For most publishers the worst outcome is not a slow week of revenue — it is waking up to an AdSense suspension. Ad networks police invalid traffic, ad-placement policy, and click quality on your behalf, and they do it with a blunt instrument: a ban. PubSentry exists to keep you on the right side of that line. The Account Safety surface answers one question directly — how close are you to a ban, and what specifically moves you away from it.
This page explains how PubSentry measures ban-risk, what feeds the number, and where each driver is honest about what it can and cannot see yet.
If you have not installed yet, it is one async tag:
<script async src="https://pubsentry.com/t.js" data-site="st_xxxx"></script>Why ad-network bans happen
Three surfaces drive the overwhelming majority of invalid-traffic suspensions, and they map cleanly onto what an ad network can observe about your site:
- Invalid traffic. Bots, datacenter origins, automation, and declared crawlers loading your ad slots. When invalid impressions become a meaningful share of your traffic, the network's own IVT detection flags the account.
- Ad-setup policy. Placements that violate program policy — ads too close to interactive elements, deceptive layout, more units than the page warrants, sticky/overlay abuse. These get caught by automated policy scans, not by traffic at all.
- Click quality. Suspicious click patterns — clicks that arrive too fast, repeat from the same entity, or correlate with invalid sessions. Invalid clicks are the single most-cited reason for AdSense action.
PubSentry's job is to keep all three healthy before the ad network ever has to act. The first two it can actively prevent at the source; the third it measures and surfaces. The Account Safety score fuses all three into one readout.
The Ban-Risk Meter
The Account Safety screen in the dashboard shows a single composite score from 0 to 100, where higher is safer. It is a transparent weighted blend of the three risk surfaces — there is no black box, and every input is a real measured value pulled from your own traffic, never an invented or industry-average number.
The weights:
| Driver | Weight | What it measures |
|---|---|---|
| Invalid traffic | 50% | Your measured IVT rate and how much was blocked |
| Ad Setup Risk | 30% | Your ad-policy grade (A–F) from the placement scanner |
| Click quality | 20% | The share of tracked clicks classed as abusive |
The composite is simply Σ (driver_score × weight), rounded. Invalid traffic carries the most weight because it is both the most common cause of suspension and the surface PubSentry actively controls in real time. The dashboard's "how it's computed" gloss shows you the same math — the mechanics live next to the number, not buried.
A plain-word status sits under the gauge so the color is never the only signal:
- 85–100 — Protected. Well clear of ban-risk thresholds.
- 60–84 — Watch. Some risk surfaces need attention.
- Below 60 — At risk. Account safety needs action now.
Score drivers, and where they're honest
Each driver renders its own sub-score, its weight, and a one-line detail, and clicking it routes you straight to the screen that explains it.
Invalid traffic (50%) comes from the same scoring engine that runs your block-before-serve gate. The driver detail shows your live IVT rate and the count blocked. Obvious invalid traffic — known bots, datacenter origins, automation tells, declared crawlers — is caught at near-100% recall with zero false positives. Sophisticated invalid traffic from clean residential IPs with no behavioral tells is only partially caught today, by design: blocking it aggressively would risk blocking real humans, which is the one failure PubSentry treats as worse than missing a bot. That gap closes as the reputation network corroborates entities across sites and the ML calibration layer (M9, coming) tightens the engine on real traffic.
Ad Setup Risk (30%) is the placement-policy grade from the tag's content and layout scanner. The scanner inspects how your ads are positioned relative to interactive elements and content, and assigns an A–F grade with a list of findings. A grade of A maps to a 100, F to a 25 — so a clean ad setup contributes fully, and each finding pulls the contribution down. This driver is fully measured the moment the tag loads; it does not need traffic to populate.
Click quality (20%) is the abusive share of clicks tracked on your ad units. If no clicks have been tracked yet, the driver shows an explicit Not tracked yet state rather than a fake 100 — PubSentry never shows a score it has not measured. Once clicks start flowing, the abusive rate drives this contribution down as it climbs.
That "Not tracked yet" behavior is the principle in miniature: before any traffic, the whole Ban-Risk Meter shows Awaiting first traffic, not a flattering placeholder. Honesty about what we have and have not measured is a product rule, not a nicety.
What raises your score
The dashboard does not just grade you — it tells you the move. Any driver below 85 surfaces in a "What would raise my score" panel, sorted weakest-first, each with an imperative action and a direct link to the screen that fixes it:
- Cut invalid traffic → switch Protection Mode to Block so invalid visitors are stopped before the ad fires, and review the rules and IP management screens for any specific abuse pattern.
- Reduce ad-setup risk → work the policy findings list down; each fixed placement lifts your grade and the 30% it contributes.
- Lift click quality → investigate the abusive-click clusters; the click-abuse protection drops invalid clicks at the source.
Because invalid traffic is half the score, moving from Measure to Block mode is usually the single largest lever — it converts detection into prevention.
Score trajectory over time
PubSentry persists a daily snapshot of your safety score and plots the trend on the same screen. There is no back-fill — history starts the day you install, and the chart only plots days we actually measured. Two snapshots are enough to begin a line; until then it shows History starts now. This gives you a defensible record of your account's safety posture trending in the right direction, which is exactly what you want on hand if a network ever questions your traffic.
Alerts and proactive defense
A score you have to remember to check is a score you will miss the day it matters. Ban-risk works best as a tripwire: configure an alert so a sharp move in your IVT rate, a dropped policy grade, or a click-abuse spike notifies you before it compounds. Today notifications are delivered in-dashboard and to a configured webhook; the email channel and a full public API with subscribable webhooks are coming. See the Alerts documentation for setting thresholds and backtesting them against your own history.
The throughline: account safety is not one feature, it is the sum of keeping invalid traffic out, your ad setup clean, and your clicks honest — measured continuously, weighted transparently, and acted on before the ad network ever has to.