Go-live checklist: configuration checks (items 1–6)

Goal: in about five minutes, rule out the six configuration faults that make a launch fail silently — a paused ancestor, a rejected creative, a pair that is linked but not matched, a rule that excludes your traffic, a leftover test cap, a zero rate. The numbering is the Go-live checklist’s own; each item says who runs it.

Steps

1. Statuses — the whole chain is Active (everyone, on their own side). Serving requires every level to be active: advertiser, campaign, banner, site, zone, placement. Walk both trees and check the status indicators; a single paused ancestor silently stops everything below it. Watch for “hide inactive” filters in the tree rail — an entity you cannot see may still be the inactive culprit.
2. Approval states (network manager with the Pending tabs; they are not on the stock role, so otherwise your network administrator). On networks that review content, check Pending banners and Pending sites. A pending site still serves, and so does a creative under the soft review mode; under the strict mode a creative stays switched off until approved, and a Rejected creative is switched off. Make sure nothing is deactivated or rejected.
3. Linking — verify from both ends (network manager). Open the banner’s Matched placements card and the placement’s Matched banners card. Both must list the counterpart. The Linked placements card showing a row is necessary but not sufficient — matched means the pair is actually servable (size, type and rules all agree). The pill in the banner’s and the placement’s header should read Serving · N; Not serving means nothing is wired to the other side yet. The placement’s Matched banners card is on the stock Network Manager role (Matched advertisers plus the placement’s Matching rules card); the banner’s Matched placements card needs Matched inventory, which no stock role holds — without it, rely on the serving pill and step 7.
4. Targeting sanity (everyone, on their own levels). For every level that has targeting rules: confirm they describe real traffic. Empty targeting serves to everyone — that is the safe default. Classic self-inflicted blocks: a country list that omits your test location, device targeting that excludes your test browser, time-of-day windows read in a different time zone than you assume (Time Range reads the network’s time zone — UTC unless your network has been set to another zone — unless Client local time is set). Remember targeting stacks across levels — a campaign rule AND a banner rule must both match, and the site, zone and placement pre-targeting must match too.
5. Capping and limits won’t strangle the launch (network manager, and the advertiser where the cards are shown). Check Frequency capping and Limits at campaign (and, where you hold them, placement) level: a leftover test value like “1 impression per user per month” produces exactly one ad and then silence, and a filled total limit has already switched the entity off — it stays off after you raise the cap, until you switch it back on. Include any pinned action rows your network added to the card. And if a row carries an amber ⓘ, your network stops that type of limit before the cap — Delivery stops at 98% of this limit (network setting) — so a 1,000-impression cap ends at 980; plan the figure with that in mind.
6. Pricing is set (network manager, and the advertiser on the campaign). A campaign with the wrong pricing type or a zero rate may serve but produce useless analytics and billing. Confirm the Pricing card matches the insertion order.

What next

• Go-live checklist: serving checks (items 7–10) — diagnostics, preview, a test page and the report
• Go-live checklist — the whole pass, the once-per-network checks and what to do when a step fails