Research for Better B2B SEO Decisions
Before you approve an SEO budget, you need to know what the evidence supports. The same applies when an engineer compares suppliers or a marketing manager asks an AI tool to explain an agency proposal.
SEO Yolo sells SEO services to industrial and commercial B2B businesses. We have a commercial interest in this subject. Our research programme examines supplier suitability, proposal interpretation and evidence for SEO investment.
This page tracks our questions, project statuses and reporting standard. Our first industrial supplier-search research note is published, with its 120-row study data and methodology available to inspect. The proposal and SEO-investment projects remain in progress.
AI Interpretation of SEO Proposals
Status: initial runs complete. Analysis and report preparation in progress.
If you ask an AI tool to review a proposal, which details does it recover accurately? Does changing how the same information is organised change its answers about fees, scope, responsibilities or missing evidence?
Our first pilot uses three synthetic proposal cases in two information-equivalent text formats. Three runs per combination produced 18 completed API calls covering 144 question-level answers.
The calls used Claude Sonnet 5.5 through its API. This doesn't test ChatGPT or the consumer Claude app.
The proposals are constructed examples. They aren't submissions from real agencies, and the pilot doesn't study real buyers or purchasing decisions. Native PDF layout, page design and document extraction aren't being tested.
Until there's a report to inspect, use our questions to ask an SEO agency to structure your own conversation with a supplier.
Industrial Supplier Search: Published Capability And Buying Requirements
Status: research note published, 16 October 2026. All 120 requirement rows have received machine-assisted source inspection.
Our industrial supplier AI-search research note compares current supplier evidence with five historical compressed-air buying cases. Fifteen searches produced 30 candidate appearances. The note separates supplier discovery, published capability and the evidence still needed for a complete buying requirement. Later pages resolved some previously uncertain capabilities, but don't establish what the original model saw. Read the findings, 120-row study data and methodology and source limitations. These are exploratory textual observations without human engineering endorsement or a consumer-app test. Control Gear, an existing direct agency client, remains included.
Evidence Behind Case Studies and SEO Investment
Status: source feasibility checks underway. No findings published.
We've retrieved public case-study pages to check whether they can support a study of starting points, work completed, measurement periods and commercial outcomes. Questions about attribution, other changes and relevance to your business remain open.
Your own economics need a separate check. Our SEO break-even calculator calculates required additional wins and qualified-enquiry thresholds from your assumptions. It doesn't forecast returns.
What Each Research Report Must Show
You should be able to inspect how we reached a finding. Our reporting standard is to include:
- The question. What we tested and which decision it could inform.
- The source and sample. Where the material came from, how it was selected and what was excluded.
- The dates and model. When the work ran and, for AI tests, the model version and relevant settings.
- A repeatable method. The inputs, prompts, scoring rules and steps used to produce the analysis.
- The denominator. How many pages, cases, runs or answers sit behind each count or percentage.
- The limitations. What the study cannot establish and which explanations remain open.
- The supporting material. Data, examples and code we're permitted to publish, with restrictions explained.
We'll distinguish observations from our interpretation and make revisions visible. Small controlled tests can help refine a question. Extending a finding to real procurement or SEO returns will need evidence from those settings.
The first report, industrial supplier AI search, includes its findings, method and supporting data. We'll add further reports here when they're ready.