I'm going to show you a structured method for auditing industrial B2B websites from the buyer's evidence perspective — not SEO traffic, but whether procurement, engineering, and quality decision-makers have enough proof to move forward.
This portfolio covers two executions: one self-use operational validation, and one public-site analysis of a global hydraulics manufacturer.
The problem1:00
Navigate to index.html, "The problem" section.
Traditional SEO reports answer traffic and keyword questions. They don't answer whether buyers have traceable proof for high-stakes claims.
Industrial B2B buyers need evidence for certifications, performance limits and tolerances, capacity and lead times, named customer cases, and warranty terms.
If those aren't complete, consistent, and decision-grade, the inquiry cycle extends — or the buyer disqualifies the supplier without ever making contact.
Method overview2:00
Navigate to methodology.html and walk the seven-step flow on screen.
Seven steps, and the key insight is: define evidence standards before inspection.
Buyer Questions — framed from the buyer's perspective, not the supplier's marketing
Evidence Requirements — what counts as sufficient, decided before looking at content
Claim Registration — statements linked to evidence, sufficiency tracked per question
Human Review — certifications, capacity, performance, customer cases flagged for mandatory review
URL Actions — which page, what to change, why it matters
Retest — new snapshots compared against original hashes
The critical principle: Unknown is a valid output. When evidence is missing we preserve Unknown rather than guessing. Industrial buyers need to know what can't be verified publicly.
Case demonstration3:00–4:00
Track A — job interviews
Open case-eigentime.html.
This is a self-use operational dry run on my own professional site — validating the workflow where I had full authority to review findings.
14 URL-level actions with specific pages and changes
Strongest evidence was public resources with versions, dates, and SHA-256 hashes. Weakest was business-results closure and client authorization — exactly what you'd expect from a capability-demonstration site.
Gate status: STOP. Correct result — this isn't a completed client engagement, and the scorecard doesn't pretend it is.
Track B — client discovery
Open case-industrial-hydraulics.html.
A public-site desk audit of a global hydraulics manufacturer, conducted without authorization. The company didn't commission or review this work, and it's anonymized here.
Highest-value finding: high-risk claims are repeated more strongly than their evidence. ISO certifications, OEM relationships, capacity numbers, 100% testing claims — mostly first-party statements without certificate IDs, approval records, or test protocols.
Another material issue: templated location pages presented as factories in cities across three continents, with no public facility evidence. That's a trust risk the moment a buyer checks.
Three provisional P0 actions: publish a complete warranty policy; convert specifications to revision-controlled datasheets; define torque terminology and the safe selection rule.
Gate status: STOP. No client authorization, no approved facts, no adopted actions.
This is the anonymized structural sample — Demo Hydraulic Co. doesn't exist, every figure is invented. What's real is the structure.
Source registry with privacy classification
URL snapshots with timestamps and content hashes
Evidence requirements defined before inspection, including an explicit forbidden_inference column
Priority score is a visible formula — =M+N+O+P — not a number a model asserted
Time, adoption, and payment sheets left deliberately empty
The scorecard returns STOP, and that's the point. Every high-risk item needs client-authorized fact approval, which doesn't exist in an unauthorized desk audit. The formulas aren't tuned to produce a flattering number.
Boundaries1:00
Privacy-first — public pages only. No passwords, CRM, raw inquiries, or PII needed for first-round value.
Unknown is valid — when evidence is insufficient, we say Unknown. Not estimate, not infer.
Human review required — certifications, capacity, performance, customer cases are flagged, never auto-verified.
Validation signals I track: evidence gaps that weren't already in the client's backlog; accepted recommendations with an owner and target date; implementation and retest; value produced without raw CRM; actual labor time measured.
I don't count "the report looks professional" as validation.
Closing0:30
For interviews
I can walk through the workbook structure or specific findings in more detail. The two execution summaries and the decision register are also available if you want to see the full scope and gate logic.
For client discovery
If this fits what you're trying to solve, I can scope a first audit — public pages only, no passwords or CRM. You'd get the evidence gap analysis and URL action list, then decide what to implement.
Likely questions
How long does an audit take?
Depends on scope. 30–50 URLs with 15–20 buyer questions runs roughly 8–12 active hours. Not yet reliably measured across enough client cases to state as a norm.
Can you verify certifications?
No. I can identify whether certificate numbers, validity dates, issuing bodies, and verification links are present. Actual verification requires the client's quality or compliance team.
Do you write the content fixes?
I produce URL-level action lists: which page, what's missing, what to add, why it matters. Content creation and CMS implementation stay with the client's team.
What if we don't have certifications or test data?
The audit flags the gap and marks status Unknown or Missing. It doesn't fabricate evidence. You decide whether to add proof, remove unsupported claims, or leave the gap documented.
How is this different from the SEO audits we already get?
SEO audits optimize for search engines and traffic. This optimizes for buyer evidence sufficiency. The overlap is technical hygiene; the difference is buyer-question framing, evidence classification, and high-risk human review.
Why not just use AI to fix everything automatically?
High-risk claims require human fact approval. Automated changes create liability. The audit identifies gaps; implementation stays with the people who own the business facts.