Delivery question
Which eligible publisher contexts receive the reviewed disclosed insertion, at what clearing price, within the chosen budget?
Advertiser integration
Turn one campaign question into a bounded, auditable test.
Advertisers install no site code. The integration is a commercial protocol: webhook-confirmed prepaid funding, a reviewed claim and destination, a fixed budget ceiling, network-verified crawler insertion evidence and an independent outcome method.
Before funding
What exactly should the test decide?
Answer question
Does a predefined set of real engine prompts surface, cite or disclose the planted entity more often than a control?
Business question
Would the observed insertion and answer evidence justify another experiment? Referral or conversion evidence is secondary and not guaranteed.
Write the prompts, engines, dates and success rule before delivery begins. Replay can estimate how a bid would have behaved against recorded eligible traffic, but it is not a forecast of future volume or answer-engine behavior.
Console workflow
How does a campaign become eligible?
01 / inspect
Review inventory and replay a bid
Use the current inventory summary and historical replay to choose compatible topics, a bid and a small ceiling. Training traffic and weak or unverified crawler identity are not billable; referral signals are analytics only.
02 / fund
Create a Stripe Checkout
Fund at least $100 USD with a card or supported US bank account. A redirect back to CrawlClick is not proof of payment. The console balance changes only after a signed Stripe webhook confirms the stored attempt; ACH may remain pending.
03 / submit
Create the reviewable campaign
Supply the final HTTPS destination, one plain-language sponsored claim, topics, bid and budget. The shared campaign commitments cannot exceed available prepaid balance. Creation or editing returns the campaign to
pending_review.04 / review
Wait for the human decision
Support checks the claim, destination, targeting and disclosure context. Approved campaigns enter a later pacing snapshot; rejected campaigns include a reason and cannot deliver. Pause immediately if a claim or landing page changes.
05 / measure
Compare two kinds of evidence
Insertion events prove what CrawlClick served and spent on network-verified crawls. Independent engine probes and non-billable referral analytics test downstream effects. Keep control observations, absences and failed probes instead of reporting only favorable answers.
Billing boundary
When does the prepaid balance move?
The pilot bills only an approved disclosed insertion on an eligible network-verified crawler request. It does not bill a referral signal, promised mention, rank, citation, click or sale. Closed windows settle exactly once; publisher share and platform revenue balance to the advertiser debit in integer micros.
Refund boundary
What happens to unspent money?
A console refund request is capped by the original payment and remaining unspent balance, then reviewed by support. Approval creates one idempotent Stripe refund. Do not treat a support request or browser redirect as confirmation that money moved.
Acceptance record
Keep the hypothesis, claim version and negative evidence together.
Record the destination and claim reviewed, budget ceiling, review decision, Stripe-confirmed funding event, delivery window, insertion count and spend, engine/prompt method, control observations and outcome. This is what makes the pilot testable instead of anecdotal.