Each monitor encodes one QA rule a trafficker used to check by hand. The rule runs every ten minutes or so against the live ad-server API — and deep-links straight back into FreeWheel or GAM the moment it fires.
Real engineer. Your stack on screen. No deck.
Every monitor encodes one QA check — the kind your infrastructure monitoring (Datadog, Splunk, the rest) doesn’t see. Datadog watches your servers; VM watches whether your ad server config — including the SSP and partner-deal rules it holds — is actually doing what you sold. Here are four checks a trafficker used to do by hand — and now doesn’t.
Rule written once: "alcohol creative · 06:00–21:00 · broadcast region X." If the partner template drifts, or the trafficker edits by hand, or a new creative inherits an old flight — the monitor catches it before the regulator does.
Rule written once: "per-household frequency cap · 3 / 24h · segment Y." Caps drift through template inheritance more often than people think. The monitor reads the live config, not the dashboard summary.
Rule written once: "segment Z · min 2 impressions · reach threshold." When a segment gets renamed upstream and the rule resolves to empty, nothing errors — the campaign just misses. The monitor compares target to runtime, not target to target.
Rule written once: "bid-floor · partner P · ≥ $X · deal tier T." When the partner-side change lands through their UI, your dashboard shows nothing. The monitor reads the bid-facing config and fires the moment the floor moves past its threshold.
One monitor — a placement-name format check — catching nine drifts in a single day. Each row is a placement that doesn’t match the convention, with a deep link back to the line item to fix.
The growth is the team’s, not ours.
What a deployment typically looks like at six months: three free monitors cover the first month — usually alcohol daypart, household freq cap, one creative-compliance rule. By month three the count is often past twenty: every escalation that surprised the team is now a rule, and every rule that keeps firing has been sharpened. By month six the library can cross eighty — the same first-month seat the deployments below this section grew out of.
VM runs against your existing FreeWheel or GAM account. No data leaves your system. No write access. No new ad-server workflow for your team.
Service-account credentials to your FreeWheel Admin or GAM Network. Scoped to the endpoints VM actually reads. Revocable by you, any time.
Every monitor checks its rule against the live config every ten minutes or so. Not nightly. Not quarterly. Fires as an alert by email or Slack, named with the campaign and the partner.
Every fire is a link straight to the specific line item, creative, or package in FreeWheel or GAM. The trafficker clicks once and lands on the exact object to fix — not a summary, not a search.
Validation Monitors connect to FreeWheel or GAM. Anomaly Monitors (the sister surface) cover a wider set of ad-ops systems including SpringServe — see the Platform overview.
These aren’t pilots. These are monitor libraries in steady-state use.
“On a Friday night, a monitor caught a daypart-config drift that would have run an alcohol spot against a morning audience. The regulator wasn’t the first to know.”
“A 24-million over-delivery on a major title would have walked out the door as make-goods. The monitor caught the trajectory mid-flight, while there was still time to fix the pacing. Partner didn’t need to know.”
“The monitor fires with the number attached. The escalation call starts with what we’re actually losing per hour, not with "something looks off on a dashboard."”
Three categories of money-loss your reports don’t catch. Each one VM watches at the live config — before the loss is the loss.
Alcohol daypart breaches. Political-creative violations. Freeze-window slips. These don’t show up as anomalies — they show up as fines that start at six figures. VM catches the config drift that creates the breach, hours before it goes live.
Frequency-cap drift doesn’t error. It doesn’t alert. It just spends. A 3% over-cap on a $50M campaign is $1.5M of waste — and your reports won’t catch it until reconciliation does, six weeks later. VM watches the live config, not the dashboard summary.
Underdelivery against a sold campaign isn’t fixable after flight — it’s a make-good, and the inventory you owe is free. VM catches the pacing trajectory mid-flight, while there’s still time to fix the cause and hit the guarantee.
Most teams that start with three monitors don’t stop at three. The same data that catches drift becomes the data that explains anomalies, validates partner deliveries, and answers the questions ops directors used to chase across four dashboards. That’s the Operations Control Platform — and Validation Monitors are how teams begin operating it.
