> For the complete documentation index, see [llms.txt](https://docs.adalysis.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.adalysis.com/product/inside-adalysis/how-to-avoid-performance-blind-spots.md).

# How to avoid performance blind spots

{% embed url="<https://youtu.be/C4wyUxYx8w0>" %}

Adalysis ships with sensible defaults. But "sensible" doesn’t mean one-size-fits-all. We'd recommend using the [performance monitor templates](https://docs.adalysis.com/tools/performance/how-to-create-performance-monitors#how-to-add-monitors) for a fast setup, but customizing them for your accounts. Here are some examples:

<table data-view="cards"><thead><tr><th></th></tr></thead><tbody><tr><td><mark style="color:$tint;"><strong>Small account</strong> — minimums too high</mark><br><br>Example: At least 20 conversions/week<br><br>Result: If the account average is lower, the minimum is rarely met and the alert never fires.</td></tr><tr><td><mark style="color:$tint;"><strong>Large account</strong> — default change thresholds too high</mark><br><br>Example: 25% increase in spend<br><br>Result: Smaller shifts aren’t flagged, even though they’re relevant.</td></tr></tbody></table>

### Four decisions to get right

{% stepper %}
{% step %}

#### Choose the right campaigns

Brand vs. non-brand, top-funnel vs. bottom-funnel — set monitors where a change would require action. Using labels means new campaigns are always included.
{% endstep %}

{% step %}

#### Decide on your key metrics

What actually matters for this account — conversions, revenue, ROAS, CPA?
{% endstep %}

{% step %}

#### Set a relevant % change and time frame

For example: 10% over 7 days, or 25% over 14 days.
{% endstep %}

{% step %}

#### Set minimum thresholds

If the account gets 20 conversions/month, a threshold of 10 over 7 days is too high. Consider the account size and data volume.
{% endstep %}
{% endstepper %}

### &#xD;The cascading drop trend

Here's a scenario that often catches people off guard. The performance monitor looks for a 10% drop over 7 days. Each week’s data is under the threshold, but the cumulative drop is a problem.

<figure><img src="https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2FQQgEswUj2bQLGBRshwH5%2F10%25%20threshold.png?alt=media&amp;token=2e7851e4-9476-4461-8f53-a6d65aa81fc6" alt="" width="375"><figcaption></figcaption></figure>

The fix: Duplicate your alert and set a second version for the 30-day window. You’ll catch both sudden drops and slow bleeds.

{% hint style="info" %}
Agree with your team on who should receive notifications and act on them. This helps avoid unneeded messages while ensuring control and speed of response.
{% endhint %}