Home

Lab

Experiments, notes and systems in progress.

An editorial space to document hypotheses, decisions, failures and learning — without turning prototypes into validated products.

Growth systems

Models connecting strategy, measurement, channels and experience within one operating cadence.

  • Frameworks
  • Field notes
  • Decisions
  • Trade-offs

Web growth

Experiments across architecture, speed, content, trust and conversion.

  • Performance
  • SEO
  • UX
  • Experimentation

AI & automation

Controlled prototypes that expand capacity while documenting limits and failure.

  • Prototypes
  • Evaluation
  • Guardrails
  • Learning

Experimentation discipline

Building quickly does not remove the obligation to evaluate well.

The Lab reduces the distance between a question and useful evidence. Not every experiment needs to become a product; each should produce a decision or documented learning.

Start with a hypothesis

Define expected behaviour, the constraint and what must be observed to continue, adjust or stop.

Build the smallest useful proof

The prototype should answer the question with the smallest possible surface, data and dependencies.

Test the hard cases

Averages hide critical failures. AI and automation tests include risky intents, fallback, ambiguity and human intervention.

Give the experiment a state

Building, testing, paused, archived and private prevent an experiment being mistaken for a product or commercial validation.

Boundaries

Curiosity with privacy and honesty.

Internal tools may be described editorially but are not public by default. Screenshots, data and links only enter after privacy review and approval.

  • No credentials, tokens or protected URLs
  • Fictional or anonymised data
  • No invented metrics
  • Failures and archive decisions preserved
  • Human oversight described

Questions about the Lab

Are all Lab projects available?

No. Some are private, incomplete or awaiting approval. A public editorial status never implies access to the tool or internal data.

How do you decide whether an experiment continues?

The decision depends on utility, risk, reliability in critical cases, operating cost and learning quality — not only a positive average.

Why document archived experiments?

Because a decision to stop can reveal criteria, limits and judgement that a success-only case would hide.

Explore research tracks

Web Growth

Architecture, content, performance, trust and conversion.

AI & Automation

RAG, evaluation, guardrails and human-in-the-loop.

Work

Public cases with explicit context and status.