Performance alerts documentation

#What Performance Alerts does

Performance Alerts is a daily monitoring layer that compares your current Google Ads performance with a baseline period and fires when a metric moves enough to matter. It is not a replacement for regular account reviews, but it closes the gap between those reviews by surfacing significant changes as soon as the data is available.

Each alert rule defines a single metric or set of metrics, the direction to watch (increase, decrease, or both), a percentage deviation threshold, a comparison window, and an entity scope. When all conditions are met during the daily scan, Adnine sends an email and stores the alert in your dashboard history.

Common alert jobs

Catch performance drops

Alert when conversions, conversion value, ROAS, impressions, or search impression share falls below your threshold.

Catch spend spikes

Alert when cost, CPC, or CPA increases sharply. Useful for spotting budget overruns or bidding anomalies early.

Focus on what you own

Scope alerts to the campaigns or ad groups your team manages so every notification points to an area you can act on.

#Supported metrics and alert conditions

You can monitor one metric or several metrics within the same alert rule. For each metric, set the direction and the percentage deviation required to trigger the alert.

Setting up metrics, direction, and percentage deviation thresholds for a performance alert rule.

Metric families you can monitor

Pick the metrics your team would actually act on rather than alerting on everything.

Metric family	Examples	Typical alert direction
Spend and delivery	Cost, impressions, clicks, search impression share	Increase for spend spikes, decrease for delivery or visibility drops
Conversion volume	Conversions and other conversion counts	Decrease: when lead, sale, or action volume falls
Efficiency	CTR, CPC, CPA	Decrease for CTR; increase for CPC or CPA
Value and return	ROAS, conversion value	Decrease: when revenue or return weakens
Video and display	Video views, display metrics	Increase or decrease depending on campaign goal

Alert condition settings

Setting	What it controls	Example
Direction	Whether the alert fires on an increase, a decrease, or either.	Alert when conversions decrease.
Threshold	The minimum percentage deviation before Adnine fires an alert.	Alert when conversions drop by at least 20%.
Comparison window	The baseline period used to measure deviation.	Compare the last 7 days with the previous 7 days.
Delay window	Recent days to skip so late conversion data has time to settle.	Analyze data only through 2 days ago.
Entity scope	The account, campaign, or ad group level to evaluate.	Monitor only branded campaigns.

#Comparison windows and daily scan timing

Adnine runs the alert scan once per day. The scan takes the current analysis window, compares it with your selected baseline, and evaluates the percentage deviation for each monitored metric. Today is excluded from analysis by default to avoid alerts triggered by incomplete partial-day data.

Comparison window options

Option	How it works	Use when
Previous period	Compares the current window with the immediately preceding window of the same length.	You want to detect short-term changes after budget edits, creative updates, tracking changes, or market shifts.
Same period last year	Compares the current window with the matching date range from the previous year.	The account has strong seasonality and year-over-year context is more meaningful than the prior few days.

For example, if conversions typically arrive up to two days late, set the delay window to two days. Adnine will then analyse performance only through two days ago, giving the conversion pipeline time to fill before the comparison runs.

#How to set up an alert rule

Choose the metric Start with metrics your team would act on: conversions, cost, CPA, ROAS, conversion value, search impression share, or clicks.
Set the direction and threshold Choose whether increases or decreases matter for this metric, then set a percentage large enough to filter out normal daily volatility.
Pick the comparison window Use previous period for operational monitoring or same period last year for seasonal accounts.
Add a delay window if needed If conversions or revenue data arrive late, skip the most recent one or two days so the comparison uses more complete data.
Scope the alert to an entity level Select account, campaign, or ad group level. Then add name-based filters (contains, does not contain, equals, not equals, or regex) to target the campaigns and ad groups you own.
Configure recipients Add the email addresses of the people who should receive and act on alerts for these accounts. Each team can configure its own recipient list.
Review and refine after the first runs Compare triggered alerts with real account outcomes. Adjust thresholds, scope, or filters if alerts are too frequent or too quiet.

#Scope and campaign filters

Alerts can run at account, campaign, or ad group level. The right level depends on what you want to learn: account-level alerts signal macro changes, while campaign and ad group alerts help pinpoint where a problem started.

Entity levels

Level	Best for	Watch out for
Account	Overall health KPIs: total conversions, total cost, blended CPA/ROAS, and search impression share.	Large accounts can hide campaign-specific problems behind a stable account average.
Campaign	Budget pacing, delivery shifts, campaign-level CPA or ROAS movement, and seasonal changes.	Low-volume campaigns may need wider date ranges or softer thresholds to avoid noise.
Ad group	Narrow diagnosis of a specific theme, keyword cluster, or creative set.	Too many granular alert rules with tight thresholds will generate alerts that are hard to prioritise.

Filters let you include or exclude specific campaigns and ad groups by name. Supported filter logic: contains, does not contain, equals, not equals, and regex. Use these to build alert coverage around naming conventions, campaign types, markets, brands, or product lines without creating a separate rule for each one.

#Email alerts and dashboard history

When a monitored metric crosses its threshold, Adnine emails the configured recipients and creates a dashboard record. Email is the immediate routing channel; the dashboard is the durable log your team uses for follow-up, recurring issue detection, and client updates.

What each alert includes

Detail	Why it matters
Impacted entity	Shows the account, campaign, or ad group where the change was detected.
Metric and direction	Names the KPI that moved and whether it increased or decreased.
Current and baseline values	Shows both values and the percentage deviation used to trigger the alert.
Analysis time frame	Makes clear which date ranges were compared and whether a delay window was applied.
Customer ID and account name	Helps teams with many managed accounts route the alert quickly.

Dashboard records keep every triggered alert available for review after the email has been sent.

#Alert logic and data quality

Before any alert is sent, Adnine simulates the alert condition to confirm the rule is correctly configured and is not firing on incomplete partial-day data. Each alert is calculated using a consistent percentage deviation formula. Generated alerts are stored in a logs table so your team can audit past alerts and track recurring issues over time.

Built-in data safeguards

Safeguard	What it protects against
Simulation before sending	Prevents alerts from firing on invalid or misconfigured rule conditions.
Today excluded by default	Avoids false alerts from partial-day data that will change before the day closes.
Optional delay window	Reduces false positives from late conversions, offline imports, or delayed tracking.
Consistent deviation formula	Keeps alert decisions comparable across different metrics and entity levels.
Persistent alert log	Creates a historical record for review, trend spotting, and team accountability.

#Team access and permissions

Performance Alerts respects team and user-level account permissions. Only customer IDs a user has access to are included in the daily scan, and only authorised users can create or modify alert rules for those accounts.

Permission behaviour

Access rule	Result
User has access to a customer ID	That account can be included in alerts the user is allowed to manage.
User does not have access	The customer ID is excluded from that user's scans and should not appear in their alerts.
Team configures recipients	Alerts can be routed to any email addresses the team designates for that workflow.
Rule creation requires authorisation	Only users with the right permission level for an account should create or modify its alert rules.

#Troubleshooting common issues

Frequently asked questions

An alert fired but the account looks fine. What happened?

A threshold can be crossed without a genuine problem. Check for conversion lag, a tracking gap, a day with unusual ad serving, or a comparison window that happened to include an outlier period. Review the specific dates shown in the alert before making account changes.

I am getting too many alerts. How do I reduce noise?

Raise the percentage threshold so only larger deviations trigger alerts, narrow the entity scope with name filters, use a longer date window to smooth out daily volatility, or remove metrics your team does not act on.

Why does the comparison look wrong when seasonality is high?

If you are using the previous period comparison during a seasonal peak or trough, the baseline will naturally look different from the current period. Switch to same period last year for accounts where year-over-year context is more accurate than week-over-week.

My account uses offline conversions that arrive a few days late. Will that cause false alerts?

Yes, if the delay window is not set. Add a delay window equal to the typical conversion lag (for example, two days) so Adnine only analyses dates where the conversion data is likely complete.

Can I alert on a metric for some campaigns but not others?

Yes. Use the campaign-level scope together with contains or does not contain filters to include specific campaigns by naming pattern. Create separate rules for different campaign groups if they need different thresholds.

I set up an alert but have not received any emails. Is it working?

Confirm the recipient email addresses are correct and that the configured threshold is being met. If the metric has not deviated by the required percentage during any daily scan, no alert will be sent. That may mean the threshold is too high or the account is performing consistently.

#Best practices

Start with high-impact metrics: conversions, cost, CPA, ROAS, conversion value, and search impression share.
Set thresholds large enough to filter out normal daily volatility. A 5% daily swing is rarely actionable.
Use campaign and ad group filters to route alerts to the team member responsible for that area.
Add delay windows for any account where conversion data takes more than one day to fully arrive.
Use previous-period comparisons for operational changes; switch to year-over-year for seasonal accounts.
Keep the number of alert rules manageable. One well-designed rule per KPI beats a dozen overlapping ones.
Review dashboard history weekly to spot recurring patterns instead of treating each alert as a one-off event.
Revisit thresholds after major budget changes, campaign restructures, tracking updates, or seasonal shifts.

#Related guides

Related guides

Explore the Adnine tools that complement performance monitoring.

Performance alerts