SEO and AI Search Optimization
for Brands and Founders
I'm Khadija Zaman. I help brands and founders rank on Google and earn AI visibility in ChatGPT, Gemini, Perplexity, and AI Overviews. As AI Search Manager at Wellows, with 8 years in search across agency and in-house roles, I automate the work so visibility compounds instead of resetting every month.
- ChatGPT
- Gemini
- Perplexity
- AI Overviews
Figures updated 9 Sep 2026 · GA4 and Search Console
Method I ran a permutation null against my own published citation study and revised two headline findings that failed it. How I tried to break my own study →
Featured in Search Engine Land. Its guide to topic clusters for AI search runs on the Query Fan-Out Generator I built at Wellows. The article credits Wellows; the COO’s post on the experiment behind it names me.
Podcast The Growth Show by PipeRocket Digital: “How to Make Your B2B Brand Citable by AI, Not Just Rankable on Google”, 25 June 2026.
- ChatGPT
- Gemini
- Perplexity
- AI Overviews
How do I structure topic clusters for AI search? Search Engine Land
What does Khadija Zaman do?
Khadija Zaman runs SEO, AEO and GEO as one system at Wellows, so a brand ranks on Google and is cited inside ChatGPT, Gemini, Perplexity and AI Overviews. At Wellows that took LLM referral sessions from 176 a month to roughly nine times that in eight months with no paid spend (GA4, July 2025 to February 2026).
Ranking used to be the finish line. Now it's the entry ticket.
At Wellows, LLM referral sessions grew roughly 9× in 8 months on $0 paid (GA4). The work behind that number was built around three changes in how search systems pick their sources.
-
Change 01
Entities, not keyword lists
Search systems resolve the entities on a page (the brand, the product, the problem it solves) and judge how completely a site covers them. That's why every content plan I build starts from a central entity, not a keyword export.
Keyword listsEntity maps -
Change 02
Passages, not positions
ChatGPT, Gemini, Perplexity, AI Overviews and AI Mode quote a few passages and cite a few sources, and they rarely agree: in my cross-engine study, 79.6% of cited websites appeared on one engine only. How that finding held up.
Rank positionCited passage -
Change 03
Measured referrals, not assumed ones
Visitors sent by an AI assistant blend into GA4's referral and direct traffic unless someone builds a segment for them. I build that segment first, so the work has a baseline to beat. The Wellows numbers.
Total sessionsLLM segment in GA4
None of the three replaces technical SEO. A passage can't be cited if the crawler that fetches it is blocked, so every audit I run starts with access.
Six steps from a prompt to a cited source
This is the pipeline I design pages for. Fan-out is the step most teams never see, and the one I built a tool for at Wellows: one prompt becomes several searches, and each one is a separate chance for a page to be retrieved.
- PromptA buyer asks: which CRM suits a ten-person agency?
- Fan-outThe engine rewrites it as sub-queries on pricing, team size, integrations and reviews. Try it.
- EmbeddingEach sub-query is encoded as a vector.
- RetrievalA search index returns candidate pages; their passages are split into chunks and scored for closeness to each sub-query.
- RerankA second model scores those chunks against the prompt and keeps the best few.
- AnswerThe reply is written from the kept chunks, and their sources are cited 1
Most of my work lands on steps 02 and 04: covering the sub-queries, and writing chunks worth retrieving.
Retrieval works on chunks, so I write in chunks
Before retrieval, a page is split into chunks and each chunk is embedded on its own. Whatever context a chunk needs has to travel inside it.
A paragraph that opens with "as mentioned above" loses its subject once it's split off. That's what Layer 3, Information Extractability, of the R2A framework scores, and what the Citeability Checker flags. It's the first thing I rewrite on a page that ranks but never gets cited.
The work, not the pitch
Three case studies with full write-ups, and four further results with headline figures only. Every case-study number names its source and window.
AI visibility from near-zero: 9× LLM traffic in 8 months on $0 paid spend
Wellows, in-house, Jul 2025 – Feb 2026. LLM referral sessions from a 176/month baseline (GA4); average position 28.9 → 17.3, last 3 months vs the previous 3 (Search Console).
Read the case study →Losing 70% of organic clicks, then recovering to position 8.2
A high-volume publisher, confidential. Five workstreams run by one owner. Search Console, 12 months to May 2026.
Read the case study →Stuck near position 55 for two months, into the 20s in days
Confidential property. A mid-September step change that held for eleven weeks. Search Console, July – November 2025.
Read the case study →Three SEO tools built at Wellows, acquired entirely through organic search
Query Fan-Out Generator, LLM Query Builder, AI Humanizer, all live on Wellows, plus six open-source SEO projects on GitHub. Headline figures only, reconciled on the tools page.
Topic cluster architecture for an affiliate site, six months
Intent-mapped clusters and internal linking, no technical overhaul. Headline figures only; no full write-up on this site.
LLM referral growth for a confidential client
A separate engagement from the Wellows study. GA4; the client and baseline are not published. Headline figure only; no full write-up on this site.
KIVA launch: #1 U.S. rating on All About AI, all organic
Positioning, launch content, SEO-led distribution and partnerships. Headline figures only; no full write-up on this site.
AI search visibility, and the SEO that makes it land
Getting a brand cited inside ChatGPT, Gemini, Perplexity and AI Overviews is the job. Semantic SEO, content architecture and automation are what make the citations arrive and keep arriving.
AI Search Visibility (GEO / AEO)
Entity frameworks built from Google's query fan-out patent and Koray Tuğberk GÜBÜR's semantic model, so content is structured the way retrieval systems select and quote it. Grounded versus non-grounded citation experiments that track explicit citations and implicit brand mentions separately, because they move independently. Free tools built as LLM entry points, answers placed where models read (Reddit, Quora, G2), and LLM referral traffic segmented in GA4 so the result is measured, not assumed.
The method is public: AI search optimization, engine by engine, the R2A Content Framework that scores a page from retrieval to action, the retrieval-readiness checklist, how SEO, AEO and GEO sequence, the study I tried to break, and the free tools built as LLM entry points.
Semantic & Technical SEO
What gets a page retrieved in the first place: site architecture, crawl and indexation governance, schema, Core Web Vitals, and topic coverage complete enough that the site is a likely source.
Content Architecture
Topic clusters and intent mapping built for how models interpret authority, so a site earns citations across a topic rather than on one lucky page.
Measurement & Automation
Python, BigQuery and n8n pipelines that pull GA4 and Search Console data, segment LLM referrals, and report without anyone exporting a spreadsheet. Auditable, so a number can be checked.
Field notes on SEO & AI search
Practical breakdowns on ranking in Google, getting cited by ChatGPT, Gemini and Perplexity, and the automations that make it repeatable.
Built for Two Readers: Why Every Machine Layer on This Site Is Generated From the Page
The build behind khadijazaman.com: Eleventy, a Git-backed CMS, and markdown twins, answers.json, CSP and lastmod all generated from the rendered page.
Read the post →The LLM Technical SEO Checklist: Seven Stages From Crawler Access to Citation
A technical SEO checklist for AI search: seven stages in order, every check with a pass rule and a P0 to P2 priority, backed by sources and run on a live site.
Read the post →Hiring an AI search or
SEO manager?
Full-time or contract, in-house. Tell me about the role and what you're building. I reply within 48 hours with a direct read, not a pitch.