2026 has been unusually rough.
A string of outages across Meta, Shopify, Cloudflare, and AWS has made this one of the worst stretches for DTC advertisers in recent memory. If your campaigns struggled this year, the infrastructure was working against you.
See the full breakdown →If it feels like Ads Manager isn’t telling you everything,
you’re right.
The conditions affecting your performance are mostly hidden:
Surface the signals Ads Manager leaves out.
Try BreezewayEvery day, Breezeway measures how e-commerce advertisers on Meta are performing across cost-per-purchase and return on ad spend. We track a panel of ~45 businesses — all with a solid history of profitable spend on Meta — and compare each one against its own recent history, controlling for the day of the week, since Sundays and Mondays naturally perform differently.
The chart shows whether the panel as a whole had a good, bad, or very bad day relative to what’s normal. A high score means advertisers were paying more than usual to get a purchase. A low or negative score means they were getting purchases cheaper than usual.
Why does this matter to you? If your campaigns had a rough day, this tells you whether it was a you problem — or whether every advertiser on Meta was struggling at the same time. If the panel had a bad day and so did you, it’s likely a platform-wide event. If the panel looked fine but your numbers were down, that’s worth investigating on your own account.
The purple lines are the alert thresholds. When a day’s score crosses the solid line, we call it Bad. When it crosses the dashed line, we call it Very Bad.
To know whether a day is bad, we first need to know what a good day looks like. We established our baseline using a clean reference period — August 10 through November 30, 2025 — a stretch when Meta ad delivery was stable and performance across the panel was healthy and consistent.
From that period, we calculated the typical day-to-day variation in performance: how much does a Monday normally bounce around? How much does a Sunday? Each day of the week has its own rhythm, so we measure each business against its own same-day history rather than a single flat average.
Normal means the panel is performing within its usual range. Bad means performance has drifted meaningfully outside that range. Very Bad means it has moved far enough that it would only happen by chance on roughly one day in a hundred under normal conditions.
The thresholds don’t move with recent bad days — they’re anchored to that clean reference period — so a prolonged rough stretch doesn’t quietly redefine what “normal” looks like.
A Bad Day is when the panel as a whole lands in the bottom 5% of performance compared to our baseline period — August through November 2025, when Meta ad delivery was stable and results across the panel were healthy.
Under normal conditions, with everything working as it should — Meta, Shopify, the internet in general — you’d expect to see a Bad Day roughly once or twice a month. That’s just the natural variation in any advertising system.
When Bad Days are happening more frequently than that, it’s a signal worth paying attention to. It likely means something outside your control is affecting performance broadly — a platform delivery issue on Meta, a Shopify outage, or some other infrastructure problem that’s making it harder for buyers to convert.
A string of Bad Days means your campaigns are not necessarily broken. It means the environment you’re advertising in is working against you — and that’s worth knowing. It also means that if you’re tempted to make big changes to your campaigns in response to poor numbers, you may want to wait and see whether the platform recovers on its own first.
A Very Bad Day is rarer than a Bad Day. Under normal conditions, with Meta, Shopify, and the supporting infrastructure all working properly, you’d expect to see a Very Bad Day only a couple of times a year.
When a Very Bad Day occurs, the panel’s performance has dropped so far below baseline that it’s well outside the range of ordinary variation. If the score for that day exceeds 3, that’s a particularly strong signal — it means the drop was large enough that it almost certainly wasn’t random noise. Something broke somewhere.
When you’re seeing Very Bad Days more frequently than a couple per year — or when the scores are consistently high — it points to a recurring problem that’s well beyond any individual advertiser’s control. The likely culprits are a widespread issue with Meta’s ad delivery system, a Shopify outage, a failure in one of the systems that support those platforms, or a broader internet infrastructure problem. None of those are things you can fix by tweaking your ad creative or adjusting your budget.
We’re seeing significantly more Very Bad Days in 2026 than we should. That pattern tells us that problems affecting advertisers broadly — outside the control of any business, their ads, or their website — are happening with unusual frequency.
The panel is a subset of businesses that use Breezeway to manage their Meta advertising. Not every business qualifies — we apply two filters to make sure the signal is meaningful.
First, each business must have a track record of profitable spend on Meta. Businesses that are still finding their footing, or that have historically struggled to make Meta work, are excluded. We want the baseline to reflect what healthy, well-run Meta advertising actually looks like.
Second, each business must average at least $500 per day in ad spend. Below that threshold, day-to-day results get noisy — a single good or bad order can swing the numbers — and the signal becomes unreliable.
Good catch. We call that the elusive “5th Quarter.” A lot of e-commerce businesses pull way back on advertising after Christmas, and the ones that don’t get unusually cheap clicks and conversions as a result. Less competition in the auction means Meta has to work less hard to spend your budget — and buyers who are still shopping are often highly motivated.
If you’ve got a product or service you can sell between Christmas and New Year’s, keep this in mind for 2026.
Good question. 2026 has seen an unusual run of major platform outages that are well outside the norm.
Meta had significant ad delivery disruptions in late January (Jan 22–26), another outage in early March that locked advertisers out of Ads Manager, and further issues in mid-March. These line up directly with the Very Bad Days you see on the chart.
Cloudflare — whose infrastructure quietly underpins a huge share of the internet — had two major outages in quick succession: a 3-hour event in December 2025 that took down Shopify and AI services worldwide, followed by a 6-hour BGP routing failure on February 20, 2026 that made roughly 20% of all websites unreachable.
AWS CloudFront had a DNS failure on February 10, 2026 that cascaded into outages across 20+ downstream services.
Shopify had at least four significant incidents in early 2026. A widespread outage on February 15 caused 500 errors across merchant admin and mobile. On March 9, admins, checkouts, and storefronts were all throwing errors. March 12 brought a platform-wide outage affecting admin, point of sale, and Shopify Payments caused by a DNS configuration change. And March 17 saw another round of disruptions. When Shopify's checkout is broken, even a perfectly delivered Meta ad results in zero purchases — which is exactly what our CPA signal picks up.
This cluster of failures is not normal. It reflects a period of unusual fragility across the core infrastructure that e-commerce advertising depends on — and it shows up clearly in the data.
We’re a self-funded startup and we can’t afford the “m.”