing the work? If yes, it won't get cited. Original data is the most defensible moat, cited 30 to 40% more. Own a monopoly on a fact.
R · Relevance	“AI doesn't quote vibes.” “Powerful automation platform” is invisible; “automates lead routing for B2B sales teams on HubSpot with 50+ reps” gets cited. The steal-able move: ask the AI what sub-questions it wants answered for your topic, then answer all of them. Answer first, explain second.
C · Credibility	Claiming vs proving. Case studies are the highest-impact asset: “we increased traffic” vs “340% in 8 months, 2,100 leads, $1.4M pipeline.” Named authors matter (E-E-A-T); Google's own raters check the About page ~6 times. Anonymous expertise is a contradiction in the AI age.
E · Evaluation	Citation share (how often AI quotes you) vs AI-referred traffic (humans who click), two different numbers that don't move together. Build a prompt library of 20 to 50 real buyer questions and run them monthly. Most of your AI traffic is hiding in the “Direct” bucket.

Fixes to the merged deck

Re-merging the original SOURC-E visual slides gives a fuller deck. A few workshop leftovers came with them.

The “48 HOURS” leftover

The About slide still reads “your host for the next 48 hours,” from the two-day workshop. For a webinar it should be “for the next 60 minutes.”

Workshop framing

The two IA-example slides still say “Hawk Academy Workshop: Melbourne, 11th + 12th June,” and “paste them into Claude” reads like a workshop exercise. Cut those slides or strip the references.

Slide numbering

The deck is numbered to 25, but the re-inserted detail slides aren't renumbered, so the footers jump (05/25, then unnumbered, then 09/25). Renumber to the true count.

Duplication

There are now two “SEO is alive” treatments and two query-fan-out treatments. Pick one per idea so the deck doesn't say the same thing twice.

On adding Sophie Brannon as co-host

The title now reads “Harry Sanders & Sophie Brannon.” Two things follow from a two-host webinar.

Introduce her. The About section only covers Harry. Sophie is a co-author of the book and a recognised SEO leader, so a co-host adds authority rather than diluting it, but only if you introduce her credentials. Add a short Sophie bio to the About moment.
Split the talk-track. The current script is written for a solo presenter. A two-host webinar needs handoffs: for example Harry owns the open, the deal, why-it-matters and the offer, while Sophie teaches two or three pillars. Tell me the split you want and I'll rewrite the talk-track with handoff cues.