Niche Hunter — AI Creative Reverse-Engineering for Primores
Niche Hunter — AI Creative Reverse-Engineering for Primores
TL;DR: Primores needed a glossary/super-niche for its content strategy. Niche Hunter evaluated three candidates and found one clear GO: “AI creative reverse-engineering for performance marketers.” The skill caught two framing errors that would have wasted months — brand-name collision in autocomplete data, and a SERP shape that favors products over articles.
Context
Client/Project: Primores.org (internal) Market: US English Run date: 2026-04-24 Skill version: tools/niche-hunter
Starting point
Primores has two sides:
- Commercial: AI-powered ad creatives, performance marketing, and custom AI flows for eCommerce brands
- Content/demand-gen: The public AI wiki (~88 pages) positioning Primores as practical-AI authority
The question: what glossary/super-niche should the wiki exhaust to build glossary/topical-authority? “AI for business” is too broad. “AI marketing” is a war zone. We needed an Audience × Problem × Context intersection narrow enough to dominate in 3-6 months.
Phase 1 — Candidate Niches
Three candidates based on Primores’ existing IP and buyer profile:
| Candidate | Audience × Problem × Context | Hypothesis |
|---|---|---|
| A: AI creative reverse-engineering | Performance marketers at eCom brands × scaling ad creative × using AI to reverse-engineer winning competitor ads | Maps directly to existing ad-alchemy skill + fitme.lt case study |
| B: AI TikTok/IG distribution | Social media managers at eCom brands × cross-posting content × AI-assisted scheduling | TikTok/IG is where eCom audiences are; natural service extension |
| C: AI for marketing agencies | Agency owners at 5-50 person shops × delivering more output per FTE × AI workflow automation | Primores has internal IP; could productize for other agencies |
Phase 2 — Five-Axis Validation
Candidate A — AI Creative Reverse-Engineering
| Axis | Score | Evidence |
|---|---|---|
| Size | ✅ Pass | 151 unique queries from refined seed ai meta ads (after removing brand-collided pool — see “What the Skill Caught”) |
| Competition | ⚠️ Mixed | Broad framings (ai meta ads) dominated by AdStellar.ai (5-6 of top 10). But narrow framing (ai creative reverse engineering ads): 10 different domains, indie tools, GitHub repos, LinkedIn posts. No authority on the practitioner-workflow angle. |
| Commercial Density | ✅ High | Performance marketers running paid social at eCom brands are Primores’ exact buyer. Query pool has buyer-intent signals (“agency”, “agent”, “for ecommerce”). |
| Expertise Fit | ✅ High | Literally what ad-alchemy does. Real case study (fitme.lt × Tastier). Founder’s ad-tech background (ex-Adform VP Eng). E-E-A-T signals are true, not aspirational. |
| AEO Gap | ✅ Moderate-to-wide | Current content is indie tools and one-off blog posts. No one is publishing systematic workflow content on “how to reverse-engineer a competitor ad with AI as a repeatable process.” |
Verdict: GO — but only under the narrow framing. “AI meta ads” is a trap (AdStellar eating it). “AI creative reverse-engineering as a practitioner workflow” is wide open.
Candidate B — AI TikTok/IG Distribution
| Axis | Score | Evidence |
|---|---|---|
| Size | ❌ Soft-fail | ~35-40 unique candidates. Social-media scheduling is a mature category. |
| Competition | ❌ Product-intent SERPs | ai social media scheduler top 10 is entirely vendor pages: Blaze, Postiz, Sintra, Short.ai, Quso, Ocoya, Buffer, SocialPilot. Every result is “buy this tool”, not “read this article.” |
| Commercial Density | ⚠️ Low-medium | Scheduling-tool buyers aren’t Primores’ buyer. Creator-economy traffic is even less aligned. |
| Expertise Fit | ⚠️ Medium | No scheduling-specific IP in the wiki. |
| AEO Gap | ❌ Narrow | LLMs answer “best AI social media scheduler” confidently with tool lists. The market is settled. |
Verdict: SKIP. Wrong SERP shape. Even if you could rank, traffic wouldn’t convert. Product-intent queries need product pages, not articles.
Candidate C — AI for Marketing Agencies
| Axis | Score | Evidence |
|---|---|---|
| Size | ✅ Pass | 148 queries from ai for marketing agency — clean pool with workflow queries. |
| Competition | ✅ Fragmented | best ai tools for marketing agency: 10 different domains, classic listicle SERP. Room for better-structured content. |
| Commercial Density | ❌ Low | Audience is agencies, not eCom buyers. Traffic here won’t convert on Primores’ current service offering. |
| Expertise Fit | ⚠️ Medium-high | Primores built AI workflows for themselves; wrapping for agencies is plausible but not yet done. |
| AEO Gap | ✅ Moderate | LLMs give generic answers (“use Jasper, Surfer, Zapier”) rather than workflow-specific guidance. |
Verdict: MAYBE — winnable SERP but wrong audience. Only GO if Primores opens an agency-focused service line.
What the Skill Caught
Three framing errors that would have wasted months:
1. Brand-Name Collision
Seed ai ad creative produced 98 candidates — initially looked strong. But reading the pool revealed:
ad creative ai loginad creative ai cancel subscriptionad creative ai pricing
The pool was dominated by queries about the AdCreative.ai product, not the topical category. Any article targeting these would compete with product pages (impossible) and wouldn’t serve Primores’ audience.
Lesson: Don’t trust volume alone. Read the actual candidates.
2. Wrong-Shape SERP
Candidate B’s seed ai social media scheduler looked reasonable (~50 candidates). But the SERP revealed the issue: every top-10 result is a vendor page. Google serves products, not articles, for these queries.
Writing a 2,000-word blog post would never rank — not because competition is too strong, but because the intent mismatch is fundamental. Content strategy isn’t the right move for every fragmented SERP.
Lesson: SERP shape matters more than competition strength.
3. Audience/SERP Divergence
Candidate C (AI for agencies) has a fragmented, winnable SERP. But the audience behind those queries isn’t Primores’ current buyer.
This is the “niche looks clean, commercial density fails” pattern — the axis most often fudged in practice, and the one that sinks content programs that technically rank but don’t convert.
Lesson: Winnable ≠ valuable. Match the audience to your buyer.
Recommendation
Primary niche: AI creative reverse-engineering for performance marketers
This is the clean GO:
- Maps directly to existing IP (ad-alchemy skill, fitme.lt case study)
- Narrow practitioner-workflow framing is wide open
- Audience is exactly Primores’ current buyer
- Topical authority compounds with the ad-alchemy service
Strategic timing: AdStellar.ai is colonizing broad ai meta ads SERPs. There’s a 6-12 month window to own the narrower “reverse-engineering as workflow” framing before they notice and spread into it.
Secondary niche (conditional): AI for marketing agencies
Only if Primores opens an agency service line. Otherwise skip — wrong audience.
Skipped: AI TikTok/IG distribution
Wrong SERP shape, settled AEO, audience misalignment. Don’t revisit unless launching a scheduling product.
Phase 3 — Article Map
Full article map generated for the winning niche:
| Role | Count | Examples |
|---|---|---|
| Pillar | 8 | AI creative reverse-engineering methodology, Meta Ad Library mastery, Visual deconstruction framework, AI UGC ads guide |
| Cluster | 62 | Per-platform guides, tool walkthroughs, localization patterns, creative testing strategies |
| FAQ | 32 | ”Can AI reverse-engineer a competitor’s ad?”, “Is it legal?”, “How long does it take?” |
| Glossary | 16 | Creative formula, focal hierarchy, framing archetype, copy skeleton |
| Total | 118 |
Build Phases
| Phase | Articles | Timeline | Focus |
|---|---|---|---|
| A — Quick wins | 15 | Weeks 1-4 | FAQ + glossary. Fast to rank, create internal-link targets. Includes terminology-owning article for “AI creative reverse-engineering” (zero autocomplete = category-defining opportunity). |
| B — Authority core | 55 | Weeks 5-12 | All 8 pillars + top clusters. After Phase B, Google treats Primores as the voice on this topic. |
| C — Completion | 48 | Months 4+ | Exhaust the niche, build defensive perimeter. |
Key autocomplete insights
- “AI creative reverse engineering” as exact phrase returns zero autocomplete suggestions — the category doesn’t have canonical language yet. Primores can coin and own the terminology.
- Meta Ad Library is an under-rated sub-niche (146 unique queries from one seed). Treated as Pillar 2.
- AI UGC ads is a live sub-niche (65 queries) with clear eCom-buyer intent. Treated as Pillar 4.
Outcome
Phase A Progress (as of 2026-04-24)
15 Phase A articles drafted and ready for review:
- A-01 through A-15 covering glossary terms, FAQ answers, and Meta Ad Library how-tos
- First 15 articles committed to
articles/folder
Next steps
- Publish Phase A articles to wiki
- Begin Phase B pillar development
- Monitor AdStellar content velocity monthly
Key Takeaways
- Narrow framing beats broad framing. “AI meta ads” is being colonized. “AI creative reverse-engineering” is wide open. Same topic, different outcomes.
- Brand collision hides in autocomplete. Volume numbers can lie when a product name overlaps with a category.
- SERP shape > competition strength. Product-intent SERPs can’t be won with articles. Don’t fight the intent.
- Audience alignment is the final gate. Winnable SERPs with wrong audiences still fail.
- Zero autocomplete = terminology opportunity. If no one’s searching for it yet, you get to name it.
Related
- tools/niche-hunter — The skill used
- glossary/super-niche — What makes a niche “super”
- glossary/topical-authority — The strategy this enables
- cases/ad-alchemy-creative-reverse-engineering — The ad-alchemy case study referenced
- experiments/ad-alchemy-competitor-piggyback — Original experiment
Sources
- Niche Hunter run:
runs/primores-20260424.md - Article map:
runs/primores-article-map-20260424.md - Phase A drafts:
runs/drafts/(66 articles)