These answers cover how-to procedures.

Reading them, no, and the onboarding at /start prices a plan for your own market before you join the waitlist. Joining the waitlist takes a short form and no card. James is setting the first customers up himself, so he reads every entry and comes back to you, and nothing is bought until the plan, the ads and the start date are confirmed with you.

Because the refusals are the steps most likely to happen and most likely to be misread as errors. An underpowered geo-lift test, a contaminated control, a market larger than the calculator prices, an ambiguous supplier outcome: each of those stops the sequence deliberately and returns the specific thing that would fix it.

Count each kind separately, with its own number of people per buying team, and add the results. A plan is priced on the total.

It depends on your baseline variance, your conversion volume, how many independent markets you have and how large an effect you would need to detect, so no single figure applies. The power analysis computes exactly that before the test runs, which is why it runs first. A multi-catchment retailer can often power a test that a single national campaign cannot at any budget.

No. Nothing has run live, and a geo-lift is graded causal only under conditions sandbox data cannot meet by construction. The engine, the power gate and both refusals run in the product today, and none of their output is yet evidence about a real campaign.

No. It produces observed response: something that genuinely happened on a property you own, correlated with the campaign by an immutable key. It carries the label observed and is explicitly not a causal claim, and the report says so in its own note. Establishing that the media caused the response requires a powered geo-lift test, which carries a separate label behind a separate gate.

No. It predicts what the clearance bodies and publishers will flag, using deterministic rules against the UK codes. It cannot read a substantiation document and says so explicitly: where you declare that you hold evidence, it records the declaration beside the claim and states that it has not assessed quality. Clearcast, the CAA and the publishers make the decision, and they will ask for the documents themselves.

No. AdBuyMCP runs the hosted MCP server. You use a compatible client whose OAuth identity and exact callbacks have been approved, then sign in through Supabase, select your approved workspace and consent. Joining the waitlist takes a short form and no card. James is setting the first customers up himself, so he reads every entry and comes back to you, and nothing is bought until the plan, the ads and the start date are confirmed with you. MCP access follows that workspace's enabled capabilities and grants no spend authority.

The hosted server supports OAuth, but a claude.ai connection still needs a compatible, approved client registration with its exact callbacks. No external claude.ai connector is listed or accepted here yet. Confirm that the particular Claude product supports pre-registration before trying it; the dashboard buyer's client ID is not an external connector identity.

Nothing that commits money. No agent tool exists for 11 consequential operations, which are human-only signed REST actions, among them approvals, provider-attempt resolution, mandate replacement and revocation, mission closure, workflow acknowledgement, asset-job resolution, dossier saving and verification, and issuing or revoking a live execution grant. What an agent can do is prepare, propose, draft and surface, all of it recorded. The design intent is that an agent removes work rather than removing accountability.

No. It records which data was used, on what basis, and what conditions come with it. That is the raw material for a compliance argument rather than the argument itself. Turning these conditions into enforced launch gates is unfinished work, and AdBuyMCP makes no claim of UK GDPR compliance on your behalf.