Film day: the VSL and the confirmation video.

If you run a business doing at least a million a year, you already know you should be showing up when someone asks an AI who to use. You've probably run the search yourself, for the thing you actually sell, and watched it name two or three companies that aren't you. What you don't have is any idea what to do about it. And if you've just rebuilt your site, or you're about to, this is more urgent for you than anyone, because that is the most common way a business drops out of the answer. What makes it worse is you're already paying somebody for search, every month, and nobody can give you a straight answer on whether it's working. The report comes back green. The revenue is flat. In thirty minutes we'll tell you exactly where you stand and what's causing it, free, and in writing. And if you run marketing and this got forwarded to you, it's for you more than anyone.

Direct to camera, no music, no logo, no name yet. Slow through the first three sentences and let "companies that aren't you" land properly, that is the moment the whole video hangs on. Then tighten and speed up through the three pains. Flatten right out for the offer so it sounds like a fact rather than a pitch. Last line warm, small smile. If there is one place to do a second take, it is the pause after "aren't you".

Migration added as a qualifier, per Harry's note. It earns the place: a rebuild is now the lead cause in the diagnosis, it is the highest-converting angle in the ad bank, and it is the one circumstance where the viewer already knows something changed and has been waiting for someone to explain it. Naming it here and again in beat 3 is deliberate, because the hook qualifies and the diagnosis explains. "The ad bill climbs every year and none of it compounds" came out to pay for it; the compounding argument is now made properly in 3b rather than glanced at twice.

I'm Harry Sanders. I started StudioHawk a decade ago when SEO got serious and built one of the first SEO only companies with direct access to specialists. Now a hundred and twenty specialists across Melbourne, London and Atlanta, running AI SEO for QuickBooks, Weber and HelloFresh, and we've been named best large SEO agency seven times. Here's what that looks like in money, not rankings. Case-Mate, ninety-five percent more revenue. Clarks, a hundred and one percent year on year in organic revenue. Tapt, six hundred and eight percent return on investment. Then Bondi Sands, Aje, Acuity, Equity Estates, The Entourage, Dogline, Evolve, Paragon, SnowsBest. All of them written up on our site, with the actual work, not just the number. And we only do this. No paid, no social, no web build. AI Search is the whole business, which is the reason we can go deeper on it than a full-service agency ever will.

This is the overwhelm beat and the screen carries it, not the voice. Build the wall from studiohawk.com/case-studies: every case study as a card, real thumbnails and real headline figures, filling the frame while the names are said. Three money results land one card at a time, large. Then the whole grid at once for the name sweep, and hold it two seconds past the last name so the eye can wander. No animation on the figures, no logo carousel sliding, it is a wall not a parade. Deliver the nine names at a clip, flat, no emphasis on any one of them, then slow right back down for "we only do this". BEFORE BUILDING THE CARD: confirm the exact count. Two reads of the page returned seventeen and eighteen, so the script deliberately says "all of them" rather than a number. If you want a number on screen, count it on the day. Also confirm The Entourage is still live on the page, it appeared in one read and not the other.

Harry's founding line merged in rather than promoted over the top. His edit was written against the version of this beat from before the case study wall, so it carried no case studies, and taking it verbatim would have deleted the wall he asked for in the same session. Merged, both survive. The line earns its place on SOP-03 step 2, which wants relevant experience rather than backstory: "one of the first SEO only companies with direct access to specialists" is a differentiator, not an origin story, and it sets up the no-account-manager promise in beat 5 four beats early. The ranking behind the wall is unchanged and it is the one the script already argues, money before rankings, because this video says out loud that it measures revenue. Case-Mate, Clarks and Tapt are the only three published money metrics on the page. The nine names after them are recognition and category spread. Sylvane and The Motor Enclave stay held back for the close. This beat is now the second longest in the script. If it needs to come down, the split to make is money here and the name sweep moved to the close, not cutting Harry's opening.

So why aren't you in the answer? It's usually not that you need more content. We built a framework called SOURCE, built using thousands of campaigns and millions of data points that we have collected over a decade, which maps issues down to either Structural, Offsite, Uniqueness, Relevance, Credibility, and Evaluation. Most SEO is still behind, so none of these show up in the report you're being sent, which is exactly why it's green but you are not getting more leads.

Overlay the SOURC-E whitepaper as the six are named, one pillar highlighting as each word is said, and hold the full framework on screen through the last sentence. This is the only place in the video where a graphic is doing the teaching, so give it room and do not cut away early. Slower than everything around it. Stop dead after "more leads". HARRY, WHICH FILE: there is still no SOURC-E whitepaper on the hub or the public site. Candidates are the Format Leverage study, 1,203,748 AI referral sessions across 600-plus businesses, and the SOURC-E deck the webinar teaches from. Send the PDF and this gets pointed at it. ONE SPELLING NOTE for the overlay: the framework's S is Structure everywhere else in the funnel, and the script says Structural. Fine spoken, but the card should use the canonical six so it matches the scorecard and the webinar deck.

Harry's rewrite, promoted verbatim with one typo fixed. It trades three named causes for the framework itself, which is a bigger claim and a better one: three causes is a diagnosis, a framework built on a decade of campaigns is a reason to believe the diagnosis. It also puts SOURC-E at position three instead of position five, so the name is established before the offer uses it. Two consequences worth knowing. The offer beat no longer has "three things" to point back at and now points at the six. And the hook still promises that a rebuild or migration is the most common way a business drops out; that is now carried implicitly by Structural rather than said out loud, which is consistent but quieter. If retention dips between the hook and the offer, putting the rebuild back into this beat as the example of a Structural failure is the first thing to try.

One thing about timing, and I'll be straight rather than dramatic about it. This is the biggest change to how people find a business since Google. Nine and a half billion visits a month across the AI tools now, up seventy percent in a year, and that's Similarweb's measurement, not ours. It's early enough that the answers aren't settled yet. That matters because being cited compounds. Once these models learn to name you, it reinforces: you get named, which creates more places that mention you, which makes you easier to name next time. Which means the reverse is also true.

Flat and unhurried. No music swell, no leaning into camera. The whole beat fails if it sounds like pressure. Put the Similarweb attribution on screen as a plain card while you say it. End on "the reverse is also true" and cut, without explaining it.

Harry's trim kept exactly: the beat ends on the implication rather than spelling it out, which is stronger and shorter. Worth noting what went with it, because it was load-bearing: "there is no deadline on this call and I am not going to invent one" was the line that separated this from a closing-window pitch, and the ICP's first fear is not knowing who to trust. The beat still refuses to name a deadline, so the protection is structural rather than stated, but if this tests as pushy that line is the first thing to put back. The growth figure answers Harry's note about volume data, and it sits here rather than in the offer because this is the beat where a number is an argument. Similarweb's own panel, published 29 July 2026, 9.5 billion average monthly visits across generative AI platforms, up 70% year on year from June 2025 to May 2026. Attribute it out loud every time. It is a third party measuring a market, which is a different and better claim than a vendor measuring itself.

So here's the offer. Thirty minutes with a senior strategist, free. Before it we crawl your site, grade it against SOURCE, the six signals that decide whether a search engine or an AI recommends you, and against the one point two million AI referrals we've measured ourselves, and tell you which of the six is actually stopping you. Then you get a custom ninety day action plan. You leave with all of it in writing whether you hire us or not. And we measure the thing you care about, which is revenue, not rankings.

The scorecard on screen while the deliverables are listed, six rows, some green some red, then the ninety day plan as a real document with dates on it, not a mockup. The SOURC-E overlay has already done its work in beat 3, so do not repeat it here. Even pace, no selling in the voice. This beat is a list and it should sound like one.

The framework is named here on Harry's note, and only named. Saying SOURC-E and then leaving it as shorthand would put the video in direct conflict with its own argument, because the funnel elsewhere attacks agencies that run on "acronyms, dashboards nobody reads, and a quiet implication that not following is your problem". So the script says the name once, in the same breath as the plain-English definition it already had, and the overlay carries the six words. Two spoken words instead of forty, and the viewer sees the framework rather than being told to remember it. The custom ninety day action plan stays the biggest upgrade in this beat: it turns the audit from a diagnosis into something with a next step, and it is what a marketing manager takes to their owner.

What I want you to do - in a calculator, write down your ad spend and agency spend last month and multiply it by 12x. That is your rent, the second you stop paying you lose the traction, AI SEO is owning, it compounds, real estate that you can't buy and audience that you are losing because you can't reach them where they search now. Compare that to the cost if you want us to run the plan afterwards, it starts from $2000 a month, month to month.

Slow right down and let them actually do the sum. A real calculator on screen, or a pen and paper, while you say the first line. Say the price at normal pace with no apology in the voice, then stop.

Harry's rent argument, given its own beat rather than tacked onto a 164-word offer. It reframes the retainer against money already being spent, which the corpus supports: 88% of calls raise price unprompted and the objection is almost never the number, it is what the number buys. Owning versus renting is the one claim paid media can never make. Two sentences run on when read aloud, the one starting "That is your rent" and the one starting "Compare that", so they want a pass before the teleprompter session. Left exactly as written until Harry says otherwise.

Three things worth knowing. There's no lock-in contract, no twelve or twenty-four month tie. You get direct access to the specialist doing the work, not an account manager relaying messages, because the person who knows your site should be the person you can actually talk to. And we only do SEO, so nobody's going to try to sell you a rebrand halfway through.

Flat and factual. Three points, three beats, no flourish.

All three differentiators, said as facts rather than promises. The direct-access one answers a real pain in the corpus: one prospect had five account manager changes in eighteen months and said no knowledge was ever handed over.

A few things people think but don't ask, so let me answer them. If I stop paying, does it all disappear? No, and that's the whole difference between this and ads. Would this even work for my business? We do this for a phone case brand, a shoe retailer, a race track and a bookkeeping firm. The SOURCE signals are the same in all of them, what changes is which one is broken. Should I have sorted this out years ago? Probably, and everyone says that, and it's the least useful thought available to you, because the only version of this you can actually act on is today's, imagine going back in time and being at the start of SEO, this is your chance but for AI Search. And do I have to need you forever? No. Plenty of our clients aim to bring this in-house eventually, and we'll teach you as we go so you have the option to.

Slower here. This is the trust beat and it fails if it sounds like a list being read.

Harry's addition kept verbatim, and "would this even work for my business" added on his note, answered with four real clients in four unrelated categories rather than with reassurance. Case-Mate, Clarks, The Motor Enclave and Acuity, all published. Naming the categories instead of the logos is deliberate: it lets the viewer find themselves in the list rather than being impressed by it, which is what the objection actually needs. Four questions now, each removing a real barrier, and the last one names the ICP's stated secret desire out loud, which no agency ever does.

Most people we do this for already have someone. Bring them to the call. Either they already knew about everything we find, which is worth knowing, or they didn't, which is worth more. You're welcome to take the list straight to them and never speak to us again. That happens, and it's genuinely fine.

Generous delivery. Any hint of a smirk here loses the room and the forward.

52% of the corpus has or had an agency and 34% of calls have more than one person on the prospect side, so this video gets forwarded. It has to be safe to forward.

One thing before you apply. You should already be doing something with SEO, an agency now or one you've had before. If you're starting from absolute zero, we're not the right first call, and I'd rather say that here than take the meeting.

Straight, unapologetic, and quick. It is a filter, not an apology.

Cut from 18 seconds to 12. It used to restate who this is for, which the hook now does in its first line, so all that is left is the part that does work: the ICP's hard filter that they are not starting from scratch. SOP-03 step six only asks for who it is and is not for, and saying it twice weakens both.

The button under this video is a short application. We read every one, and if it's a fit you book straight onto the calendar.

One CTA. No alternatives, no second option offered.

Check the question count on the live form before filming and say the number if someone has counted it today.

Sylvane sell air quality gear against Home Depot and Lowe's. They thought they had no chance against the giants. On top of that, they were moving platforms, which is exactly where most businesses lose ground, and came out with brand mentions inside the AI models up six hundred and twenty-one percent, and AI referral revenue up a hundred and forty-five percent month on month. Their review of working with us was four words: five stars, no notes. The Motor Enclave rebuilt their whole site and came out two hundred and eighteen percent up in visibility instead of down. Worst case here, you spend thirty minutes and finally find out whether you're in the answer or not. Best case, you find out what's been quietly costing you while it's still cheap to fix.

Two results only, named, no logo carousel. Put Ryan Dobrin's name and title under the quote, it is published and it should look like it. Pull both write-ups from studiohawk.com/case-studies so the cards match the live pages. Close on the last sentence and cut. No outro card.

Cut from five results to two, because the three money numbers moved up to the credibility beat. What is left is the two that argue something specific rather than just proving competence. Sylvane is the only AI-native result in the bank, the only place we can say brand mentions and AI referral revenue in the same sentence, and the only attributed quote we hold besides Ashlie McKinnon's. The Motor Enclave is the rebuild, which is now the lead cause in the diagnosis and the qualifier in the hook, so it closes the loop the video opens. It carries its visibility figure only, because that case study publishes no revenue number. Say 621% as brand mentions and 145% as AI referral revenue month on month; overstating either turns the strongest proof we have into a liability.

You're in, and booking that call was the right move. Thirty minutes on Zoom with a senior member of my team, talking about where AI search is sending customers in your market and what to do about it.

Warm, direct to camera. This plays muted first, so the opening frame should show a caption card: "You're booked. Watch this before your call."

So you know exactly who you're talking to: the same team that runs search for QuickBooks, Weber and HelloFresh, and the framework behind Samsung finding us through ChatGPT. The call itself is a straight conversation: where AI search is sending your customers, what we'd do about it in your specific market, and whether we're the right team to run it. Camera optional, nothing to prepare.

One thing to do right now, and it takes twenty seconds. Your confirmation email just landed. Hit reply and tell us the single thing you most want out of this audit. More leads, less ad spend, a specific competitor that keeps showing up where you should be. The person running your call reads every reply.

SOP-32 manual-touch move: the reply starts a real thread with the setter/closer before the call.

And so you know exactly what this is: it's a conversation, not a pitch. If we can genuinely help, we'll tell you exactly how. If we're not the right fit, we'll tell you that too and point you at the better path. Either way you leave knowing where you stand.

Last thing: a senior person is holding that slot for you, and we book a limited number of these. If something comes up, use the link in your email to move it rather than miss it. See you on the call.

Honest scarcity only: senior time genuinely is the constraint. No countdown mechanics.

You're booked, and the call is in your calendar. Before it, eight minutes on why this is worth doing now rather than at some vaguer point later. I'm not going to invent a deadline, because there isn't one on a free audit and you'd smell it. Instead, we went back through our own sales calls, hundreds of them, people in almost exactly your position, and four things came up often enough to be worth your time. Then I'll tell you what to bring so you get a plan out of the thirty minutes instead of a conversation.

Straight to camera, sitting, no desk set dressing. This is the most human piece in the funnel and it should look like it was made for one person. Say "I'm not going to invent a deadline" plainly and move on, no pause for effect. No music under this beat.

SOP-28 wants the confirmation page to answer, not to sell, and SOP-18 puts a personal video at the centre of show rate. The sample size is deliberately not named on camera, on Harry's note: reciting 222 makes it sound like a research paper and puts the viewer in the audience rather than in the conversation. Saying we went through our own calls does the same job in half the words. The counts stay on this page, where the evidence needs to be checkable, and out of the mouth, where it does not.

Here's the first one, and it's the least dramatic and the most useful. One in five people we speak to asks us, unprompted, how long this takes. It's the most common question we get, and the answer is that it depends on the site, but it's months rather than weeks, and that is true no matter who you hire. Which means the size of the result isn't set by how hard you push once you start. It's set by when you start. That's uncomfortable, and it's also the only genuinely good news in this video, because it's the one variable that's entirely yours. Everything else, the competition, the algorithms, what the models decide to quote, none of that is in your control. The date is.

Slow. This is the argument the whole video rests on and it should sound like something you'd say to a friend rather than a prospect. On screen, one card: one in five asks how long it takes. Nothing else. No countdown graphics anywhere in this video.

The strongest real urgency available, and it is entirely honest: 46 of 222 asked about time to results without being prompted, which is 21%. It reframes delay as a cost without threatening anyone, and it is the one urgency argument that survives contact with a sceptic, because they already believe the premise. Jeremy's SOPs permit urgency only where it is genuine, and this is as genuine as it gets.

Second. More than a third of those conversations had a website rebuild, a migration or a replatform in them somewhere. Some were three months out. Some had just gone live. The difference between those two groups is the entire reason I'm making this beat. One person put it better than I could: they went live with a pile of new product pages, then worked out afterwards what they'd broken, and, in their words, here we are. That's the sequence to avoid. A migration is the single fastest way to become unreadable to a search engine and to an AI, and it is also completely survivable if the mapping is done before the switch instead of diagnosed after it. So if you've got a rebuild, a rebrand or a replatform coming, say so on the call, because it changes the order of everything we'd recommend. And if you've just done one and things feel quieter since, say that too. That's not a coincidence, and it's usually fixable.

Screen: a real crawl of a post-migration site, dead URLs and redirect chains visible, cursor moving. Do not name the client. Say "here we are" as the quote it is, with a beat before and after it.

79 of 220 calls raised a migration, rebuild, replatform or relaunch, which is 36%, the second most common circumstance in the corpus after competitors. It is also the highest-converting angle in the ad bank and the lead cause in the VSL's diagnosis, so this beat closes the loop for anyone who came in on a migration ad. The quote is verbatim from a prospect and the client is deliberately unnamed.

Third. More than half brought up a competitor without being asked, and more than a dozen could name the exact rival who is outranking them right now, today. Here's why that matters more than it used to. Being cited compounds. Once a model learns to name a company, that creates more places that mention them, which makes them easier to name next time. The gap doesn't stay the same size while you think about it. It widens on its own, slowly, in a direction that isn't yours. That's not me being dramatic, it's just how a feedback loop works. Before your call we'll pull the actual answers where a competitor is being named instead of you, with the question and the engine, so you're not taking my word for any of this. You'll see the list.

Screen: a live AI answer naming competitors, streaming, unedited, including the pause before the names appear. Do not speed it up. This is the single most persuasive frame in the funnel and it costs nothing to shoot fresh. Re-shoot monthly, a stale screenshot is a liability.

114 of 220, 52%, is the most common theme in the entire corpus, and 16 could name a rival outranking them that day. The compounding argument is the honest version of the ICP's fear of being left behind and it is the same mechanism the VSL's timing beat uses, so the two reinforce rather than repeat. Note what this beat does not do: it never says the window is closing.

Fourth, and this is the one that changed how I think about the whole thing. Nearly a third brought up AI search themselves. We didn't raise it. And seven of them, seven actual people, told us on the call that they found us through it. They asked ChatGPT who the best SEO agency was, and it said us, and that's why they were sitting there. Think about what that actually means. The mechanism I'm describing already worked. It worked on somebody with your job, with your budget, deciding who to trust with a real amount of money, and they never saw an ad. That's the whole product in one sentence. The only question left is whether it's working for you or for the company that comes up when someone asks about your category.

Screen: the prompt typed live, the answer streaming. Then cut back to Harry for the last two sentences with nothing on screen at all. Deliver "and they never saw an ad" flat, do not lean on it.

7 of 220 said they found StudioHawk through an AI answer, which is 3%. Small number, enormous argument, and the reason to state it as a count rather than a percentage: seven real people is concrete and 3% sounds like a rounding error. This is the only proof in the funnel where the product demonstrates itself on the buyer, and it is currently used nowhere else.

One more thing, quickly, because more than a third of those calls were with people already running paid. If that's you, you already know the shape of it. The leads stop the day the card stops. The cost per lead goes one direction over time, and it isn't down. None of it accumulates, so year three costs more than year one for the same result. I'm not going to tell you to turn your ads off, that would be stupid advice and we don't even do paid. But on the call it's worth putting the two numbers next to each other, what you spend a month to rent demand, against what it costs to own a position that keeps working. Bring the first number. We'll do the comparison honestly, including the cases where the answer is that ads are still your better buy.

Screen: nothing, or a simple two-column comparison drawn live. No graphs. Say the last sentence like you mean it, because it is the line that makes the rest of the beat believable.

81 of 220, 37%, were running paid and comparing cost per lead. This is Harry's rent-versus-owning argument from the VSL, made once more with the corpus number behind it, and the concession at the end is doing real work: admitting that ads sometimes win is what buys permission for everything before it. SOP-25 and the ICP both point the same way, the buyer's first fear is being sold to by someone who would say anything.

So, the call itself. Thirty minutes, and a senior strategist, not a salesperson reading a script. Before it we crawl your site and grade it against SOURC-E, our six signals, and we pull the answers where you're being left out. You'll get the score, the citation gap, and a ninety day plan in the order we'd actually do it. You keep all of it in writing whether you hire us or not, and that's not a trick, it's just cheaper for us to be useful to a hundred people and work with three of them than to chase a hundred. If you do want us to run it afterwards, it starts from two thousand dollars a month, month to month, no lock-in. I'm telling you the price now so it isn't a reveal at the end of a call you've already given half an hour to.

Screen: the scorecard, the citation list and the ninety day plan as real documents. Price said at normal pace with no apology in the voice.

SOP-28 says the confirmation page pre-empts what the call would otherwise have to spend time on, and SOP-03 says the price is stated plainly and early. Saying it here rather than on the call is the single biggest show-rate and close-rate lever in the sequence, because 88% of the corpus raises price unprompted and the ones who feel ambushed by it disengage before the second half.

Four things, and they take about ten minutes. One, three search terms, written as a sentence a real customer would type, not keywords. Two, go and ask ChatGPT who it recommends for that thing in your area, and write down what comes back, including the names that aren't yours. Three, if you're running ads, the monthly number. And four, if someone else runs your marketing, get them on the call. Not because you need permission, but because they're the one inside the site, and a plan that arrives second-hand dies in the handover. Do those four and the thirty minutes goes on what to do. Skip them and we'll spend half of it gathering information you already had.

On screen as a numbered list, held long enough to screenshot, because some people will. Say it slowly enough to be written down.

This is the agency half and it is also a show-rate mechanism: someone who has done ten minutes of preparation is materially more likely to turn up, and someone who has run the prompt has already seen the problem for themselves. Getting the marketing manager on the call comes straight from the corpus, where the owner buys and the manager executes, and the split is where good plans go to die.

Whatever you decide, you leave with the scorecard, the list of answers where someone else is named instead of you, and the ninety day plan. In writing, yours to keep, and yours to hand to whoever you like, including another agency. That's a deliberate decision on our part. If the plan is good enough to be worth stealing, that tells you more about whether to work with us than anything I could say in this video.

Flat, no smile on the last line, it is stronger straight. Documents on screen one more time, briefly.

Risk reversal, and the line about handing it to another agency is the strongest trust move available here because it is verifiable within thirty minutes of the call ending. It also quietly answers the corpus's most common private worry, which is being locked into something.

That's it. Block the time, do the four things, and bring the person who'd actually do the work. If something comes up, reply to the confirmation email and move it rather than skipping it, because the audit gets built either way and it's a waste for nobody to read it. See you on the call.

Warm, small smile, then cut. No outro card, no logo sting, no music.

The reschedule line is deliberate and it is worth defending against anyone who wants it removed for sounding soft. A no-show costs the build we have already done; a reschedule costs nothing and keeps a qualified person in the funnel. SOP-06 exists because the alternative is losing them silently.

You're in. Your seat is locked for the live session, and I'm glad you grabbed it.

Straight in, warm, energy up. Same framing as the landing VSL so it feels like one conversation.

First thing, and do it while you're here: hit the add-to-calendar button right under this video. The Zoom link arrives by email closer to the day, and we'll text you before we start. But the calendar entry is what actually gets you in the room, so do it now.

Point down at the calendar button under the video. This is the whole reason the video exists.

Second: set up a free Claude account at claude dot ai. Two minutes now saves ten on the call. And on the day, join from a computer with your website open, because this is a working session, not a watch-along. You'll be grading your own site while we go.

Matter of fact. These mirror steps 2 and 3 on the page.

Here's what the two days get you: the SOURC-E framework end to end, your own site graded against all six letters on day one, and the full SOURC-E toolkit handed over and installed on day two. You'll leave knowing your weakest letter and exactly what to fix first.

Paint the payoff. Builds anticipation, which is what show-up actually runs on.

Quick word on who's teaching. My agency runs SEO for brands like QuickBooks, Weber and HelloFresh, and Samsung found us through a ChatGPT recommendation. This session is that same method. And it's not a stealth sales pitch: you paid seven dollars, and you're getting the work.

Flat, no flex. The due-diligence beat for anyone about to Google you.

So: calendar, Claude account, computer on the day. That's it. See you live.

Count them off on fingers. End with eye contact, hard cut.