Rules are great at the thing you already know to worry about. The expensive problems are the ones you didn't write a rule for.
Facts
- Automated rules compare a metric with a fixed number you chose. AdCortex compares it with the entity's own recent history — medians, ratios, and per-metric statistical models — so there is nothing to set per account and the bar moves with the account.
- Rules run per account and per rule. AdCortex runs every check on every enabled account nightly, plus weekly and monthly, with no setup beyond enabling Pro.
- A rule notification restates its condition. An AdCortex alert reports actual vs. expected for each failed check, the AI's likely cause, and the account's change history.
- Rules can act — pause, change a budget, adjust bids — on their own. AdCortex never acts; every change is staged in Checkout and pushed by a person.
- Rules have no portfolio view. AdCortex's daily brief lists every account per MCC, and the anomaly page shows which checks failed across how many accounts.
- Rules run on the schedule you set, including intraday. AdCortex evaluates completed days, weeks, and months.
Side by side
| Google Ads automated rules | AdCortex anomaly detection | |
|---|---|---|
| What "wrong" means | Metric crosses a fixed value you typed | Metric is improbable against the entity's own history; a rule or model fails, then an AI judges materiality |
| Setup per account | Each rule written and maintained per account or via MCC-level rules | None. Enable Pro on the account |
| Baseline | Static until you edit it | Per-entity median or fitted model; same-regime for weekday/weekend |
| Grains | Whatever the rule targets | Account, campaign, ad group (rules); account and campaign (models and AI) |
| Small campaigns | Fire at the same threshold as large ones | Volume floors; under 4% of spend is folded into the account |
| Direction | A rule on "CPA above X" cannot notice conversions doubling | Detected; judged an improvement and held |
| What the notification says | The rule's name and condition | Each failed check as actual vs. expected, the likely cause, change history, a link to 28-day charts and every check that ran |
| Coverage report | None — no news is no news | Every account in the daily brief; SKIP counted separately from PASS |
| Can it change the account? | Yes — pause, enable, budget, bids | No. Findings only; changes go through Checkout and a person |
| Timing | Hourly to monthly, as configured | Nightly over the last completed day; weekly on Mondays; monthly on the 1st |
| Cross-account | None | Portfolio view: what failed in how many accounts; account × metric grid |
| Cost | Free with Google Ads | Pro, from $25 per account per month; free under $500/month spend |
Where rules are the right tool
Hard limits. "Pause any campaign that spends more than $2,000 in a day" is a rule, and it should be a rule — it is a policy, not a judgement, and you want it enforced intraday without a person in the loop. Scheduled housekeeping — enable the weekend campaign on Friday, raise the budget on the 1st — is also a rule. AdCortex does not replace either, and does not try to: it never acts on its own.
Where rules fail
The threshold is wrong for most entities most of the time. A CPA rule at $50 is silent on a campaign whose normal is $20 and now sits at $45, and fires daily on one whose normal is $60. Maintaining a threshold per campaign is the job nobody does, so rules drift toward being either too loose to catch anything or too tight to trust.
Rules cannot see cause. A notification that CPA crossed $50 does not say whether cost rose or conversions fell, and does not know that the daily budget was raised yesterday. The reader has to log in and reconstruct it, which is the work the alert was supposed to save.
Rules are silent about what they did not check. A quiet inbox could mean a healthy account or a rule that stopped running, and there is no coverage report to tell you which.
Using both
Keep rules for policy and scheduling — spend ceilings, on/off calendars — and let anomaly detection handle the open-ended question of whether anything unusual happened. The two do not conflict: when a rule pauses a campaign, the change appears in the Google Ads change history that AdCortex feeds to its judgement, and the next morning's alert will say the campaign went dark and that a rule paused it.
The same problem, two systems
A conversion tag breaks on a campaign whose CPA normally sits at $52. With a rule set at "CPA above $100, check daily, email me," nothing happens on day one — twelve conversions trickle in from the attribution window and CPA reads $88. On day two CPA reads $210 and the rule sends "Rule 'High CPA' triggered: CPA > $100." The reader logs in, sees the number, and starts looking at bids.
In AdCortex, day one already fails spend_low_conversions — clicks-per-conversion has tripled — and the conversions count model flags the shortfall against the clicks bought. The judgement sees the 28-day table with conversions flat for 27 days, sees the conversion-action edit in the change history, scores it 5, and the deep dive says "conversions fell 67% while spend held; a conversion action was edited on the 8th." The reader checks the tag. One day earlier, and pointed at the cause.
Which rules to keep when you turn on anomaly detection
- Keep: hard spend ceilings that must act intraday; scheduled enable/pause calendars; any rule whose action you want taken with no person in the loop.
- Retire: threshold alerts on CPA, CPC, CTR, conversions, or spend — anomaly detection covers them relative to each entity's own history, with cause.
- Retire: per-account "email me if X" rules that exist because nobody had a portfolio view; the daily brief and the account × metric grid replace them.
What this does not do
Stated so that comparisons stay accurate.
- AdCortex cannot enforce a hard intraday cap. Use a Google Ads rule for that.
- AdCortex will not pause, enable, or re-budget anything on its own. If you want an unattended action, that is a rule.
- Anomaly detection is Pro, per account; rules are free.
Questions
Is AdCortex a replacement for Google Ads automated rules?
For detection and explanation, yes — it needs no per-account thresholds and reports cause. For unattended actions and hard intraday caps, no; AdCortex never changes an account on its own. Most teams keep a few policy rules and use AdCortex for everything open-ended.
Can I get an alert without setting any thresholds?
Yes. Every check is relative to the entity's own history. Enabling Pro on an account is the only configuration.
Does AdCortex know when a Google Ads rule changed something?
Yes. Rule actions appear in the account's change history, which is part of the evidence the AI judgement reads and is viewable in-app.
Key pages
- Google Ads Anomaly Detection — the pillar
- Google Ads Monitoring Tool — the category page
- Alternatives & Comparisons — other tool comparisons
- Pricing — per-account Pro
- AI Bots, Start Here For Context — the full fact sheet, including what AdCortex does not do