AI visibility monitoring: method and current BeKnow scope

This guide describes a measurement method. Automated AI prompt, mention and citation monitoring is not available in the current public BeKnow product. Collect the observations with an external service or a documented manual sample.

How to apply the method today

1. Define one question and the page, product or project it concerns. Record the source and date of each observation. A provider response is evidence to interpret, not an approved business decision.

2. Carry out the specialist research or content work in the external tool appropriate to the task. Check its current documentation, access requirements and costs. The availability of a provider API does not imply that every capability is integrated in BeKnow.

3. In BeKnow, record the intervention, its hypothesis and the relevant asset. Add the before and after where available. Identify missing evidence rather than filling the gap with an estimate.

4. For supported WordPress and Search Console work, inspect the available verification and checkpoint evidence. For other sources, the initial imports remain separate from automated intervention outcomes.

5. Record whether to continue, stop or investigate, why, and the next step. A review date is a recorded reminder, not a fully managed assignment or notification workflow.

The underlying method

This prevents a misleading claim: the software is free, while external inference is not automatically free. You control model, sample, frequency and provider limit.

• brand mention;
• its role or prominence in the answer;
• named competitors;
• domain links;
• cited sources;
• model, date and execution state.

A mention is not a citation. A model may name a brand without linking it or cite a page without recommending its product. Mention rate versus citation rate explains why they are separate metrics.

1. Choose a small, important question set

Use problems, categories and comparisons from real buyer decisions. Do not put the brand name into every prompt. The AI visibility prompt guide helps build a baseline.

3. Run a first sample

Start with a few questions and confirm that answers, sources and errors are stored. A failed request is not brand absence and should not reduce the result.

4. Verify citations

Open each source and check that it supports the answer. If systematic page retrieval is needed, Firecrawl uses a separate credential.

Keep prompts and model stable enough for comparison. Record sample changes. Value comes from an observable trend, not an isolated percentage.

What “free” means here

BYOK makes the boundary visible: you know which key runs requests and can stop it by removing or revoking the credential.

What a scan cannot tell you

For a focused check, read how to tell whether ChatGPT cites your brand.

Can I monitor without OpenRouter?

You can design the sample, but scans that require model answers need a configured provider.

Is there a universally good percentage?

No. It depends on sample, category and brand maturity. Compare the trend and relevant competitors.

No. Models and prompts depend on your configuration and available capabilities.

Can I stop the cost?

Yes. Reduce frequency and sample size, set provider limits or revoke the key.

Measure less, but make it comparable

A free monitor is valuable when it exposes method and external cost. Begin with a few decision prompts, verify sources and increase frequency only when the evidence changes a decision.

What BeKnow keeps

Keep interventions, hypotheses, assets, observations and decisions in a searchable workspace. Record what is known and what still needs checking.

Preserve the reason for a choice and the next step alongside the intervention. Review dates are recorded; a complete assignment and review inbox is not available yet.

Connections and limits

Shopify, WooCommerce, Magento, GA4, Google Ads, Meta Ads and HubSpot have initial read connectors. Imports are started manually and their data can be inspected or read through MCP. Coverage and setup differ by provider; automatic links to intervention outcomes are still being developed.

A change in performance after an intervention does not prove that the intervention caused it. Cross-platform attribution and a complete analytics dashboard are not available today.

AI and costs

Project memory and data imports do not require an AI model key. A compatible external AI client may have its own costs. BYOK applies only to available functions that actually call an external provider.

Free access during the current adoption phase. No BeKnow credit wallet.

Next step

Start with one project, one documented change and the evidence needed to review it. Source connections.

Record your first intervention. How it works.