Meta Ads Had Two Great Days, Then a Dead Week: What to Check Before You Turn Them Off

Saturday and Sunday the account did 4x. Monday it did 1x, and by Thursday you’ve had three days with one sale between them, on the same ads and the same budget. You’ve opened Ads Manager more times this week than you’ve opened your email, and you’re one bad morning away from pausing everything.

If that’s you, you have company. Reddit is full of store owners describing the same pattern this year. One owner wrote in April 2026 that they get “1-2 good days followed by entire week of nothing, which eats the cash that the 1-2 good days made.” A reply from someone running a fashion brand described “1-3 days of normal or even higher than normal performance followed by a complete collapse, usually 3-4 days and repeat.” In another thread, someone said they’re “basically playing craps at this point.” These are individual owners talking about their own accounts, not data about Meta as a whole. But if your week looks like that, you’re not imagining it.

The problem is that a “dead week” in Ads Manager can come from three very different places, and each one needs the opposite response. Turning everything off fixes one of them and makes the other two worse. So before you touch anything, work out which one you’re in.

Step zero: is the week actually dead?

Open Shopify, not Ads Manager. Write down, for each of the last 14 days:

• Total orders and revenue (Shopify Analytics)
• Sessions from Meta (Shopify’s sessions by referrer or by UTM source, if your ads carry UTMs)
• Add-to-carts and reached-checkout counts
• What Meta reports as purchases for the same days

Then compare 7 days to 7 days, not day to day. Meta’s default attribution counts a purchase up to 7 days after a click (Meta: About attribution settings), so the last few days in Ads Manager are always unfinished and will usually go up. A Tuesday that looks like zero on Wednesday morning can look fine by Sunday.

Now you have two lines side by side: what Meta says happened and what Shopify says happened. Where they agree and where they don’t tells you which of the three cases you’re in.

Case 1: the platform is resetting or wobbling

What it looks like:

• Shopify orders and Meta-reported purchases drop together.
• The drop started within a day or two of something you changed: a new ad added to the ad set, a budget change, a new optimization event, a paused-then-restarted campaign.
• The Delivery column says Learning or Learning limited.
• Or nothing on your side changed, but CPM jumped across every ad set at once, not just one ad.

Meta says an ad set is in the learning phase until it gets roughly 50 optimization events within 7 days of its last significant edit, and that performance is less stable during that time (Meta: About the learning phase). Significant edits include changing targeting, creative, the optimization event, or the bid strategy; adding a new ad; and large budget changes. A store doing 4 sales a day on one ad set spends most of its life near that line, so every edit can push it back into learning.

The account-wide swing is harder to pin down. In a May 2026 thread, a buyer who said they manage more than $400k a month listed CPM spikes of 50% or more and CPA “easily doubling” on accounts with no structural change. Meta doesn’t explain these swings, and you can’t fix them. What you can do is not make them worse.

What to do:

• Leave it alone for 3–5 days. Write down what you want to change and hold it.
• Batch your edits. If you need to change a budget and add two ads, do it all in one sitting so the ad set re-enters learning once, not three times.
• Raise budgets in steps, not by doubling. The 20–30% step most buyers use isn’t a Meta rule, just a habit that keeps changes small.
• Check Meta Status if the drop is sudden and account-wide. If Meta lists an incident for Ads Manager or the Marketing API that day, the day’s numbers aren’t worth reacting to.

What not to do:

• Don’t duplicate the ad set “to reset it.” The copy starts learning from zero.
• Don’t pause everything and restart next week. An ad set paused for 7 days or more goes back into learning when you turn it on.
• Don’t judge on one day. Decide on rolling 7-day numbers, and only against rules you wrote down before the week started.

Case 2: tracking is losing purchases

What it looks like:

• Shopify orders are roughly flat, or down a little, but Meta-reported purchases fell off a cliff.
• Sessions from Meta and add-to-carts look normal.
• The gap opened on a specific day.

A beginner described this cleanly in a January 2026 thread: after a Meta outage around January 9, they had 9 sales in the store, all from ads by their account, and Meta tracked 5. Their ROAS “tanked from 5.0 to 0.8” on the dashboard. Several people in the replies said the same thing happened to them that week.

Purchases go missing for ordinary reasons too:

• A theme or checkout change. A new theme, a new checkout app, a post-purchase upsell page, or a third-party checkout that sends people to another domain can stop the purchase event from firing.
• Apple privacy changes. Every iOS release tends to tighten tracking a bit more. People on Reddit have been arguing all year about how much iOS 26 affected Meta’s click tracking. I can’t tell you the size of that effect, and neither can they. What you can check is whether your own gap grew after a given date.
• A changed attribution setting, or comparing a 7-day-click column with a 1-day-click column without noticing.

What to check:

• In Events Manager, look at the Purchase event’s volume by day. If it fell on a specific day while Shopify orders didn’t, that’s your date.
• If you send events from both the pixel and the Conversions API, check that Events Manager isn’t flagging deduplication problems.
• Place a real test order on your phone, through an ad link, with your normal payment method. Watch for the Purchase event in Events Manager’s test tool.

What to do: fix the tracking. Then, for the days with the gap, judge the ads by Shopify revenue against total ad spend (MER), not by ROAS in Ads Manager.

What not to do: don’t kill ads that Shopify says are selling because Meta has stopped seeing the sales. And know that the algorithm optimizes on what it sees. If it’s missing a third of your purchases, delivery will get worse for a while even after you fix it, which is one more reason not to pile edits on top.

Case 3: real demand, or the creative is worn out

What it looks like:

• Shopify orders are down, and Meta-reported purchases are down by about the same amount.
• Nothing changed in your setup, and Meta Status is clean.
• One of these patterns shows up in the ads:

For the last one, look outside your account: your email and organic orders, Google Trends for your category, and whether last year’s same weeks looked similar. If every channel dipped together, Meta isn’t the cause and turning it off won’t help.

What to do about worn-out creative: this is the one case where the fix is new ads. After Meta rebuilt ad retrieval around its Andromeda system (Meta Engineering, December 2024), most DTC accounts run broad audiences and let the system choose who sees what. The practical result is that what the ad shows and says does most of the targeting. That same $400k-a-month buyer put it bluntly: the “5% we can still control” is “creative velocity + diversity.”

Diversity is the part people skip. Ten versions of one video with different music still look like one ad. A new pain point, a new opening line, or a new kind of proof is what reaches new people. The week-by-week method is in Creative Fatigue.

Add new creative as new ads next to the old ones. Don’t swap the creative inside the ad that’s still spending. Turn off the tired ones only after the new ones are spending.

Put it on one page

Two habits make all three cases easier to read:

1. Keep a change log. Date, ad set, what you changed. Most “Meta broke” weeks turn out to start the day after an edit nobody wrote down.
2. Watch the total, weekly. Total revenue, total ad spend, MER, new customers, cash in the bank. Daily ROAS swings with whichever day a big order landed. How to set your break-even and read MER is in Does Your Ad Math Add Up?.

And one honest caveat: sometimes it really is Meta, and nothing on this page will make a bad week good. The point of the checks isn’t to make the swings go away. It’s to stop you from making a two-day wobble into a two-week reset by tearing the account apart in the middle of it.

What to do next

Before you change anything this week, fill in the 14-day table from step zero: Shopify orders, Meta sessions, add-to-carts, checkouts, and Meta-reported purchases, side by side. It takes 20 minutes, and it will tell you which of the three cases you’re in.

If it comes out as Case 3 and the bottleneck is making new angles fast enough, that’s the part AutoWhisper does. Add your product and it generates videos built on different angles (pain points, openings, proof) rather than copies of one, at 52 credits per full video, with 65 free credits for new accounts. If you connect your Meta ad account, the ads it builds start paused; you decide when they go live and what they spend. Create an account and start with the product whose ads just went quiet.