AI Advertising Accelerator 2026 · Week 4

Deep Research

Find your market before you write a single word. Three hours of structured research replaces six months of writing a book that doesn't sell.

The shift

Most authors spend 6–12 months writing a book and hope people buy it. You're going to spend three hours researching what people are already buying — then write that book.

The difference is hope-based vs. data-based. You're choosing data.

The framework is four phases. One tool per phase. One deliverable at the end: a positioning brief that tells you exactly how to structure your book before you write a word.

The 4-Phase Framework

Phase What it answers Tool Time
1. Discovery What's actually selling in my niche right now? Perplexity Deep Research 30 min
2. Verification Is this data real or hallucinated/stale? Comet (agentic browser) 45 min
3. Pattern decoding What's the formula behind what's working? Claude Code 60 min
4. Live ad intelligence What hooks are converting in paid traffic? Comet + Facebook Ad Library 30 min
Why different tools for each phase

In 2026, prompting one chatbot is no longer the move. Perplexity has live web search and citations. Comet (Perplexity's agentic browser) browses websites as you — past the bot blocks. Claude Code excels at structural pattern-finding across large amounts of text. The Facebook Ad Library is a goldmine of what's converting right now in paid. Each phase needs its own tool.

Worked Example: Gut Health & Digestive Wellness

To show you what good output looks like, here's the framework run live on a real niche — gut health, the #1 selling category in healthcare and wellness publishing right now. Replace the keywords with your niche and the output structure is identical.

1

Discovery

Tool: Perplexity Deep Research  ·  Why: live web access, citations on every claim, will say "I don't know" instead of hallucinating

The prompt

You are a book market analyst. I'm writing a book in the
[YOUR NICHE] space.

Find me the 10 books that have the strongest signals of CURRENT
sales momentum (not all-time bestsellers — books moving NOW in 2026).

For each, return:
- Title + author
- Author credentials (MD/PhD/RD/journalist/none)
- Approximate review count (Amazon + Goodreads)
- Year published
- Subtitle (verbatim)
- The author's backend offer (program/course/coaching) if visible
- Any 2026-specific signal (recent press, podcast tour, new edition)

Hard rule: if you cannot find a citation for a data point,
mark it [UNVERIFIED] — do not estimate.

Output as markdown table sorted by review count descending.

The output (10 candidates)

TitleAuthorCredentialsRatingsYear
Gut: The Inside StoryGiulia EndersMD (in training)56,9492014 (new ed. 2026)
How Not to DieMichael GregerMD51,1102015
Wheat BellyWilliam DavisMD (cardiologist)31,4982011
Grain BrainDavid PerlmutterMD (neurologist)24,8002013
Fiber FueledWill BulsiewiczMD (gastroenterologist)12,6752020
The Mind-Gut ConnectionEmeran MayerMD (UCLA)9,7042015
Brain MakerDavid PerlmutterMD9,1982015
Spoon-FedTim SpectorMD / Professor5,8552020
The Clever Guts DietMichael MosleyMD5,6752017
Super GutWilliam DavisMD3,4822022
2

Verification

Tool: Comet (Perplexity's agentic browser)  ·  Why: Amazon and Goodreads block scrapers. Comet browses as a logged-in user — not a bot.

The prompt

For each book in the table I'm pasting below, open its Amazon page
and pull these data points TODAY:

- Current Best Seller Rank (overall + category, with timestamp)
- Total review count
- Reviews in the last 30 days (look at recent review dates, count them)
- Most recent review date
- Current price (Kindle + paperback)
- Author's website URL

Flag any book where:
- BSR > 100,000 (= barely selling)
- No reviews in last 60 days (= momentum dead)
- Mark these [STALE] in the output.

Output as markdown table. Add a "Momentum Score" column (1-10)
based on review velocity vs. lifetime average.

What you're looking for

Strong current momentum: books with BSR under 50K, fresh reviews in the last 30 days, an active author podcast or new edition, a visible backend offer.

Stable but not breakout: books with steady review velocity but no new media wave.

[STALE] flag: books that show on bestseller listicles but were published 5+ years ago with no new reviews this quarter. Selling once does not mean selling now.

The lesson

Don't trust the first AI list. Half of "top books in 2026" articles include books published in 2014 with stale review streams. Phase 2 is what separates the books that are selling from the books that used to sell.

3

Pattern Decoding

Tool: Claude Code (with web access)  ·  Why: best at structural pattern-finding. Feed it 10 book descriptions, subtitles, and cover lines — it finds the formula in minutes.

The prompt

You are a positioning strategist. I'm pasting in 10 verified
bestselling book listings from the [YOUR NICHE] niche, including:
- Title
- Subtitle
- Author credentials
- Cover description
- Amazon book description (first 200 words)
- Backend offer (if known)

Find the patterns. Specifically:

1. TITLE STRUCTURE — what formats appear repeatedly?
2. SUBTITLE FORMULA — what's the dominant promise structure?
3. THE ONE PROMISE — if each book had to fit on a billboard, what's the single line?
4. AUTHORITY SIGNALS — how is credibility established?
5. BACKEND OFFER STRUCTURE — what's the value ladder behind the book?
6. COMMON METAPHORS — what mental models do they use?

Then identify GAPS:
- What promise is overused (= crowded)?
- What angle is missing (= opportunity)?
- What credentials combo is rare in the top 10 (= differentiator)?

Output as a positioning brief, not a list of facts.

The output — what we found in gut health

Title structure (3 dominant patterns):

PatternExamples
Single-word visceral nounGut, Wheat Belly, Super Gut
Anti-establishmentThe Diet Myth, The Plant Paradox, Missing Microbes
Clinical authorityThe Mind-Gut Connection, The Microbiome Solution

Subtitle formula — the dominant pattern is the Triple-Promise List:

[Action verb] [Outcome 1], [Outcome 2], and [Outcome 3]

  • "Lose Weight, Restore Your Health, and Optimize Your Microbiome" — Fiber Fueled
  • "Lose Weight, Heal Your Gut, and Live Lectin-Free" — Plant Paradox Cookbook
  • "Taking Control of Your Weight, Your Mood, and Your Long-term Health" — The Good Gut
  • "Reprogram Your Microbiome, Restore Health, and Lose Weight" — Super Gut

The triple-promise is so dominant in this niche it's almost mandatory. Always: weight + something internal (gut, mood, energy) + something aspirational (longevity, freedom).

The one promise (billboard test): Every winning gut book reduces to "Fix your gut → fix everything." The "everything" varies — weight, mood, fatigue, brain fog, autoimmune — but the architecture is identical.

Authority signals: 8 of the 10 top books are authored by an MD, PhD, or Professor. The two exceptions are scientists writing as journalists. You need credentials or a credentialed co-author. No exceptions in the top 10.

Backend offer structure — the value ladder:

Every successful author in this space runs the same ladder:

Book ($15–30)
Online course / masterclass ($300–2,000)
Coaching or membership ($2K–10K, or $30–100/mo)
Often: supplement line
+ podcast as content engine

Examples in this niche:

  • Will BulsiewiczPlant Fed Gut Masterclass (7-week certified program) + Microbiome Method Coaching + podcast
  • Steven Gundry — Gundry MD supplement empire + Gundry Health telemedicine app + Dr. Gundry Podcast
  • Megan Rossi — Bloating Masterclass + London clinic + product line
The most important insight

The book is the lead magnet. The money is downstream. If you write a book without designing the value ladder behind it, you'll leave 95% of the revenue on the table.

Gap analysis — where the opportunity is

Crowded angles (avoid head-to-head):

  • Plant-based / fiber-first (Bulsiewicz owns this)
  • Anti-lectin / restriction-based (Gundry owns this)
  • Bloating-specific consumer book (multiple players)

Missing in the top 10 (the open lanes):

  • A practicing clinician writing from inside an active practice — not a bestselling-author-side-hustle
  • A book positioned for healthcare professionals who treat gut patients (B2B angle)
  • Functional medicine integrated with conventional GI — the bridge book
  • The concierge-medicine angle: "what your specialist won't tell you"

If your real-world story matches one of these missing angles, you have an opening that the current top 10 cannot reach.

4

Live Ad Intelligence

Tool: Comet + Facebook Ad Library  ·  Why: longevity = conversion. Ads running 90+ days are working. Ads running 5 days are tests.

The prompt

Open the Facebook Ad Library at facebook.com/ads/library.
Search keyword: [niche keyword — e.g., "gut health", "bloating"]
Filter: Active ads only.

For each advertiser running 5+ ads OR ads that have been
running 90+ days:
- Take a screenshot of the longest-running ad
- Note: advertiser, hook (first 5 words), ad format
  (image / VSL / UGC), CTA

Bring me the top 10. I'm looking for what's CONVERTING
(longevity = converting in paid traffic).

What we found in gut health

Hook patterns running 90+ days:

  • "NEVER eat these [3/5/7] foods" — running for 5+ years across multiple variations
  • "Why your bloating is actually [X]" — symptom into reframe
  • "If you're over 50 and still tired, watch this" — age plus symptom
  • "I'm a doctor. I'd never eat this." — authority plus contrarian

Format patterns:

  • Long-form Video Sales Letters (VSLs), 30–60 minutes — the dominant format. Once a VSL converts, it can run for years.
  • UGC testimonial reels (real customers, phone-shot) outperform polished ads
  • Doctor-on-camera explainer (15–90 seconds) used as top-of-funnel before VSL
Compliance note

Facebook allows symptom language ("bloating", "fatigue", "brain fog") but blocks condition language ("IBS", "Crohn's", "leaky gut"). Top-performing ads stay symptom-side. Know this before you spend a dollar.

The replay-ability rule: A working hook in this niche can run for years with a strong authority figure on camera. Don't reinvent — refine.

Synthesis — What you now know

  1. The triple-promise subtitle is the proven formula — weight + internal benefit + aspirational outcome. Don't reinvent the structure.
  2. You need MD / PhD / RD credentials — or a credentialed co-author. Eight of the top 10 follow this rule.
  3. The book is a lead magnet, not a product. Design the value ladder (course → coaching → membership → supplements) before you write Chapter 1.
  4. Lead with the symptom, not the condition. Most readers identify with "bloating" or "fatigue" before they identify with anything clinical. Marketing follows the symptom.
  5. The open lanes are the practitioner-led angles — clinic-backed, integrative, professional-to-professional, or concierge-style. None of those are in the current top 10.