AI Transparency

Every analysis runs through the same pipeline, in the same order, for every page:

• We retrieve the public content of the URL you submit and parse it into structured sections.
• Deterministic signal resolvers measure objective properties of the page — structure, clarity markers, proof elements, calls to action and more.
• AI evaluators assess the qualitative criteria defined in our published specification, returning structured, validated output rather than free-form text.
• A weighted scoring model combines deterministic signals and AI judgements into category scores and the overall Big Wow Score™.
• Findings and recommendations are generated against the same framework, so two similar pages are judged the same way.

The AI is constrained by a versioned specification: the criteria, rubrics and weights are defined by us, not improvised by the model. That is what makes results repeatable. When the specification changes we version it, so you can tell whether a score moved because your page changed or because the framework did.

We use commercially available large language models from third-party providers, accessed over their APIs from our servers. We may change model versions to improve accuracy, cost or latency. Each request is logged with its model version so results remain auditable.

• Requests are sent server-side; your credentials are never exposed to model providers.
• We do not permit our providers to use your content to train their models.
• Identical requests may be served from an encrypted cache scoped to your account.

• The public content of the page you submitted — text, structure, headings, calls to action and metadata.
• The context you provide, such as your goal, audience or industry.
• The framework instructions for the specific criterion being evaluated.

• Your password, payment details or authentication tokens.
• Your analytics, CRM data or actual conversion numbers — unless you choose to describe them.
• Other customers' pages, reports or data.
• Content behind logins, paywalls or authentication on the page you submit.

Be aware of what the system cannot do:

• It evaluates the page as presented; it cannot measure your traffic quality, offer strength or pricing fit.
• It cannot guarantee a conversion uplift. Recommendations are hypotheses that should be tested.
• AI output can occasionally be wrong, over-confident or miss context specific to your business.
• Highly dynamic, JavaScript-heavy or gated pages may be partially extracted.
• It is not legal, financial, medical or compliance advice.

You are the decision maker. Review recommendations against what you know about your business, prioritise them, and validate meaningful changes with a test where you have the traffic to do so. Report feedback on any finding you believe is wrong — it feeds directly into how we refine the framework.

You own the content you submit and the output generated for your account. We use your data to deliver and support the service, as described in our Privacy Policy. We do not use your content to train our own models or to improve results for other customers.