What works
The meaning and customer needs already supported.
Copy Lab
Turn customer persuasion needs into measurable copy experiments.
Copy Lab is a governed way to identify what a customer group needs from a page, develop an exact treatment for an exact piece of copy and test the hypothesis through measurable, reversible experiments. At the same time, it turns product pages into structured, decision-ready assets that AI systems can understand and recommend.
The starting point
Start with the customer group, not “better wording” in the abstract.
For a defined audience, Copy Lab compares what people may need to see, feel confident about or imagine before acting with what the current page actually communicates.
- What the customer may need
- Assurance · desire · clarity · relevance · social proof · ease of choice
- What the page currently provides
- Accurate features, but potentially too little confidence, meaning or motivation.
The gap creates a persuasion hypothesis. It is not a claim of proven causality.
Diagnosis
Explain what the wording does, and what it leaves unresolved.
The assessment considers the audience, product, brand and page context together. It is a structured reading, not an automated quality score.
What is missing
Where persuasion value may be left on the table.
Whose need
The defined customer need that is insufficiently addressed.
What may close it
The language, evidence or framing worth testing.
Bounded proposal
Change a defined piece of copy for a defined reason.
Every proposal changes one exact control on one exact page. This generic illustration shows the shape of a hypothesis, not a client example.
- Illustrative control
- Built from hard-wearing material for everyday use.
- Illustrative treatment
- The bag made for the jobs ordinary bags avoid.
Current
Accurate functional reassurance, but limited expression of distinctive capability.
Proposed hypothesis
For a high-use customer group, framing durability as capability may increase relevance and desire.
Every real proposal retains its current and proposed wording; product-description page and destination mapping; governed identity and copy references; an explanation of what changed and why; its technical-readiness record; and independent approval and rollback status.
Change intensity describes how materially the wording changed. It is not a quality score or a prediction of performance.
Persuasive to people. Legible to machines.
Turn product pages into assets AI systems can understand and recommend.
Copy Lab improves the signals that help search, retail and AI systems classify a product, compare it with similar options and retrieve it for a relevant customer need. The copy remains natural, useful and consistent with the brand.
Name the product clearly
Use explicit product types instead of relying on vague pronouns or surrounding page structure.
Add useful retrieval anchors
Make relevant materials, use cases, settings, formats and differentiators easy to identify.
Support comparison and choice
Explain suitability and differences so systems and customers can distinguish similar options.
Validate
Check that the output meets the agreed copy structure and quality rules.
Repair
Correct accepted copy that still needs human or structural refinement.
Fallback
Retain the original when a proposed rewrite does not clear the quality threshold.
Machine readability is a practical measure of clarity. It does not promise that any search or AI system will recommend the product.
Visible, useful product information supports retrieval readiness. It does not replace crawlability, structured site architecture, accurate product data, internal linking, schema or technical SEO.
Review batch
Review the biggest changes first without losing sight of the wider set.
The workspace can surface a bounded five-item “most changed” batch, while keeping every eligible rewrite available to search and browse.
Most changed
Five proposals for focused review.
All rewrites
Browse by page group, product, SKU or destination.
Full context
Current copy, proposed copy, rationale, forces and readiness detail.
Commercial decisions
The client decides which exact wording should move forward.
Select
Include a rewrite in the proposed approval pack.
Deselect
Remove a provisional selection.
Exclude
Keep the rewrite out of the current process.
Request changes
Return the wording for revision with a clear reason.
Workspace selections are not live approvals or launch commands. They form the basis of a commercial approval pack confirmed by the named client content owner.
Governance
Approved wording is not the same as a ready or authorised experiment.
Commercial copy approval
Is this exact wording approved?
Technical readiness
Can we safely target, measure and stage it?
Launch authority
Should the client actually run the test?
A rewrite may be commercially approved but technically incomplete. A technically ready rewrite still cannot launch without explicit client authority.
Experiment preparation
Move into the experimentation platform only when copy and mapping are ready.
Convert sits here as the delivery and measurement layer. It is not the principal client experience and it does not replace the governed decisions made in Copy Lab.
Control and treatment
The approved exact wording.
Page and selector mapping
Where the change can be safely applied.
Allocation and success metric
How the comparison will run and be judged.
QA, launch and rollback
How the test is staged, authorised and reversed.
This is the intended end-to-end workflow; it does not mean that Copy Lab automatically configures or launches live Convert experiences.
Evidence and learning
Record what worked, for whom and in what context.
The approved control and treatment are compared with real visitor behaviour under client authority, agreed allocation, metrics and safeguards. Results return to Copy Lab as qualified learning tied to the customer group, page context and wording hypothesis.
Run
Launch under client authority with agreed allocation, metrics and safeguards.
Measure
Read aggregate outcomes against the agreed evidence threshold.
Decide
Retain the control, adopt the treatment, investigate further or stop and roll back.
What the evidence supports
The observed result and the interpretation that can reasonably be made.
What it does not prove
A winning phrase is not a blanket rule that will work for every audience or page.
What to do next
Adopt, refine, retest or explore a different persuasion gap.
Human-qualified learning keeps the next decision proportionate to the evidence.
Where Copy Lab is today
The secure review and governance middle of the system is built.
- Available now
- Protected review workspace, bounded batches, rewrite comparison, selection controls and governance separation.
- Not yet automated
- Source-Sheet ingestion, approval-pack delivery, live Convert configuration and interpretation of experiment results.
The current alpha supports controlled client review. It does not present unfinished automation as a live capability.
Test readiness
Persuasion cannot rescue a broken experiment.
Copy Lab adds a customer-led hypothesis and stronger governance. It still depends on reliable measurement, sound test design, enough traffic and technically correct implementation.
Measurement
Working analytics, agreed metrics and a valid baseline.
Experiment design
Controlled changes, suitable traffic and an appropriate duration.
Technical quality
Correct targeting, stable pages, staging QA and rollback readiness.
The 10 Forces do not replace analytics, CRO practice or technical execution.
FAQ
Frequently asked questions
Is Copy Lab an AI copywriter?
No. Copy Lab is a governed experimentation service. Technology supports diagnosis and proposal development, while named people approve the exact wording, technical preparation, launch and interpretation.
Does a higher change-intensity score mean better copy?
No. Change intensity only describes how materially the wording differs from the control. It does not score quality or predict commercial performance.
Can a selected rewrite go live automatically?
No. A workspace selection is provisional. Commercial approval, technical readiness and launch authority remain separate decisions.
What role does Convert play?
Convert is the delivery and measurement layer used after the exact copy is approved and the page mapping is technically ready. Copy Lab remains the governed client experience.
How does Copy Lab make product pages easier for AI systems to understand?
It makes important product facts explicit: what the item is, what it is made from, where and how it can be used, who it may suit and how it differs from nearby options. It then validates that structure before the copy moves forward.
What if a page does not have enough traffic?
Dark Horse will recommend a more suitable evidence route rather than force an underpowered test. The hypothesis can be retained for a higher-traffic page, combined evidence or later experimentation.
Commercial next step
Which customer group and page should Copy Lab examine first?
Bring a customer group, a set of product pages and a measurable commercial question. Dark Horse will identify the persuasion gap and recommend a bounded first review batch.