Per-buyer auto-approve

​

Overview

A per-buyer auto-approve override lets you keep mediaBuyApproval: manual for your storefront while opting one trusted buyer into automatic approval for their media buys. It’s the structured way to express “auto-approve buyers I trust, review the rest.” While an override is enabled, that buyer’s media buys skip both Brief Acceptance screening and your review queue and forward to your ad server. It is a deliberately narrow, money-path control:

• Media buys only. This never affects creative review. The buyer’s creatives still follow your creativeApproval setting — creative review is a separate gate and is untouched.
• Relax-only. It can only remove review for a named buyer. It takes effect only while mediaBuyApproval is manual; when the storefront is auto, every buyer already auto-approves and the override does nothing. It can never put a buyer under review.
• Acceptance-policy bypass. The override removes both the create-time policy screen and the human Approval Mechanics step for that buyer’s media buys. Their buys auto-forward even when the evaluator would otherwise have held them. Use it only when you trust the buyer to submit acceptable buys.
• Known buyers only. You can only opt in a buyer that has submitted a media buy to your storefront (whatever its outcome). An unknown or mistyped account id is rejected — you can’t pre-arm a stranger.

​

Lifecycle and audit

Disabling an override is enabled: false — the row is kept, not deleted, so your change-log retains who turned a buyer’s trust on or off and when. Each write records the operator who made it. Because the override lives independently of the storefront-wide setting, it survives an auto → manual flip: if you opt a buyer in, switch the storefront to auto, then later switch back to manual, that buyer’s override is still in effect. List your overrides after tightening back to manual to confirm which carve-outs are live.

​

Policy-bypass audit

Every override exposes policyBypassCount and lastPolicyBypassAt. The count increases when an enabled override auto-forwards a buy that the policy evaluator would otherwise have held; the timestamp records the latest such bypass. Ask your assistant to show your per-buyer auto-approvals. It opens the same portable Page in Murph, Claude, ChatGPT, or another compatible MCP client. The Page lists each opted-in buyer, its policy-bypass audit, previously revoked trust, and every transacting buyer eligible for a new override. You can add, edit, restore, or revoke trust in the Page; enabling always shows the exact money-path consequence before it saves. If you provide an exact buyer and confirm the effect in a prompt, an assistant can call set_buyer_auto_approve directly. Exploratory configuration opens the Page through list_buyer_auto_approvals; neither path hands work back to chat.

​

Identity

The storefront is always resolved from your authenticated identity (customer_id) — never from the request body — so you can only ever set overrides on your own storefront. The buyer is identified by buyerCustomerId; reads echo the resolved buyer name so you confirm the right counterparty.

​

Fields

The list response also includes eligibleBuyers, derived only from buyers that submitted a media buy to this storefront. Each candidate includes its resolved name, mediaBuyCount, and lastMediaBuyAt, so an operator does not need to find or type a numeric buyer id.

​

Related per-buyer levers

Per-buyer auto-approve is operational trust only — it is not a commercial term and is keyed by the buyer’s account id (not a domain). It is independent of the pricing levers, which are keyed by buyer domain:

​

Tasks

​

Related