# Recommendations (Beta)

Adsap checks your Meta and Google Ads accounts every day and separates what broke from what could be better, with the evidence behind each one.

Canonical: https://adsap.ai/docs/guides/recommendations
Updated: 2026-09-13

> **Rolling out.** If you do not see **Recommendations** in the menu yet, your account does not have it switched on. The checks still run in the background, so you miss nothing in the meantime. You can also ask for them through the AI Copilot today.

Every day, and once a week for the slower-moving checks, Adsap reads your Meta and Google Ads accounts. It checks them against a fixed list of things that commonly go wrong or get left on the table. What it finds appears at **Performance, then Recommendations**.

Yesterday's numbers are checked every day and sorted into two lists: alerts for what broke, recommendations for what is worth doing. Each card shows the evidence and the money at stake.

## Two lists, not one

The page has two sections, and the difference between them matters:

- **[Alerts](/guides/recommendations/alerts)** mean something broke or changed. Spending stopped, tracking went silent, a switched-on ad is not showing. These need attention now, so they sit at the top.
- **[Recommendations](/guides/recommendations/findings)** are things worth doing. An ad that is losing money, a search term wasting budget, a keyword worth promoting. Nothing is on fire, but there is money in acting.

They stay apart on purpose. An alert mixed into a list of suggestions reads as a suggestion, and an account that stopped spending is not a suggestion.

## What each card shows

- **What was found**, written out in plain language.
- **The evidence**: the actual numbers from your own account, so you can check the conclusion instead of taking it on trust.
- **The money at stake**, where your spend lets Adsap work it out.
- **The suggested next step**, and for some cards a button that does it. See [One-click Apply](/guides/recommendations/apply).

Every checked figure comes from your account's own data. Nothing here is a platform score or a generic benchmark.

## The weekly note

At the top of the page, on weeks where there is something to say, you get a short written summary of the account. It carries the label **AI Generated**, because it is.

Adsap writes it only from the findings below it. It cannot see, and cannot repeat, anything else. Adsap checks every number in it against the findings data before you see it. If a number fails that check, the note does not show at all.

It appears only in weeks where there are active findings. A quiet week gets nothing rather than filler.

## When nothing is wrong

A quiet account says so:

> Nothing needs your attention right now. This account is checked every day.

That is the honest answer, not an empty state to fill. If a connection is broken, you get a reconnect notice instead, because silence from a broken connection is not the same as a clean account.

## Clearing cards

**Mark as done** hides a card you have decided not to act on. If the same problem happens again later, it comes back as a new card instead of staying hidden forever.

Cards that stop being true move to a **No longer an issue** section, struck through, with the date they cleared. The ones you fixed and the ones that resolved themselves both end up there.

## Through the AI Copilot

Ask **"what needs my attention?"** and you get the same two lists, alerts first, across both platforms at once. It works even while the page is not switched on for your account.

## Emails

You can have these delivered by email instead of checking the page. Both are strictly opt-in from **Settings**. See [Email notifications](/guides/email-notifications).