Verity Agentic: 5-Model Panel Review
Five reasoning models from five labs, and the two findings they got right that I had missed. Enter hub password to continue.
Wrong password
verityagentic.ai · panel review · 2026-07-30
What Five Models Saw
Opus 5, Gemini 3.1 Pro, GPT-5.6, Grok 4.5 and GLM 5.2 each read the same five pages. Every finding below was checked against the live site before it was written down, including the two that turned out to be real bugs.
The two findings that matter most
1. VERIFIED BUG: the site publishes two different price sheets
Three of five models raised this, and two made it their number one priority ahead of everything else. It is real:
| Offer | pricing.html | for-coaches-and-therapists.html |
|---|---|---|
| The Bottleneck Audit | From $500 | From $500 (consistent) |
| Done-For-You Workflow | $3,000 to $8,000 typical | From $1,500 |
| AI Ops Retainer | $1,500 to $3,000/month | From $500/month |
The build differs by a factor of two at the floor. The retainer differs by a factor of three.
Why the panel weighted it so heavily: the pricing page's own headline is "Pricing, plainly stated" and "No mystery pricing." A buyer who lands on the coaches page, sees $1,500, clicks to pricing and sees $3,000 does not conclude that one page is stale. They conclude the price depends on which door you came through. GLM-5.2 put it well: this is an active leak rather than a missed opportunity, and it is losing people who were already sold.
This predates the current work. It was not introduced by the recent changes; the coaches page has carried its own numbers all along.
Needs Annette's decision, not a guess. Two coherent readings, and picking wrong publishes wrong prices: - The coaches numbers are intentional entry-level pricing for a simpler single-workflow vertical, in which case pricing.html needs to acknowledge that floor ("from $1,500 for a single workflow, $3,000 to $8,000 typical"). - The pricing page numbers are the real ones and the coaches page is stale, in which case it gets corrected upward.
Given the recent decision to focus on small projects, the first reading is plausible, which is exactly why this should not be resolved by inference.
2. VERIFIED MISS: the new sales page is nearly orphaned
Four of five models flagged it independently. Confirmed by crawling the built site:
- Pages linking to
bottleneck-audit.html: onlypricing.html. - It is not in the nav.
- The homepage mentions "Bottleneck Audit" exactly once, inside an FAQ answer, with no link.
So the single most important conversion page on the site sits two clicks deep behind a page most visitors never open. This one is my error from the previous session: the page was built, wired into the sitemap, llms.txt and the pricing ladder, and then not given a route anyone would actually travel.
Fix, and it is small: 1. Add the Bottleneck Audit to the primary nav. 2. Add a homepage section for it directly under the hero, so the paid entry offer appears above the fold region rather than being implied. 3. Link the FAQ mention. 4. Link it from the coaches page offer card.
Consensus stack
Themes ranked by how many of the five models raised them independently.
| Raised by | Finding | Status |
|---|---|---|
| 4/5 | Audit page not linked from nav or homepage | Verified, my miss |
| 4/5 | Proof is dog rescue and teaching, not the buyer's world | Valid judgment call |
| 3/5 | No sample of the Automation Makeover deliverable | Strong idea |
| 3/5 | FREE REPORT still competes in the nav | Valid |
| 3/5 | No risk reversal or guarantee anywhere | Valid |
| 3/5 | Pricing contradiction between pages | Verified bug |
| 2/5 | No way to actually pay; every CTA routes to a free call | Strong, needs a decision |
| 2/5 | Before/after table weak on mobile | Needs checking |
| 1/5 | Hero copy assumes the reader already has an AI system | Worth considering |
| 1/5 | The 3-to-5-day and 1-to-2-week promises are unbacked | Worth considering |
The rest, with my read on each
Proof doesn't match the buyer (4/5)
The case studies are 31,310 keywords for a dog rescue and a teaching business across seven countries. Genuinely impressive, and completely outside the world of a therapist wondering whether her intake can be automated. Opus put it sharply: the credentials prove Annette can build, not that she has solved this problem for this person.
This is the strongest argument yet for placing Bob's testimonial prominently. Bob runs a seminar business, is a licensed counselor, and the automation described (an assessment that scores itself and emails a personalized response) is exactly what a coach or therapist would want. He is the missing bridge between the credentials and the buyer.
My read: valid, and largely solved tomorrow when the testimonials land. But it also argues for writing one short case study in the buyer's own vertical rather than relying on the existing two.
Show a sample of the Automation Makeover (3/5)
Grok's version was the most concrete: a band on the audit page showing the actual shape of the deliverable, with five example cards (bottleneck named in one sentence, priorities in order, why that order, build-versus-DIY for each, fragility notes).
My read: this is the best new idea in the panel. Someone paying $500 for a document has no idea what arrives. A visible skeleton removes that uncertainty, and it costs nothing but a page section. It also quietly reinforces that the Makeover is a plan rather than a build, which is the exact ambiguity flagged when "Plan" was dropped from the name.
No risk reversal (3/5)
Nothing on the site reduces the buyer's downside. The fee-credited-toward-a-build line is good and already there, but it only pays off if they continue. Options range from a plain satisfaction guarantee to something narrower like "if the session doesn't surface at least three things worth fixing, you don't pay."
My read: valid, and Annette's call. A guarantee is a business commitment, not a copy tweak.
No way to actually pay (2/5)
Every CTA on the audit page routes to a free call. Opus argued this takes money off the table three times on a page whose job is to sell a $500 product, and suggested a Stripe payment link as the primary with the free call as the secondary.
My read: genuinely arguable both ways. For a considered service purchase from a solo consultant, a call first is often the higher-converting path, and it protects against bad-fit clients. But the panel is right that there is currently no way for a ready buyer to just buy. Worth testing rather than assuming.
FREE REPORT still competes (3/5)
It was de-accented from a yellow pseudo-button to a plain nav link, but it is still in the primary nav offering a second free thing. Gemini's recommendation is to remove it from global nav entirely.
My read: agree, especially now. If the Bottleneck Audit needs a nav slot, this is the slot to take.
Where I disagree with the panel
- Gemini's hero rewrite ("Stop doing the manual work a basic AI agent can handle") is weaker than the current line. "Basic" undersells the work and invites the reader to think a cheap tool would do. The existing hero is fine.
- Several models suggested softening or conventionalizing the design. Out of scope by explicit constraint, and the noir brand is a genuine differentiator in a category full of beige consultant sites.
Recommended order
- Resolve the pricing contradiction. Blocked on Annette deciding which set of numbers is real.
- Un-orphan the audit page. Nav slot, homepage section, link the FAQ mention. Not blocked on anything.
- Place the testimonials when they land, with Bob on the coaches page and the homepage, since he is the bridge to the buyer's world.
- Add the Makeover sample band to the audit page.
- Decide on risk reversal and on whether to allow direct purchase. Both business calls.
Raw per-model responses: research-notes/llm-site-review-synthesis.md. Scripts: research-notes/llm-site-review.py and -retry.py, re-runnable.
Raw responses, per model
Every word each model returned, unedited. The synthesis above is my read on it, including where I disagree.
anthropic/claude-opus-5 anchor / arbiter, most reliable verdict under pushback
1. WHAT'S WORKING
1. The opening of the Bottleneck Audit page is the best copy on the site. "You read the intake form yourself, because nobody else knows what to do with it. There's a follow-up you meant to send on Tuesday. It's Friday. A client slipped last month because the handoff lived in your head and you were on a call."
Three sentences, present tense, no abstraction, no benefit language. A therapist reads that and recognizes her own Friday. Then "That's not a discipline problem. It's a structure problem" lifts the shame off, which is the actual barrier to buying an audit of your own operations. Whoever wrote this section should write the homepage.
2. "You've probably tried these already" (bottleneck-audit.html). Naming the no-code tool, the freelancer who left without documenting, and the ChatGPT subscription is competitive positioning at the level of the buyer's actual purchase history rather than against named vendors. "You bought a tool where you needed a system" is the sharpest line on the site. And "None of those failed because you picked wrong" is doing real work: it lets the buyer stay smart while admitting the problem persists.
3. "Do I get the automations built as part of this? No, and this is the one thing I want to be blunt about." This is the FAQ answer most consultants soften into mush. Putting the scope boundary in bold, unhedged, before the sale prevents the single most common $500 dispute. The willingness to answer "no" plainly buys credibility for every other claim on the page.
4. The therapist "What I will not build" list (for-coaches-and-therapists.html). "Automated review requests tied to your sessions (a real board-complaint risk), outcome-guarantee copy, or any tool that puts client clinical data into a marketing system without a proper agreement in place."
This is the strongest differentiator anywhere on this site. It's proof by refusal, it's specific enough to be checkable, and it tells a licensed therapist that you understand the thing that keeps her awake. No generalist automation vendor can copy this paragraph without doing the reading.
5. "It will need attention eventually, and anyone who says otherwise is selling." Conceding fragility before being asked is the correct move for a buyer who has been burned. It also sets up the retainer without pitching it.
2. COPY CRITIQUE
1. The origin band ends by telling the buyer what to conclude.
Current:
"I built adoption.com in 1995, connecting families with children in crisis across three countries." / "Before Google existed." / "That's the kind of builder you want designing your AI systems."
Why it fails: the last line does the buyer's thinking for her, which reads as insecurity. Worse, there's no bridge. A Boulder therapist looking at her intake pile does not have a mental path from a 1995 adoption website to her Tuesday problem. You've handed her a credential and asked her to build the relevance herself. She won't.
Replacement:
1995 / adoption.com I built adoption.com in 1995, matching families with children across three countries. Before Google existed. The site did the matching, sequenced the paperwork, and chased the follow-ups, because there was nobody else to do any of it. One person can't hold that many handoffs in her head, and I was the one person. Taking that person out of the middle is the whole job. I've been doing it for thirty years.
2. The homepage stats section promises small-business insight and delivers enterprise trivia.
Current:
"Why most AI fails for small businesses (and how we fix it)" ... "95% of enterprise AI pilots failed to deliver measurable P&L impact" ... "92% of workers at companies with a clear AI strategy say it improved their team's productivity, 30 points higher than those without one"
Why it fails: the heading says small business, both stats are about enterprises. The second one is worse than useless for this buyer, because a solo coach has no "workers," no "team," and no "AI strategy," so the stat quietly tells her this page is for somebody else. And the section promises "how we fix it" then never says how. You've spent a full screen borrowing authority from Fortune and EY when your buyer's trust problem is about you, not about the category.
Replacement:
Most AI projects fail for a reason that has nothing to do with AI The tools work fine. What's usually missing is a written account of how the work actually moves, and without that there's nothing to hand to a system. That's as true in a one-person practice as it is in a bank.
95% of enterprise AI pilots delivered no measurable P&L impact, and they had budgets and staff you don't have. MIT NANDA Lab / Fortune, 2025
One person is usually all it takes to stall an entire process. In a business your size, that person is you, and you're also the one who has to notice.
3. Two voice bugs: the negation tic, and the we/I flip.
The negation tic first. The site defines itself by what it's not at least thirteen times:
"Real workflows, not pilots" ... "not pilots or slide decks" ... "Real systems that did the work, not slide decks about it" ... "It's a roadmap, not a build" ... "That's the normal case, not the exception" ... "That's usually the best time, not the worst" ... "What matters isn't headcount, it's whether..." ... "not a black box only I can touch" ... "in weeks, not the six-month timelines big vendors quote" ... "not just describe you"
Why it fails: two reasons. First, rhythm. Once a reader clocks the pattern, every future instance reads as a formula, and formula is the one thing plain-spoken authority can't survive. Second, and worse: half of these argue against an enemy this buyer has never met. A therapist has never been offered an AI pilot. She has never sat through a slide deck about agentic workflows. "Not pilots, not slide decks" is language aimed at a CIO, and you cut the $15,000 tier specifically to stop talking to that person. Keep the negations where the buyer has genuinely met the enemy (the freelancer, the ChatGPT subscription, the six-month vendor timeline). Cut the rest to positives.
Rewrites: - "Real workflows, not pilots" becomes "Workflows that are running by Monday." - "Real systems that did the work, not slide decks about it" becomes "I built and ran both of these myself, start to finish. They did the work while I slept." - "It's a roadmap, not a build. It tells you exactly what to do next" becomes "The Makeover is the map. It tells you exactly what to do next, and you can do it with me, with someone else, or yourself."
Second bug, same section. Inside a single homepage block you use three voices:
"We don't sell AI tools: we sell results." ... "As an AI systems consultant, Annette brings 30 years of systems thinking to every agentic AI engagement." ... "Teaching your people to run and change what I've built"
"We" from a solo operator reads as a small business pretending to be bigger, which is the exact instinct this buyer has learned to distrust. The third-person "Annette brings 30 years of systems thinking to every agentic AI engagement" is keyword copy wearing a sentence, and it's sitting three lines from genuinely good first-person writing. Pick first person and hold it on every page. Same fix for the hero's "Leader: Annette Thompson," which is a label nobody has ever used about a person they were about to hire. Make it "Built and run by Annette Thompson, solo since 1995."
4. The coaches page makes the one claim a visitor can check, and it fails the check.
Current:
"A lead funnel: a freebie or quiz, an automated email nurture sequence, a booked call. This is the exact system running on this page right now." "Built the quiz-to-email nurture system running on verityagentic.ai right now. It's not a pitch deck, it's live on this page."
Why it fails: there's no quiz on that page and no email capture on that page. The only CTA goes to contact.html. A coach reads "running on this page right now," scrolls looking for it, finds nothing, and now every other claim in that list is suspect. You've taken your best proof mechanic, a demo the buyer experiences on her own body, and pointed it at empty space. Also "It's not a pitch deck, it's live on this page" is the construction you've banned everywhere else.
Replacement, plus the build task of actually placing the quiz block on that page:
"A lead funnel: a short quiz, an automated email sequence, a booked call. I run one on this site. Take the ten-question audit at the bottom of this page and you'll watch it work on you, which is the only product demo I trust."
And in the proof list:
"Built the quiz-to-email intake that runs Verity Agentic's own front door. It takes about 90 seconds. Go use it and judge the build yourself."
3. DESIGN / LAYOUT / UX CRITIQUE
1. Your two best pages are orphans.
bottleneck-audit.html is not in the nav, not in the mobile menu, and not mentioned anywhere on the homepage. Its only inbound link is an underlined text link inside the "01" card on pricing.html. for-coaches-and-therapists.html has zero inbound links from anywhere in the dump. So the page you built to sell the paid entry offer and the page carrying your single strongest differentiator both require the visitor to guess a URL.
Meanwhile the nav gives prime real estate to BLOG and FREE REPORT, which are the two lowest-intent destinations on the site.
Fix: nav becomes ABOUT / THE AUDIT / PRICING / BLOG, with THE AUDIT pointing at bottleneck-audit.html. Drop FREE REPORT from the nav and let it live inside the blog and the quiz thank-you page. Add a homepage section for the Bottleneck Audit (copy in section 5). Link the coaches page from the homepage services block with a plain line: "Coach or licensed therapist? There's a version of this written for you." Add a matching "for agencies" page later, and link both.
2. There are three competing front doors for one free offer.
The nav button says FREE 15-MIN AUDIT and goes to contact.html. The hero button says GET YOUR FREE 15-MIN AI AUDIT and goes to contact.html. Then the readiness section says "Step one of your free audit" and "The free audit begins here" and sends people to quiz.html. Then the about page button says BOOK A 15-MIN CALL and jumps straight off-site to calendly.com, skipping both.
So the site contradicts itself about where the free audit starts, and one page lets people book Annette's calendar with zero intake, which destroys the exact promise the readiness section makes ("I read your answers before we speak. That's what makes a 15-minute call long enough to be useful").
Fix: one door, in one order, everywhere. Every free-audit CTA on every page goes to quiz.html. The quiz's completion screen delivers the score and the Calendly link. contact.html becomes the fallback for people who want to email rather than take a quiz, linked in the footer only. Change the About page button to href="quiz.html" with the same label as everywhere else. Kill the direct Calendly links entirely, because an unqualified booking costs you the scarcest thing you own.
3. The homepage spends three sections on Annette before it says what she'd do for you.
Order today: hero, pullquote about being a business operator, 1995 origin band, "Systems I've already shipped," MIT stats, then finally the services block. Three consecutive credibility blocks, all about the past, before a single sentence about the engagement. A skimmer bounces at the second one, because from her point of view she's reading a biography.
Then the page ends with ten FAQ items across two sections, five of which duplicate the Bottleneck Audit page almost word for word. The last thing anyone sees before the final CTA is a wall of doubt.
Fix, and this is reorder plus deletion, not new build: 1. Hero 2. New Bottleneck Audit block (copy in section 5) 3. Services / what I build, with the before-and-after table 4. Origin band, rewritten as in critique 1, now functioning as "here's why I can do that" instead of "here's my history" 5. "Systems I've already shipped" 6. Stats, rewritten 7. Readiness quiz 8. One FAQ section of six items, keeping "Am I too small," "What if AI isn't the answer," "How much does it cost," "How is this different from a no-code tool or a freelancer," "What happens when it breaks," "How fast, and do you work remotely." Delete the pullquote band and the "too polite to ask" heading, and move the rest of the objections onto the audit page where they belong. 9. Final CTA
Also: the before/after table needs to lose the jargon. "Automated at trigger" and "Zero-gap routing" mean nothing to a therapist. Make it one named scenario with hours attached: BEFORE "New client emails. You answer at 9pm. You send the form. You chase it twice. 40 minutes, every time." AFTER "New client emails. Form goes out in 30 seconds, chases itself twice, lands in your calendar. You look at it once."
4. Conversion plumbing is broken in ways that are cheap to fix and expensive to leave.
- bottleneck-audit.html has no
</head>and no font stylesheet link. Every other page preloads Oswald. If the CSS is inlined per page, verify it, but as written your flagship sales page renders in a different typeface from the rest of the site. On a brand built on Impact and jagged clip-paths, a typography break on the money page reads as amateur. - bottleneck-audit.html's social meta is copy-pasted from pricing.
og:image:alt,twitter:titleandtwitter:descriptionall say "Pricing | Work With Verity Agentic," and there are duplicateog:imageblocks. Share your paid-offer page in a DM and the preview sells the pricing page. - The homepage logo is
href="#". Every other page uses index.html. - Leftover one-page anchors: the hero carries
id="about"and the footer carriesid="resources". Any inbound link to#aboutlands on the hero instead of the About page. - Date wobble. About says "In 1994, I got online for the first time... Within two months, I taught myself to build database-driven websites and launched adoption.com," which puts the launch in 1994, while the h1, the timeline and every other page say 1995. The audit page says "two months after I first got online." Pick one story. The 1995 claim is load-bearing for the entire positioning, so it cannot be approximate.
- About's stats bar says "5 / Countries where I built from nothing" while the prose names Ethiopia, Kenya and Haiti. If the other two are China and the US, say so in the label. An unsupported number next to three named countries invites the reader to start auditing you.
- About's timeline says "2026–" for Verity Agentic, which reads as "hasn't started."
- pricing.html marks up FAQ questions as
<div class="faq-q">while bottleneck-audit.html uses<h3>. Make them all<h3>for outline and screen-reader parity.
4. WHAT'S MISSING
1. A way to give you money.
Every CTA on all five pages, including both buttons that say BOOK THE BOTTLENECK AUDIT and the one that says BOOK THE CALL, resolves to the same free-call page. There is no checkout, no deposit link, no paid-session calendar, no invoice path. A visitor who arrives at bottleneck-audit.html already convinced, which is the entire point of building that page, has literally no way to buy. She gets routed into a free 15-minute call, which means you now have to sell her again, live, and she has to wait days for a calendar slot that you have to give away for free.
What to add: on bottleneck-audit.html, two buttons in the price-line block and the end-cta block. Primary, pink: "BOOK THE AUDIT, $500" pointing at a Stripe payment link that redirects to a scheduling page for the 60-minute session. Secondary, outline: "NOT SURE YET? TAKE THE FREE 10-QUESTION AUDIT" pointing at quiz.html. Then one line of what-happens-next copy under them, which is currently absent from every page on the site:
"Here's what happens after you pay. You'll get the intake questionnaire straight away and a link to book your 60 minutes, usually within the same week. I read your answers first. Your Makeover lands 3 to 5 business days after we talk, with the recorded walkthrough. Nothing else gets sold to you in between."
That last sentence matters more than it looks. Non-technical buyers stall at the paid step because they don't know what they're about to be dragged into.
2. Proof of the thing you actually sell, and a sample of the thing you actually deliver.
You sell intake, follow-up and routing. Your proof is 31,310 SEO keywords, a TESOL course, an orphanage, and 4,000 dogs. All impressive, none of it an intake workflow. There is not one example anywhere on the site of an agentic workflow you have built for anyone, including yourself.
And the $500 product is a written document. You're asking a stranger to buy a document sight unseen from a business with no reviews. Show the document.
Add two artifacts:
- A redacted sample Automation Makeover, as a real page at /sample-automation-makeover.html, not a gated PDF. One anonymized client or your own business. Three to five priorities, sequenced, each with the reason it made the list, exactly as promised. Link it from the deliver-card "03 The Automation Makeover" with the line "Here's a real one, names removed." This single asset will do more for the $500 conversion than any amount of copy, because it converts the buying decision from trust into inspection.
- One workflow teardown of your own intake, on the blog: the quiz-to-email-to-booking system, screenshots, what it decides, where the human review sits. You're already claiming this system exists. Show it running.
Where the two incoming testimonials go, and what each does. Do not put them in a carousel and do not put them side by side in a band. Split them and give each a job:
- Testimonial A, the one that names hours or dollars recovered, goes on bottleneck-audit.html immediately after the deliver-grid and immediately before the price-line. Its job is to answer "is a written plan worth $500" at the exact moment the price appears. Attribution needs role and place: "Sarah K., couples therapist, Longmont CO."
- Testimonial B, the one that speaks to Annette personally, being told the truth, not being sold to, goes on the homepage directly beneath the hero, replacing the pullquote-band. Its job is to replace your own self-quote with someone else's. A founder quoting herself in a big pink quotation mark is the weakest form of proof on the page, and it's currently sitting in the highest-value slot on the site.
3. Risk reversal at the $500 step, and an anchor to price it against.
Right now the answer to "What if I don't like the plan?" is essentially "tough, you got the 60 minutes." That's admirably honest and it's the wrong answer for a business with zero reviews selling to a non-technical buyer's first-ever AI purchase. You don't need a money-back guarantee, which invites tire-kickers. You need a bounded, specific promise you fully control:
What if I don't like the plan? Here's the line I'll hold. The Makeover will name at least three specific things in your business worth fixing, in order, with the reason each one made the list. If it doesn't, tell me and I'll refund the whole fee, no argument and no forms. What I won't promise is that you'll like the answers. Sometimes the fix is a step you've been proud of.
And the $500 has nothing to sit next to. Add one line above the price-figure:
"Five hours a week of work only you can do. Put your own rate on that, then multiply it by fifty weeks. That's the number this is actually priced against."
Same problem one rung up: with the $15,000 tier gone, $8,000 is now the ceiling and reads as the expensive end. Anchor it against a hire, not against nothing: "A part-time ops coordinator costs more than this in the first quarter, and you'd still be the one training her."
5. THE #1 PRIORITY CHANGE
Make the Bottleneck Audit the homepage's offer, and make it buyable.
Two parts, both small, both on the critical path to revenue.
Part one. On bottleneck-audit.html, replace the CTA in the price-line and end-cta blocks with a real primary purchase button plus a secondary free-call fallback, as specified in section 4.1, wired to a Stripe payment link. Today there are three buttons on that page reading BOOK THE BOTTLENECK AUDIT and all three take the money off the table and put a free call in its place.
Part two. Insert a Bottleneck Audit section on the homepage as section two, directly under the hero and in place of the self-quote pullquote band. Copy:
From $500, credited toward a build
Before you automate anything, find out what's actually in your way.
Most of what eats your week isn't one big broken thing. It's four small handoffs that only work because you remember them, and you can't see them clearly from inside the job.
The Bottleneck Audit is sixty minutes on how work really moves through your business, then a written plan called the Automation Makeover: the three to five things worth fixing, in the order worth fixing them, with the reason each one made the list. It lands in 3 to 5 business days with a recorded walkthrough. If you go on to have any of it built, the fee comes off the price.
You can take it and hand it to someone else. Plenty of people should.
[BOOK THE AUDIT, $500 →] [See what the Makeover looks like]
Why this beats everything else on my list. The homepage currently has one thing to buy and it's free. Five pages, eleven buttons, and every single one leads to a fifteen-minute call on Annette's calendar. That's a funnel with no cash register in it and a solo operator's diary as the only bottleneck, which is a striking thing to build for a business whose entire proposition is finding bottlenecks.
Everything else I've flagged is optimization. The negation tic, the enterprise stats, the three-doors problem, the broken meta tags: all real, all worth fixing, none of it matters if a convinced buyer still can't pay. And there's a compounding effect. The moment the paid audit is purchasable from the homepage, the qualification gate you know you're missing on the upper tiers partly builds itself, because the people who pay $500 to be diagnosed are the people worth quoting $8,000 to, and the ones who only ever want the free call sort themselves out without costing you an hour each.
You removed the $15,000 tier to focus on small projects. Right now the site doesn't sell any project at all. It sells a phone call, four times a page.
google/gemini-3.1-pro-preview top abstract + scientific reasoning
Here is your specific, blunt critique of the Verity Agentic pages. We are looking strictly at how a non-technical SMB operator moves from initial pain to opening their wallet, and right now, friction in your own copy and routing is getting in their way.
1. WHAT'S WORKING
- The compliance distinction on the Therapists page. Defining a strict boundary with "What I will not build" is brilliant. It establishes immediate, unshakeable authority. A licensed therapist instantly knows you have read their rulebook.
- Anticipating failed alternatives. On the Bottleneck Audit page, specifically calling out "The no-code tool," "The freelancer," and "The ChatGPT subscription" directly answers the buyer's internal monologue. You prove you know what they already tried.
- Reframing limited capacity. Stating "I'm solo by design" on the Pricing page turns a potential weakness into a quality guarantee. It frames your sole attention as a feature.
- The absolute truth in your FAQ. "Automating a broken process only makes it break faster." This is a deeply authoritative reality check that proves you actually build systems in the real world.
2. COPY CRITIQUE
Critique 1: The completely mismatched enterprise stat * The quote: "95% of enterprise AI pilots failed to deliver measurable P&L impact" (Homepage) * Why it fails: You sell specifically to 1-to-10-person businesses. Solo coaches and boutique agencies ignore paragraphs containing the words "enterprise" and "P&L impact." It alienates the exact buyer you want to convert. * Replacement copy: "95% of AI initiatives fail because owners buy a tool before they map their process. We fix the process first."
Critique 2: The weirdly robotic hero eyebrow * The quote: "Agentic AI consulting by Annette Thompson, AI systems consultant" (Homepage) * Why it fails: It reads like raw SEO keyword stuffing drafted by an algorithm. It completely undermines the veteran human authority established everywhere else on the site. * Replacement copy: "Internet builder since 1995. AI systems for operators."
Critique 3: Clunky delivery mechanics * The quote: "Before we speak, so the 60 minutes goes on solving rather than on me asking what you do." (Bottleneck Audit page) * Why it fails: The phrase "goes on solving" is awkward syntax that makes the reader stumble. Keep the mechanical steps painfully simple. * Replacement copy: "You fill this out before we speak. We spend our 60 minutes fixing the problem, not covering basic background on what your business does."
Critique 4: The passive hero copy * The quote: "Stop doing work / Your AI system should be doing." (Homepage) * Why it fails: Your buyer doesn't have an AI system yet, so telling them their system "should be doing" it creates a logical disconnect. The phrasing feels passive. * Replacement copy: "Stop doing the manual work / A basic AI agent can handle."
3. DESIGN / LAYOUT / UX CRITIQUE
Element: The competing main-nav CTAs * The flaw: You place a "FREE REPORT" text link right next to the "FREE 15-MIN AUDIT" button in the primary header. You give an overwhelmed, non-technical buyer two different free actions to evaluate. This causes hesitation. * The fix: Remove the "Free Report" from the global utility navigation completely. Keep your main navigation aggressively focused on the single designated entry point: the audit.
Element: The Before/After table (.ba-table)
* The flaw: On mobile viewports, multi-column horizontal layouts with central dividing arrows (.ba-divider) break comprehension. The text squishes into vertical pillars that are difficult to read side-by-side.
* The fix: Use a media query to stack the Before
openai/gpt-5.6-sol OpenAI current flagship voice
1. WHAT'S WORKING
-
The homepage hero names recognizable work instead of selling abstract AI.
“Client intake, follow-ups, routing, the busywork that eats your week” gives a non-technical buyer immediate handles. That’s much stronger than leading with agents, orchestration, or model names. -
The free audit has a credible mechanism.
The homepage section “Start with 10 questions, then we talk” explains why a 15-minute call can still be useful: “I read your answers before we speak.” That reduces fear of wasting time on a generic discovery call. -
The Bottleneck Audit makes the deliverable unusually tangible.
The five-part deliverables grid is the strongest sales material on the site. “Three to five priorities, sequenced, each with the reason it’s on the list” tells buyers what will actually arrive. The recorded session and walkthrough also reduce the risk of paying $500 for a thin PDF. -
The maintenance objection is answered honestly.
On the homepage, “It will need attention eventually, and anyone who says otherwise is selling” sounds like an experienced operator. It also creates a legitimate reason for the retainer without pretending automations run forever untouched. -
Annette’s proof is specific and connected to operating constraints.
“31,310 keywords ranked, $0 on ads,” “7 countries,” and “4,000 rescue dogs” are memorable. More importantly, the Bottleneck Audit page explains the common thread: building without a budget, team, or playbook. That connection makes the history relevant rather than merely impressive.
2. COPY CRITIQUE
1. The homepage eyebrow reads like an SEO field accidentally displayed to humans
Current copy:
“Agentic AI consulting by Annette Thompson, AI systems consultant”
Why it fails:
It repeats two near-identical search terms and leads with language the target buyer probably doesn’t use. It makes the first line feel machine-written, which is especially damaging for an AI company.
Replacement:
AI systems for businesses with 1 to 10 people
Keep the existing headline beneath it. The eyebrow’s job is to identify the audience, not carry keyword density.
2. The homepage pull quote asks buyers to accept an unsupported self-assessment
Current copy:
“I’m not a technologist who learned business. I’m a business operator who mastered technology. There’s a difference: and it shows in every system I build.”
Why it fails:
“There’s a difference” says nothing, the colon is awkward, and “it shows in every system I build” is self-congratulation before client proof appears. Annette’s history is much stronger than this claim.
Replacement:
“I’ve spent 30 years building systems where there wasn’t a clean process, a large team, or room for failure. That’s why I start with how the work moves, not with the latest AI tool.”
That expresses the distinction through method and evidence.
3. The Bottleneck Audit’s risk objection makes the purchase feel riskier
Current copy:
“You’ll have had the 60 minutes and the written findings either way, and $500 is the cheapest way to find out we’re not a fit.”
Why it fails:
The buyer asked what happens if the plan disappoints them. The answer amounts to “you still lose $500.” Calling that cheap doesn’t reduce the risk. It also frames the audit as a compatibility test rather than a useful standalone product.
Replacement:
What if the plan doesn’t feel useful?
Tell me within seven days of receiving it. I’ll revise it once at no charge so the priorities, reasoning, and next steps are clear enough for you to act on. You’ll still own the recording and final plan, whether you hire me to build anything or not.
Only publish this if Annette will honor it. Some defined remedy is needed here.
4. The coaches and therapists page directly contradicts the main pricing page
Current copy on the niche page:
“Done-For-You Workflow, From $1,500”
“AI Ops Retainer, From $500/month”
Current copy on the pricing page:
“Done-For-You Agentic Workflow, $3,000 to $8,000 typical”
“AI Ops Retainer, $1,500 to $3,000/month”
Why it fails:
A buyer who compares pages will reasonably wonder which prices are real. It looks like either hidden discounting or stale pricing. Both damage trust, particularly when the site repeatedly promises “No mystery pricing.”
Replacement on the coaches and therapists page:
Done-For-You Agentic Workflow
$3,000 to $8,000 typical
One or two priority workflows, such as intake, scheduling, follow-up, or lead routing, built, tested, documented, and running in 1 to 2 weeks.AI Ops Retainer
$1,500 to $3,000 per month
Ongoing monitoring, maintenance, and agreed improvements for systems I’ve built. You’ll know what I’m watching, what’s included, and how changes are approved.
3. DESIGN / LAYOUT / UX CRITIQUE
1. The homepage buries the actual buying path under too much repeated persuasion
The homepage contains the hero, founder quote, origin story, proof, industry statistics, services, offer panel, readiness section, two separate FAQ sections, and a final CTA. Several sections repeat the same claims about intake, follow-ups, routing, and systems running without the owner.
Conversion cost:
The page keeps explaining why automation matters without clearly showing what Annette sells next. The paid Bottleneck Audit, which should bridge interest and a build, barely appears outside the pricing FAQ.
Specific fix:
Shorten the sequence to:
- Hero
- Buyer symptoms
- Proof
- Three-step buying path
- What gets built
- Objections
- CTA
Replace the enterprise statistics section with a three-step offer path:
Start small. Go further only if the numbers make sense.
1. Free 15-minute audit
Answer 10 questions first. I’ll identify the most likely bottleneck and tell you whether it’s worth investigating.2. Bottleneck Audit, from $500
A 60-minute working session and a written Automation Makeover delivered in 3 to 5 business days. Your fee is credited toward a build.3. Done-for-you workflow, $3,000 to $8,000 typical
I build, test, document, and deploy the priority workflow in 1 to 2 weeks.
Give steps two and three direct links to their relevant pages.
2. Bottleneck Audit buttons promise a purchase action but lead to a generic contact path
Element:
“BOOK THE BOTTLENECK AUDIT” links to contact.html, while the page later says the first step is a free 15-minute call.
Conversion cost:
“Book the audit” implies the visitor can select and buy the paid service. If the destination instead asks them to apply for a free call, the click feels misleading. Uncertainty at the commitment point creates abandonment.
Specific fix:
Until direct audit checkout exists, change every button on this page to:
START WITH THE FREE 15-MINUTE CALL
Add microcopy directly below it:
You won’t be charged. If the Bottleneck Audit fits, I’ll explain the scope and exact price on the call.
3. The pricing cards show products but don’t help buyers choose among them
The pricing page presents three equal cards with short descriptions. There’s no “choose this when” guidance, no scope variables, and no CTA inside the paid cards.
Conversion cost:
A non-technical buyer can’t reliably distinguish when a $500 plan is sufficient, when a $3,000 build becomes necessary, or why a build might cost $8,000. Equal visual weight makes the ladder look like three unrelated services.
Specific fix:
Add these fields to each card:
- Best when
- What you leave with
- Timing
- Next step
Make each full card clickable. Label the Bottleneck Audit “Recommended first paid step.” Keep the free call visually separate above it, as it currently is.
4. Proof needs to sit beside the claims it’s supposed to validate
When the two approved testimonials arrive, don’t put both in a generic testimonial band near the footer.
Specific placement:
- Put the stronger implementation or time-saved testimonial on the homepage immediately after “Systems I’ve already shipped.” It should verify that Annette can understand a messy business and produce something usable.
- Put the testimonial that best demonstrates clarity, diagnosis, or trust immediately after the Bottleneck Audit deliverables grid and before “From $500.” It should reduce concern about paying for advice before paying for a build.
Each should include the client’s name, role, business type, what Annette handled, and a concrete result or observed change. A quote saying Annette was “great to work with” won’t carry enough weight.
4. WHAT'S MISSING
1. Buyers can’t see what an Automation Makeover looks like
The audit page describes the document well, but a $500 planning deliverable still feels abstract without a visual sample.
Add a redacted two-page example or a realistic annotated mockup showing:
- Current workflow
- Bottleneck
- Recommended change
- Why it ranks first
- Tools involved
- Human review point
- Estimated implementation range
- Expected hours or errors reduced
- What not to automate
Show it directly after the deliverables grid with:
See what you’ll receive
This sample shows the level of detail in an Automation Makeover. Your plan will use your workflows, tools, constraints, and priorities.
2. “From” pricing has no visible pricing logic
The site discloses ranges, which is good, but it doesn’t explain what changes the price. A non-technical buyer may assume the final number is arbitrary.
Add a section to the pricing page:
What changes the price?
Your quote depends on four things:
- How many tools the workflow needs to connect
- How many decisions the system has to make
- Whether sensitive data or human approval is involved
- How much cleanup, testing, and documentation the existing process requiresYou’ll get a fixed scope and price before work starts. I won’t begin a build on an open-ended hourly estimate.
Also explain why the Bottleneck Audit itself starts at $500. If there are genuine higher-scope audit versions, name the variables. If there aren’t, use a fixed $500 price.
3. The site doesn’t explain data access, ownership, or offboarding
This matters to every non-technical buyer and becomes critical on the therapist page. “Human-in-the-loop” and “governance” don’t answer the practical questions.
Add a concise “Your systems and data” section to the pricing and Bottleneck Audit pages:
You keep control of your systems
Wherever the platform allows it, your accounts, credentials, data, and automations stay in your business’s name. I document what I build and tell you which parts require maintenance. I won’t put sensitive client data into an AI model without discussing the data path, access rules, and safer alternatives with you first.If we stop working together, you’ll receive the current documentation, access list, and handoff notes included in your scope.
For therapists, name the limits more precisely. “Compliance-aware” is safer than claiming compliance, but buyers still need to know how vendors, BAAs, EHR boundaries, and protected information are handled.
5. THE #1 PRIORITY CHANGE
Fix the pricing contradiction on the coaches and therapists page immediately.
Change:
- “Done-For-You Workflow, From $1,500” to “Done-For-You Agentic Workflow, $3,000 to $8,000 typical”
- “AI Ops Retainer, From $500/month” to “AI Ops Retainer, $1,500 to $3,000 per month”
Then search the entire static site for every price and standardize all visible copy, metadata, structured data, and offer descriptions against one source document.
This beats every other change because it’s an active trust failure at the exact point where buyers evaluate affordability. The site says “Pricing, plainly stated” and “No mystery pricing,” then gives materially different numbers for the same offers. Better proof, tighter pages, and stronger CTAs can’t compensate for making buyers wonder whether the published price depends on which page they happened to open.
x-ai/grok-4.5 blunt, low-hedge independent critic
1. WHAT'S WORKING
Bottleneck Audit pain open is the best writing on the site. The sequence "You read the intake form yourself… There's a follow-up you meant to send on Tuesday. It's Friday… A client slipped last month because the handoff lived in your head" is concrete, ownable, and sounds like the buyer's Tuesday. This is the voice the rest of the site should match more often.
"You've probably tried these already" (no-code / freelancer / ChatGPT) on the Bottleneck Audit page does real sales work. It names the exact debris this buyer already owns, and the closer ("you bought a tool where you needed a system") is the sharpest single line on any page. It pre-handles the objection without asking permission.
Objection FAQs, especially "And the ones you're too polite to ask" on the homepage. "Then I'll tell you that on the free call, before you've spent anything" and "I'd rather send you away with that than sell you a system you don't need" build trust faster than any credential block. Same register on Bottleneck Audit for "Why pay for this if the first call is free?" and "Do I get the automations built as part of this?" Blunt scope is a feature.
Therapist vs coach track split on /for-coaches-and-therapists.html. "What I will not build" (no automated review requests tied to sessions, no outcome-guarantee copy, clinical data stays in the EHR) is rare and commercially smart. Non-technical licensed buyers need that more than another Oswald headline.
Proof cards with case-study links (31,310 / E-Teach Abroad) plus "I built and ran both of these myself, start to finish. Real systems that did the work, not slide decks about it." Credentials are earned, not inflated. The structure is right even if the translation to buyer outcomes is still weak (see below).
2. COPY CRITIQUE
1. Homepage hero note: "Leader: Annette Thompson"
Fails because "Leader" is empty corporate waste beside a pink audit CTA. A coach scanning the hero gets no job title, no reason to care, and a word that sounds like a badge generator. It undermines the pullquote two scrolls down.
Replacement:
Built by Annette Thompson, systems operator since 1995
2. Homepage stats setup: "Why most AI fails for small businesses (and how we fix it)" + "95% of enterprise AI pilots failed…"
Fails for this buyer. You promise small business, then cite MIT/Fortune enterprise pilots. A solo therapist does not live in P&L pilot land. The 95% figure either scares her off ("so it fails") or feels like someone else's war. The EY card about "workers at companies with a clear AI strategy" is the same miss: she does not have "workers" or an "AI strategy."
Replacement (section heading + one card, drop the corporate framing):
Heading: Why the tools you already bought didn't stick
Card 1 number: Still you
Card 1 desc: Most small practices don't fail at AI because the model is weak. They fail because the process still runs through one person, and the tool only made the busywork faster.
Card 1 source: What I see in almost every 15-minute audit
Card 2 number: Hours, not decks
Card 2 desc: The operators who get time back start with one repeated workflow (intake, follow-up, routing), map it, then automate it. Strategy without a shipped workflow is just reading.
3. Homepage services blurb: "AI strategy, workflow design, and practical agent builds for teams ready to stop duct-taping their operations together. As an AI systems consultant, Annette brings 30 years of systems thinking to every agentic AI engagement."
Fails hard. Third-person resume voice after a first-person site. "Agentic AI engagement" is vendor-speak your coach buyer does not use. "Teams ready to…" flatters readiness she does not feel.SEO stuffing is audible.
Replacement:
I map how work actually moves through your practice, then build the parts that shouldn't need you: intake, follow-ups, routing. Strategy only when it changes what we ship. You've been duct-taping this long enough.
4. About page line: "From Africa to the AI frontier."
Fails because "the AI frontier" is the exact marketing-influencer register you said you refuse. It also restarts the Ethiopia story a third time on a page that already told it in the charcoal band and will tell it again in the timeline. The buyer who opened About wants one clean thread: why you, why this work, then a CTA. She gets a loop.
Replacement (section heading + one pivot sentence):
Heading: Same conditions. Different decade.
Lead: No budget, no team, no playbook: that was Ethiopia, Kenya, and Haiti. It's also what most 1-to-10 person businesses look like on the inside. I still start the same way. Write down how the work moves. Take one person out of the middle where she doesn't need to be.
3. DESIGN / LAYOUT / UX CRITIQUE
1. Primary paid offer is orphaned from nav and homepage structure.
bottleneck-audit.html is the actual sales page for the $500 entry rung. Desktop nav is ABOUT / BLOG / PRICING / FREE REPORT. Homepage CTAs all go to contact.html for the free audit. The Bottleneck Audit only surfaces as a text link inside a pricing card. Cost: anyone persuaded by the homepage never lands on the page that sells the paid diagnostic; free-call inventory fills up with tire-kickers who never saw the makeover, deliverables, or "credited toward a build" argument in full.
Fix: Add BOTTLENECK AUDIT (or THE AUDIT) to primary nav after PRICING. On homepage, after the before/after table or proof section, add one full-width band that links to bottleneck-audit.html with its own pink CTA: "See the Bottleneck Audit." Keep header free-audit CTA; give the paid offer a named door.
2. Empty feature icons on homepage (.feature-icon spans with nothing in them).
Five features in a grid with blank icon slots look unfinished next to a noire/Impact identity that otherwise commits hard. Non-technical buyers read "half-built" as "half-serious," especially on an AI site.
Fix: Either delete the icon spans and let bold title + description breathe, or use hard geometric CSS shapes (corners, yen-like marks, simple yellow/pink rules) that match the jagged clip-path language. No Lucide-soft icon pack. Brand stays sharp; unfinished chrome goes.
3. Coaches/therapists pricing grid undercuts the main ladder.
On /for-coaches-and-therapists.html: Done-For-You "From $1,500", Retainer "From $500/month." On /pricing.html: DFY "$3,000 to $8,000 typical", Retainer "$1,500 to $3,000/month." Same buyer can open both tabs in one sitting. Cost: trust collapse, and the vertical page trains them to expect a price you do not actually offer.
Fix: Make numbers identical to the main ladder, then localize the description, not the price. Example DFY line: "$3,000 to $8,000 typical. One or two practice workflows (intake, scheduling, lead funnel) built and running in weeks. Therapist builds stay inside compliance guardrails from day one." If you truly sell a lighter coach package, name it as a separate rung with scope boundaries. Do not silently widen the range by vertical.
4. Hero proof density vs buyer translation (homepage proof + About pile-on).
31,310 and "7 countries" sit in big Impact numbers with no bridge to intake/follow-up/routing. Visually they dominate; cognitively they read as SEO and nonprofit ops. Beside a hero that promises busywork relief, the layout sells "impressive biography" over "she fixes my week." About then spends a charcoal section, an intro column, a stats bar, a story grid, and a long timeline retelling the same arcs.
Fix (homepage proof cards): Keep the numbers, rewrite the H3 and body so each ends on a buyer-shaped outcome. Example for Bone Voyage: H3 "A content system that ran without a marketing hire" + close the card with "Same pattern I use for client intake and follow-up: define the steps once, let the system repeat them." For E-Teach: close with "Certificates, recruiting, and delivery without you chasing every handoff." On About, cut one full retelling (the story-grid or the mid-page intro loop); keep Ethiopia once, hard, in the charcoal band, and let the timeline be the archive.
4. WHAT'S MISSING
1. A visible sample of the Automation Makeover.
You sell a $500 written plan in 3 to 5 days and never show the object. Non-technical buyers cannot picture "three to five priorities, sequenced, each with the reason." Without a redacted one-pager, a fake-but-labeled sample outline, or even a fixed section structure ("Priority / Hours recovered / Why first / Build vs DIY / Fragility notes"), the Bottleneck Audit stays abstract next to Zapier monthly pricing they already understand. Put a "What the Makeover looks like" band on bottleneck-audit.html before the price line: five labeled blocks matching deliverable 03, with placeholder practice content (e.g. coach intake). This is the single best trust object you do not have.
Where testimonials should go when they land tomorrow (not listing the gap, only placement):
- One short quote (outcome + role, e.g. "coach, solo practice") directly under the Bottleneck Audit price line, job: justify the $500.
- One on homepage between proof cards and services, job: bridge "her past systems" → "my kind of business."
- Do not park them only on About.
2. One ordered front door. You currently run four.
Free 15-min audit (contact.html), quiz readiness (quiz.html, homepage readiness band), free report (nav), and Bottleneck Audit CTAs that also hit contact.html. About sometimes jumps straight to Calendly. After someone books, nothing on-page says what the next email contains, whether they get the 10-question form first, or how free call differs from paid audit in practice (you explain it in FAQ only).
Buyer sequence should be explicit in one strip used sitewide:
Free 15-min call → (fit) Bottleneck Audit → build → retainer.
Quiz and free report are top-of-funnel content; they should not wear the same pink button weight as the audit. Contact page (not in the dump, but implied) needs three sentences: what to expect on the call, what to prepare, what happens if it is not a fit.
3. Risk shape after "I don't like the plan."
Current answer is honest and also conversion-hostile: "You'll have had the 60 minutes and the written findings either way, and $500 is the cheapest way to find out we're not a fit." That is not risk reversal; it is risk acknowledgment. For a non-technical solo buying diagnosis from a stranger, add one concrete, cheap-to-honor commitment on the Bottleneck page near the price: e.g. "If the Makeover does not name at least three specific, ordered fixes for your operation, I'll refund the fee. You keep the recording and the doc." You already say you would rather send people away. Productize that. Fee-credited-if-you-continue only helps people who already trust you enough to continue.
5. THE #1 PRIORITY CHANGE
Add a tangible "What the Automation Makeover looks like" band on bottleneck-audit.html (and link that page from nav + homepage), with this copy and structure.
Place it after the 01–05 deliver grid, before the price line.
Section eyebrow: What you walk away with
H2: The Automation Makeover, on the page.
Intro: Not a vibe check. A short written plan you can hand to me, a freelancer, or your future self. Here is the shape every Makeover uses.
Then five static cards (paper-band or jagged frames so brand stays intact):
-
Bottleneck named in one sentence
Example line on the card: "New-client intake still needs you to read every form before anyone else can act." -
Priorities 1 to 3 (or 5), in order
Example: "1. Intake routing. 2. No-show reminders. 3. Quiet-client follow-up." -
Why this order
Example: "Intake first, because every later step waits on a clean handoff." -
Build vs DIY for each
Example: "Reminders: DIY this week. Intake agent: build. Follow-up sequence: build or Playbook." -
Fragility and watchouts
Example: "Safe to leave alone for a year: reminder rules. Watch when the model or form tool changes: intake classifier."
Close under the cards:
Your version uses your workflows, your tools, and your actual volume. Findings in 3 to 5 business days, plus a recorded walkthrough. Fee from $500, credited if you move to a build.
CTA: BOOK THE BOTTLENECK AUDIT (still can route through free call first if that is policy, but label the destination clearly on the contact step).
Why this beats everything else on the list.
Your objection handling, pain copy, and ladder are already ahead of most solo AI consultancies. What still blocks a non-technical coach or therapist from paying $500 is not brand, not palette, and not another biography pass. It is buying an invisible document from someone whose best proof still sounds like SEO and rescue logistics. Show the object and the free-call → paid-audit path stops feeling like a pitch funnel and starts feeling like purchasing a defined work product. Nav link + homepage band to this page are load-bearing only because this band makes the page worth routing to. Fix icons, fix coaches pricing, cut About repetition next. Ship the Makeover sample first.
z-ai/glm-5.2 distinct-lab reasoner, real extra coverage
1. WHAT'S WORKING
The "You've probably tried these already" section on the Bottleneck Audit page. Three cards: "The no-code tool," "The freelancer," "The ChatGPT subscription." Each one names the exact thing the buyer has already tried and explains precisely why it failed, without making the buyer feel stupid. "You bought a tool where you needed a system" is the kind of line that makes someone nod and keep reading. This is the strongest sales argument on the entire site.
The before/after table on the homepage. Four rows, two columns, arrows between them. "You doing the work" → "System doing the work." It communicates the entire value proposition in under three seconds. The visual does the heavy lifting that paragraphs of copy would fumble.
The compliance track split on the for-coaches-and-therapists page. Two columns: "For Coaches / Market Freely. Book More Clients." and "For Therapists / Compliance-Aware, By Design." The therapist column includes a "What I will not build" list naming specific board-complaint risks. That's a genuine differentiator no generalist competitor would think to include, and it tells a therapist buyer that Annette has actually read the rules.
The "What happens when it breaks" FAQ answer. "It will need attention eventually, and anyone who says otherwise is selling." This answer does three things at once: it acknowledges the real risk, it positions competitors as dishonest, and it sets up the retainer as the logical next step. That's a single paragraph doing more work than most sales pages.
The proof section with specific numbers. "31,310 keywords" and "7 countries" with links to case studies. The numbers are specific enough to be credible and odd enough to be memorable. No round numbers, no "up to" hedging.
2. COPY CRITIQUE
"We don't sell AI tools: we sell results. AI workflow automation that actually runs." (Homepage, services-callout)
This fails three ways. First, the pronoun is wrong: Annette is solo, so "we" signals either a team she doesn't have or a corporate register that clashes with the personal brand. Second, "we sell results" is the kind of empty claim a buyer has heard from every vendor they've ever sat through. Third, it's a "this not that" construction dressed up with a colon. The buyer doesn't care what you don't sell. They care what they get.
Replacement:
I design and build the actual system that handles the work. Intake, follow-ups, routing. Running without you.
"No hype, no six-month deployments. Systems that run without you." (Homepage, hero subhead)
"No hype" is itself hype. It's the same energy as a used car dealer saying "no pressure." The buyer's guard goes up. The second sentence is fine, but the first half undermines it by doing the exact thing it claims not to do.
Replacement:
Most builds ship in one to two weeks. Then they run, and you stop being the one who has to remember.
"Thirty years of systems thinking. Applied to AI. Starting now." (Footer, every page)
"Starting now" makes Annette sound like she just discovered AI last Tuesday. It directly contradicts the credibility story the entire site is built on. A buyer who reads "adoption.com, 1995" in the hero and then "Starting now" in the footer has a cognitive dissonance moment at the exact point where they should be clicking a CTA.
Replacement:
Thirty years of building systems that work. Now building them with AI.
"I'm not a technologist who learned business. I'm a business operator who mastered technology. There's a difference: and it shows in every system I build." (Homepage, pullquote band)
This is the "this isn't X, it's Y" construction verbatim, and it occupies a full-width band immediately below the hero, which is premium real estate being spent on a distinction the buyer doesn't care about yet. They haven't seen proof, haven't seen pricing, haven't seen a single example of work. At this point in the page, they want to know what Annette does, not what she isn't.
Replacement:
I learned technology because my businesses needed it. The systems I build solve problems instead of creating new ones, because I've been the person depending on them.
3. DESIGN / LAYOUT / UX CRITIQUE
Pricing inconsistency between the for-coaches page and the pricing page.
The for-coaches-and-therapists page lists: - Done-For-You Workflow: "From $1,500" - AI Ops Retainer: "From $500/month"
The pricing page lists: - Done-For-You Agentic Workflow: "$3,000 to $8,000 typical" - AI Ops Retainer: "$1,500 to $3,000/month"
A coach who lands on the vertical page, sees $1,500, clicks through to pricing, and sees $3,000 to $8,000 will assume one of two things: incompetence or dishonesty. The retainer gap is even worse: $500/month vs $1,500/month is a 3x difference. This single inconsistency will cost more buyers than every copy issue on the site combined.
Fix: Decide on one pricing structure. If the vertical page is showing entry-level pricing for simpler builds, label it explicitly as such ("for practices with a single workflow to automate") and make sure the pricing page references that floor. If it's an error, correct it immediately. Every page on the site must show consistent numbers.
Two back-to-back FAQ sections on the homepage (10 total items).
"Questions people ask before they book" (5 items) immediately followed by "And the ones you're too polite to ask" (5 items). That's 10 FAQ items in a row, which is a wall of text that most visitors will scroll past entirely. The objection FAQs (the second set) are the more valuable ones, but they're buried below the more generic first set.
Fix: Cut the first FAQ set down to 2 items max (cost and timeline). Merge the remaining items into the second set. Or better: kill the first set entirely and rename the second set simply "Questions before you book." The objection answers are doing all the work.
The Bottleneck Audit page funnel is architecturally confused.
The page's primary CTA is "BOOK THE BOTTLENECK AUDIT" (links to contact.html). But the "How it works" section reveals step 1 is "Book a free 15-minute call" and step 2 is "Fill in the questionnaire" (after you book the audit). So someone who clicks "Book the Bottleneck Audit" lands on a contact page that actually books a free call, not the paid audit. The buyer thinks they're buying a $500 audit and instead gets routed to a free call. That disconnect will cause drop-off from buyers who were ready to pay and don't understand why they're being sent backward.
Fix: Either (a) make the page's primary CTA "Book the free 15-minute call" and frame the audit as what comes next, or (b) create a separate contact path for the paid audit that bypasses the free call for buyers who already know they want it. Option (a) is simpler and probably converts better given the audience.
The for-coaches-and-therapists page is orphaned.
There are zero internal links to for-coaches-and-therapists.html from the homepage, the navigation, the pricing page, or the about page. If this page exists for SEO (the title tag targets "Boulder coaches and licensed therapists"), it has no internal link equity flowing to it. If it exists for conversion, no one can find it unless they already know the URL.
Fix: Add a link from the homepage services section or FAQ. Something like: "I also build specifically for coaches and therapists in Boulder, with compliance-aware systems for licensed practitioners." Link the phrase to the vertical page.
4. WHAT'S MISSING
A "what a build actually looks like" page or section.
The Bottleneck Audit page has a detailed "How it works" with 4 numbered steps. But there's nothing equivalent for the Done-For-You build ($3,000 to $8,000). A buyer who gets through the audit and is considering a build has no idea what happens next. How many sessions? Do they review work in progress? What's the deliverable format? How long until it's live? The FAQ says "most first automations ship in weeks" but that's not enough for someone about to spend $3,000 to $8,000. A short "What working together looks like" section on the pricing page, with 3 or 4 steps from kick-off to go-live, would bridge that gap.
The Automation Playbook tier ($1,200 to $1,500) is mentioned in the business context but absent from the pricing page.
The pricing ladder jumps from $500 (audit) to $3,000 (build) with nothing in between. For a coach or solo operator who isn't ready to spend $3,000 but wants more than a written plan, there's no middle rung. Even if the Playbook isn't built yet, a "coming soon" placeholder with a description would capture interest from buyers who need it and would otherwise leave. At minimum, the pricing page should acknowledge that gap exists and tell people what to do if the build is too expensive right now.
Testimonials: where they should go and what each should do.
Three placements, each with a specific job:
-
Homepage, between the proof section and the services section. One testimonial from a relatable buyer (a coach or service provider, not a tech person). Its job: confirm that Annette delivers for someone like the reader. Format: a single sentence quote, name, business type, no star ratings.
-
Bottleneck Audit page, between the deliverables grid and the price line. One testimonial specifically about the audit experience. Its job: prove that $500 for a plan is worth it. The ideal quote names a specific thing the Makeover revealed that the buyer couldn't see themselves.
-
Pricing page, below the ladder and above the FAQ. One testimonial focused on ROI. Its job: make the dollar amounts feel small relative to the value. The ideal quote references hours saved per week or revenue recovered, not just "Annette is great."
5. THE #1 PRIORITY CHANGE
Fix the pricing inconsistency between the for-coaches-and-therapists page and the pricing page. Right now.
This isn't a copy polish or a layout tweak. It's a trust failure that will actively lose buyers who are ready to pay. A coach who finds the vertical page, sees "From $1,500" for a build, clicks through to the pricing page, and sees "$3,000 to $8,000 typical" will close the tab. A therapist who sees "From $500/month" on the vertical page and "$1,500 to $3,000/month" on the pricing page will assume the lower number was bait. No amount of strong copy or good design can recover from a buyer who thinks you're running two different price sheets.
The specific change: pick one set of numbers and use them everywhere. If the for-coaches page is showing entry-level pricing for simpler single-workflow builds, then the pricing page needs to acknowledge that floor. If the pricing page numbers are the real ones, update the for-coaches page immediately. Every page that references a price must show the same numbers in the same format.
This beats every other item on the list because it's an active leak, not a missed opportunity. The copy improvements will lift conversion by single-digit percentages. The pricing discrepancy is losing buyers who were already sold.