The methodology.
Evidence first, and clear about its limits.

Every Visionary Signal assessment follows the same method: structured evaluation, recorded technical evidence, an AI-assisted review, and a final review by Christopher Lee. Each finding says how sure it is, and every recommendation traces back to a recorded gap.

Philosophy

An independent method.
Five commitments, in every report.

Visionary Signal has no product to sell you after the assessment and no reason to inflate a finding. The method is built to keep it that way.

  1. 01

    Evidence before opinion.

    A finding starts with something observed or measured on the site, not with a preference about how the site should look.

  2. 02

    Say how sure.

    Every finding carries an evidence class. Every category score carries its coverage and confidence.

  3. 03

    Recommend only what the evidence supports.

    Recommendations come only from recorded gaps. A familiar fix the evidence does not support is left out, and the report says why.

  4. 04

    Define done. Do not predict results.

    Acceptance criteria say what a change must achieve and how to check it. No business result is predicted without analytics or business data.

  5. 05

    State the limits.

    What the assessment cannot establish is written into every report, not left for the reader to guess.

How an assessment is built

Four layers.
Each one checks the last.

Automation gives breadth and consistency. Review gives judgment. Neither is relied on alone.

  1. 01

    Structured evaluation

    Hundreds of individual automated checks across 9 test areas. The count grows with the pages assessed. A check that could not run is reported as such, never counted as a pass or a failure.

  2. 02

    Technical evidence

    Measured values are recorded with the finding they support, such as contrast ratios against the WCAG minimum, the amount of readable content, and whether crawlers can reach content without JavaScript.

  3. 03

    AI-assisted review

    A language-model review, run to a fixed structure, interprets what the checks found and how AI systems are likely to read and represent the business.

  4. 04

    Founder review

    Christopher Lee reviews every finding before delivery: the evidence behind it, its class, and whether its recommendation is earned.

In Christopher Lee's own experience, the UX/UI and technical review alone has taken 16 to 24 hours when done by hand. That is his experience, not a benchmark.

Optional context sharpens priorities

Business model, audience, buyer journey, key conversion actions, geography, analytics and documents, and a strategic priority can all be added at checkout. With context, findings are prioritized against your chosen priority. Without it, the assessment works from site evidence alone, and the report says so.

Every finding has a business lens

Each finding is mapped to the business area it affects, so a design or engineering issue is never reported without its consequence.

Acquisition Conversion Revenue Retention

Scores, coverage and confidence

Five scores.
Never one.

UX, UI, SEO, LLM readiness and Accessibility are each scored out of 10. Every score is shown with its coverage and its confidence.

Category scores, out of 10

  • UX 10.0 Coverage [COVERAGE] · Confidence [CONFIDENCE]
  • UI 5.8 Coverage [COVERAGE] · Confidence [CONFIDENCE]
  • SEO 9.7 Coverage [COVERAGE] · Confidence [CONFIDENCE]
  • LLM readiness 6.2 Coverage [COVERAGE] · Confidence [CONFIDENCE]
  • Accessibility 6.3 Coverage [COVERAGE] · Confidence [CONFIDENCE]

Scores from one anonymized assessment. Not typical or fixed values. Coverage and confidence are shown as placeholders because the sample does not supply them.

Why there is no overall score

An average would hide the answer.

The five categories measure different things. Any weighting that combined them would be arbitrary.

In the example, a single number would let a UX score of 10.0 blur a UI score of 5.8. Reported separately, each team sees its own score and the evidence behind it.

COVERAGE
How much of the chosen scope a category's checks could examine. Checks that could not run lower coverage. They are never counted as a pass or a failure.
CONFIDENCE
How strongly the evidence supports the score.

Evidence classification

Three classes.
Each finding says how sure it is.

The class decides what happens next. Verified findings can be fixed. Implications are acted on with judgment. Hypotheses are recorded as "no action yet" until evidence confirms them.

Verified

Directly observed or technically established.

Text contrast fails WCAG 1.4.3 AA on all 12 pages examined, with measured ratios of 2.84:1 and 2.78:1 against a 4.5:1 minimum.

Fix. Correct the affected text and background colors so normal text meets 4.5:1, then commission a manual accessibility review to confirm the full scope.

Accessibility
Implication

A reasonable consequence supported by evidence.

No pricing or engagement-model signal appears on any crawled page, so buyers cannot qualify themselves before investing time in a conversation.

Revenue
Hypothesis

An opportunity or explanation that needs validation.

The newsletter form uses a generic button label and a placeholder-only input, and article cards carry no explicit call to action. Both primary actions ask the visitor to infer what to do next.

To confirm. Establish a tracked baseline for sign-ups and article engagement before changing the copy.

Conversion · No action yet

Every finding carries five parts

SEVERITY
A four-step bar, never a color.
EVIDENCE CLASS
Verified, Implication or Hypothesis.
WHY IT MATTERS
Marked as an established effect or an interpretation.
THE FIX
Stated plainly, with the business lens it serves.
TO CONFIRM
For interpretations, what evidence would confirm them.
Anonymized findings from real assessments. The sites are never named.

How recommendations are made

Earned from recorded gaps.
Or left out, with the reason.

A recommendation exists only when a finding records a gap. Each is grouped into a remediation task with dependencies and effort in hour ranges.

  1. 01 · RECORDED GAP

    Text contrast fails WCAG 1.4.3 AA on all 12 pages examined, with measured ratios of 2.84:1 and 2.78:1 against a 4.5:1 minimum.

  2. 02 · RECOMMENDATION

    Correct the affected text and background colors so normal text meets 4.5:1, then commission a manual accessibility review to confirm the full scope.

  3. 03 · GROUPED TASK

    REC-01 [TASK GROUP], with its dependencies and effort in an hour range: [N to N] h.

In one anonymized assessment, the recorded gaps came to an estimated 116 to 423 hours across 47 grouped tasks. Not a typical or fixed value.

Not recommended, and why

The report also records the familiar fixes it chose not to suggest. Restraint is part of the evidence.

B5 · SCOPE

Structured data markup

No structured data was found, but no markup is recommended, because none of the pages shows content a markup type would describe. Markup is not recommended for its own sake.

B4 · AI READINESS

A niche industry page

A niche industry page is not recommended, because the site has no verifiable work in that industry to support it.

Acceptance criteria

Define done.
Instead of predicting results.

Acceptance criteria say what a change must achieve and how to check it. The assessment makes no predicted business results without analytics or business data.

An acceptance criterion, for the contrast finding

Normal text meets 4.5:1 against its background on every page in scope, confirmed by retesting and a manual accessibility review.

Illustration based on the contrast finding above. Retesting status is recorded in the Implementation view.

What the report will not say

  • That a change will raise conversion by a percentage.
  • That traffic or revenue will grow by an amount.
  • That a fix guarantees an outcome.

Those claims need analytics and business data the assessment does not have. When you add analytics at checkout, it informs priorities. It does not become a forecast.

Stated limitations

What it cannot establish.
Stated in every report.

A point-in-time assessment of publicly accessible pages, within the scope chosen at checkout.

What it can establish

  • What the publicly accessible site shows today, within the scope you chose.
  • Which questions essential to buying the site answers, and which it does not.
  • Whether AI crawlers can reach the content, and what AI systems can extract from it.
  • WCAG 2.2 failures that automated testing can detect.
  • Prioritized recommendations drawn only from recorded gaps, with effort in hour ranges.
  • Acceptance criteria that say what a change must achieve and how to check it.

What it cannot establish

  • Actual customer behavior, revenue, conversion rates or retention, without the relevant analytics and business data.
  • Predicted business results. None are made without analytics or business data.
  • Page speed, logged-in areas or real-user testing.
  • Competitor analysis.
  • Ongoing monitoring. It is a point-in-time assessment.
  • A full accessibility conformance review. Manual review is recommended where findings drive decisions.

Put better intelligence to work.

$1,470 for your first assessment, then $1,970 each. Or three assessments for $4,950, used within 30 days.

Questions? Send Christopher Lee a message.