Skip to content

Optional analytics

Allow optional analytics to help improve projects, writing and navigation. It uses a pseudonymous first-party identifier after consent. No names, email addresses, messages or search text are collected.

Read the privacy policy
Allen Manoj
← All work
Case studyWebsite intelligence systemIn development

Lens

Turns website signals into a prioritised fix list.

A website intelligence system for founders, agencies, and small teams that need a clearer view of technical health, trust, findability, conversion readiness, and AI visibility.

Discuss a similar systemOpen current build
Diagnostic preview
site health
82
Truststrong
AI readabilityfix
Conversionfix
Example finding: make the service, evidence, and next action explicit.
My role

Product design, systems architecture, and full-stack build

Project stage

In development

What it replaces

Replaces vague website feedback with a scored audit and prioritised fixes.

Current evidence

Current product build · System architecture

01
Overview

What the system is meant to change.

For

Founders, agencies, and small teams that need a clearer read on website trust, AI readability, speed, and conversion issues.

Replaces

Most website audits focus on SEO or speed in isolation. They rarely connect trust signals, AI readability, conversion friction, and page speed into one prioritised view, and almost none account for how large language models and AI search engines read and evaluate a page.

Produces

A prioritised set of evidence-backed findings with enough context for a non-technical owner to understand what failed, why it matters, and what to change next.

02
Architecture

Build the finding as a traceable object, not a line of generated advice.

Lens plans a crawl, extracts HTML, headers, metadata, links, and visible content, then separates deterministic checks from heuristic and AI-assisted analysis. Findings retain their evidence, confidence, rule version, affected URL, and recommended action before prioritisation.

01
Crawl
02
Extract
03
Validate
04
Score
05
Explain
06
Monitor
01

Public website

02

Evidence-backed diagnostics

03

Prioritised fixes

Design constraints
  1. 01Keep deterministic, heuristic, and AI-assisted checks visibly separate.
  2. 02Attach evidence, affected URLs, confidence, and rule version to every finding.
  3. 03Make partial crawl failures visible rather than presenting incomplete coverage as a complete scan.
  4. 04Explain recommendations in language a non-technical owner can act on.
Engineering decisions
  1. 01Evidence before explanation: Captured page signals remain attached to the finding so every recommendation can be inspected.
  2. 02Separate rule families: Deterministic checks, heuristics, and AI-assisted interpretation use different confidence and review rules.
  3. 03Version the scoring layer: Rule and weight versions are part of the scan so future comparisons do not silently mix methodologies.
  4. 04Prioritise the next action: The interface leads with severity, evidence, and the recommended fix rather than a score alone.
System behaviour and reliability
  1. 01A crawl planner selects the pages and resources to inspect.
  2. 02Extraction records technical and visible-page signals.
  3. 03Deterministic, heuristic, and AI-assisted checks run separately.
  4. 04Findings retain source evidence before prioritisation and explanation.
  5. 05Versioned scans can later support monitoring and change comparison.
  6. 06Failed pages remain visible rather than silently reducing coverage.
  7. 07Every finding carries a rule version and affected URL.
  8. 08AI-assisted recommendations remain downstream of captured evidence.
Evaluation approach
  1. 01Deterministic checks are evaluated for reproducibility.
  2. 02Heuristic checks require threshold and false-positive review.
  3. 03AI-assisted checks require groundedness and unsupported-claim testing.
03
Visual proof

Interface views for the workflow and its decisions.

Design representation — replace these views with production screenshots as the product matures.

Diagnostic overview
Diagnostic preview
site health
82
Truststrong
AI readabilityfix
Conversionfix
Example finding: make the service, evidence, and next action explicit.
Diagnostic overview

A category-level view designed to direct attention without treating one score as proof.

Finding detail
Finding · AI readability

Service evidence is difficult to verify

High priority
Evidence

Primary service page does not include a concrete proof point.

Confidence

Heuristic · review recommended

Affected page

/services/data-systems

Next action

Add a specific outcome, method, or verifiable example.

Finding detail

The proposed finding object keeps severity, confidence, evidence, affected URL, and next action together.

04
Evidence and boundaries

What exists now, what remains limited, and what comes next.

This separation is intentional: planned evaluation and reliability work is not presented as completed evidence.

Current evidence

Current product build
System architecture

Limitations

  1. 01The product and diagnostic catalogue are still evolving.
  2. 02Heuristic and AI-assisted findings require explicit confidence and review boundaries.

Next validation

  1. 01Publish the rule catalogue and category-weight methodology.
  2. 02Measure false positives across a fixed set of representative websites.
  3. 03Validate scan comparison across versioned diagnostic rules.
05 · Discuss a similar system

Send the current workflow, source material, report, or product idea.

A polished specification is not required. The rough version is enough to establish the problem boundary and useful next step.

Discuss a similar system
Next case studyAutomated Intelligence Revenue System (AIRS)