Skip to main content

Acting on an alert

The alert flags the problem. Hero AI is where you work through it. Most of what you do with an alert happens in two places: the bell dropdown for quick triage, and the side drawer for actual analysis.

Two ways to open an alert​

You can open an alert from the bell or from the inbox.

The bell lives in the top right of the app. The red badge shows your count of unread alerts. When you click the bell, a dropdown opens with your three most recent unread items. Each row has a dismiss icon: put the pointer on the row to show the small X on the right. Click the X to mark the alert as read. The drawer stays closed. This helps when you already saw the alert by email and just want to clear the badge.

The inbox lives at /alerts, also reachable from the Alerts item in the main sidebar. It shows every alert ever raised for you, with filters across the top for status (active, resolved, unread) and severity (critical, warning, info). Click any row to open the drawer.

What the side drawer shows​

The drawer slides in from the right and contains everything about the alert.

  • Severity and status badges at the top. Severity is one of critical, warning, or info. Status is one of unread, read, or resolved.
  • A small external link icon next to the badges. Click it to open the affected Google Ads account in a new tab. Use this when you want to look at the underlying data in Google's own UI.
  • The alert title and body. A plain English description of what happened and why it matters.
  • A suggested action. One line of copy specific to this alert type. For example, on search_term_waste: "Add this search term as a negative keyword for the campaign or ad group so it stops draining budget."
  • The primary action button: Analyze with Hero AI. This is the main thing you will click.
  • The metadata table. Account ID, first seen timestamp, whether and when Hero Marketer emailed the alert, and (if resolved) when it resolved. The metadata helps you understand context and timing.
  • The full context dump. The raw fields the rule checked when it fired. For a rotten_campaign alert, this includes the campaign name, spend over the last 7 days, baseline cost per acquisition from the previous 30 days, and other fields the rule used.

When you open the drawer, the alert marks itself as read automatically. You do not need to click anything for that.

The Hero AI handoff​

Analyze with Hero AI is the main action on most alerts. Here is what happens.

  1. The drawer closes.
  2. Hero AI opens in a fresh draft chat.
  3. Hero Marketer prefills the input box at the bottom of Hero AI with a context rich prompt. The prompt has the alert title, body, account, and suggested action. It also asks Hero AI to pull the relevant Google Ads data and recommend specific next steps.
  4. You review the prompt. Edit it if you want to add anything specific. Click send.
  5. Hero AI investigates the account with its normal tools. Then it responds with a diagnosis and recommendations grounded in your live data.

The handoff keeps alerts focused on detection and Hero AI focused on diagnosis. The alert tells you something is wrong. Hero AI tells you what to do about it.

When Hero AI reaches the limit of 10 chats​

Hero AI allows up to 10 chat sessions per Google Ads account. If you are at the limit when you click Analyze with Hero AI, the drawer shows an at limit notice instead of the primary button: "You've hit your Hero AI chat limit". The notice has a Manage Hero AI chats link that takes you to /hero-ai, where you can delete an older chat to free a slot. Come back to the alert afterwards, and the Analyze button returns.

The limit of 10 chats applies to each account, not each user. If you have multiple connected accounts, each one has its own 10 chats. See Manage your chats for the details.

The exception: connection broken​

For connection_broken alerts, Hero AI cannot help because the connection to Google Ads is gone. The drawer's primary action is Go to connections instead, which takes you to /settings/connections so you can reconnect the account.

When an automation has already acted​

If you turn on an automation for this alert type, it may already have acted before you open the drawer. An alert triggers both automations: Block wasting search terms acts on a search terms wasting budget alert, and Lift the budget on profitable campaigns being capped acts on a profitable campaign capped by budget alert.

When one of these automations acts on the alert you view, a banner above the suggested action block shows what happened:

  • A plain English summary of the action taken, for example "Added negative keyword: [term] to [campaign]" or "Raised budget: [campaign] from $80 to $100 per day."
  • A Revert button that undoes the action in Google Ads and records a Reverted row in Automation history.

You can still use Analyze with Hero AI whether or not an automation acted. The Hero AI handoff and the automation are independent of each other. For the full automation system, see What automations do.

Mark alerts as read​

An alert becomes read in three ways.

  • Open the drawer. This happens automatically for any alert.
  • Dismiss from the bell. Put the pointer on any row in the bell dropdown and click the X icon. The alert stays in the inbox but counts as read.
  • Mark all read in the inbox. The button in the top right of /alerts clears every unread alert in one click. It is only visible when you have unread alerts.

Read alerts stay in the inbox. Hero Marketer does not delete them. To find them later, filter the inbox by status.

Auto resolve​

Alerts resolve themselves when the underlying problem clears. If rotten_campaign fires on a campaign that then starts converting again, the next monitoring pass detects the change and marks the original alert as resolved. The drawer shows a resolved badge next to the status. The badge persists if you reopen the alert later.

Hero Marketer does not send a "good news" email when an alert resolves. The resolved badge in the inbox is enough. If you specifically want to see what clears, filter the inbox where status equals Resolved.

Manually mark an alert as resolved​

You cannot manually mark an alert as resolved from the UI. The monitoring system decides when an alert resolves, not you. It checks whether the underlying problem cleared. Say you already fixed the underlying issue, but the alert is still open. Wait for the next monitoring pass (every 15 minutes). If the problem is genuinely fixed, the alert will clear on its own.

Say an alert fires for days. The underlying issue is fixed, but the alert does not resolve. Report this as a bug. See Hero AI troubleshooting or contact support.

Unsubscribe from email​

Every email Hero Marketer sends for an alert has a one click unsubscribe link at the bottom. Click it to turn off email for that specific alert type. You do not need to log in.

The in app notification still fires. If you want to turn off both channels for the alert type, do that from Configure alert preferences.

What an alert is not​

Alerts are not predictions. They react to current and historical data. An alert that fires today is about something that happens now, or that already happened. It is not about something that might happen next week.

Alerts are not strategy. They tell you what to look at, not how to fix it. The Hero AI handoff is the bridge from alert to fix.

Alerts are also not a substitute for a dashboard review. They catch a fixed set of common silent failures. Other issues will still show up only when you actually look. The goal is to make sure you do not have to log in to find out about the obvious ones.

Next​