Research notes on search, AI discovery and how to check what works.
We investigate how search engines crawl real codebases, how generative AI interfaces retrieve citations, and how verified technical changes influence organic commercial pipeline.
This hub will hold our research and the methods behind it. Nothing here is a customer result. Where we show an example, we label it as an example.
- Methodology
- Documented methods
- Verification
- Growth Receipts
- Data Sources
- Search Console and public pages
- Approach
- No benchmark claims without a source
Topics we plan to research, and the method we follow.
- Planned
Crawl Infrastructure & Rendering Bottlenecks
We plan to look at how client-side JavaScript execution affects crawler queues, hydration delays, and indexing rates on commercial websites.
- Client-side hydration delays and crawl timeouts
- HTTP status code distributions during crawler passes
- Static HTML generation vs client hydration characteristics
- Planned
RAG Retrieval & Citation Dynamics
We plan to study how AI search engines extract facts from web documents, generate sub-queries, and select primary domains for citations.
- Dense vector embeddings vs lexical BM25 retrieval behaviour
- Entity verification patterns across public business registries
- Citation variance across model updates
- Method documented
Growth Receipts & Evidence Chains
How a website change is recorded in a Growth Receipt and checked against search performance data afterwards.
- Pre-change defect verification protocols
- Post-change search impression and click tracking methods
- What a receipt records, and what it cannot prove
Name the source.
The SEO and AI industry is crowded with unverified claims, manufactured case studies, and speculative algorithm hacks. Tractionora adheres to three strict research standards:
- 01. Say where it came from: When we state a finding, we say where it came from: Search Console, a public page we read, or a documented source. If we cannot name a source, we do not publish it.
- 02. Zero fabricated case studies: We never invent client testimonials or anonymous revenue metrics. Where customer data is confidential, we explicitly mark illustrative models as examples.
- 03. Reproducible methodologies: Our evaluation methods and schema standards are detailed across our public Methodology documentation so technical teams can verify findings on their own infrastructure.
Apply research-grade rigour to your website.
Run a rapid scan to discover technical defects, crawl errors, and missing structured data across your key pages.