# Recommendations The seven checks that find money worth moving, four on Meta and three on Google Ads, and the evidence behind each. Canonical: https://adsap.ai/docs/guides/recommendations/findings Updated: 2026-09-13 > **Rolling out.** If you do not see **Recommendations** in the menu yet, your account does not have it switched on. Where [alerts](/guides/recommendations/alerts) mean something broke, these are things worth doing. Adsap checks each one against your own account's numbers, and each card shows those numbers so you can disagree with it. Four checks look at Meta ads. Three look at Google Ads search terms and keywords. ## Meta ### This ad got only a fraction of what its ad set spent Meta is barely showing a new ad, because it puts the ad set's budget into the other ads instead. The card shows what the ad spent against what its ad set spent, over how many days, and how many other ads compete with it. When the ad has spent nothing at all, the card says so plainly, instead of implying Meta tested it and it failed. **Suggested step:** push more delivery to it, or turn it off if you no longer need it. Meta has no ad-level budget, so no button can push delivery. The turn-it-off half gets a button when the ad spent nothing, or spent with no conversions. ### Money spent with no conversions An ad spent a real amount without a single conversion, where one conversion normally costs a known amount in its ad set or the account. The card works out how many times over that amount the ad has already spent. That is what makes zero results more than bad luck. **Suggested step:** turn it off. A button does it. ### Each conversion here costs several times the usual The ad is converting, but each conversion costs several times what conversions normally cost around it. **Suggested step:** turn it off, or change the image, video, or audience before it spends more. The turn-it-off half has a button. ### Only a fraction of viewers watch past the first 3 seconds Plenty of people have seen a new video ad, but few viewers stay past the opening. The card shows what a healthy new video normally keeps. **Suggested step:** replace the first 3 seconds of the video and keep the rest. A variant of this check covers the click rate: plenty of people see the ad and almost nobody clicks. That one suggests changing the image, video, or message before spending more. No button on either. Both ask for new material, and no button can make a video. ### This ad is wearing out People have now seen the ad several times each on average, and its numbers are slipping against the period before. The card names which numbers moved and by how much, for example cost per conversion up 40% or viewers watching past 3 seconds down from one figure to another. It has two levels. **Starting to wear out** is a warning to plan a replacement. **Wearing out** means the decline is underway, and the advice gets more specific: - If people who get past the opening keep watching, replace just the opening 3 seconds and keep the rest. - If the opening still grabs attention but people drop off right after, keep the opening and redo the rest. - Otherwise, make a new version this week. No button. This one asks for new material too. ## Google Ads ### A search term cost you money with no conversions People searching a specific term clicked your ad several times over the window and not one converted, in a campaign where a known share of clicks normally does. **Suggested step:** block it, by adding the term as a negative keyword in exact match. A button does it. ### Searches containing a phrase are wasting money Several different searches sharing a common phrase together got clicks and cost money with zero conversions. The card lists example terms so you can see what the phrase catches. **Suggested step:** block them, by adding the phrase as a negative keyword in phrase match. A button does it. Check the examples before applying this one. A phrase match blocks everything containing that phrase. That is the point, and sometimes more than you meant. ### A search term is working and deserves its own keyword A term brought conversions at a lower cost than the campaign average, but reaches your ads only indirectly, so you have little control over it. **Suggested step:** add it as an exact-match keyword so you can manage it directly. A button does it. The card is honest that this gives you control. It does not guarantee more sales. ## What each card gives you - **The numbers**, from your own account, over a stated window. - **A comparison**, against the ad set, the campaign, or the account. That is what turns a number into a judgment. - **The money at stake**, where Adsap can work it out. - **How long it has been showing**, so a card you have been ignoring for three weeks says so. ## Clearing a card **Mark as done** hides a card you have decided not to act on. If the same problem recurs later, it comes back as a new card. Cards that stop being true move to **No longer an issue**, struck through with the date they cleared. ## The weekly note Weeks with active findings get a short written summary at the top of the page, labeled **AI Generated**. Adsap writes it only from the findings on the page, and checks every number in it against them before you see it. ## Through the AI Copilot Ask **"what should I fix?"** for the same list. The assistant can also apply the fixes for you through the same write paths as the buttons.