If your storefront is failing to load but your ad accounts are healthy, pause your paid campaigns at the campaign level before you do anything else, then work the diagnosis. Ad platforms have no idea your site is broken, so they will keep buying clicks at the same rate and sending them to an error page for as long as the outage lasts.
This playbook covers the specific case where the ads are fine and the store is not. It is the awkward failure mode, because nothing in your ad dashboard looks wrong: impressions, clicks and cost per click all look normal, and the damage only shows up later as a conversion rate collapse for that hour.
How do you confirm it is your store and not the ad platform?
Thirty seconds of checking saves you from pausing campaigns you did not need to pause. Do these three things in order:
- Load your own product page in a private browser window on mobile data, not office wifi. If it errors there too, the problem is not local caching or your ISP.
- Check your platform and CDN, not your ad accounts. Open the live status pages for Shopify and Cloudflare, since a CDN or edge failure produces exactly the same symptom as a platform failure from the visitor's side.
- Click one of your own live ads. If the click bills and the landing page throws a 5xx or times out, you now have proof that spend is converting into nothing.
The base rates support checking the storefront side first. Over the 90 days ending September 10, 2026, Shopify recorded 99.9% uptime with 2 major or critical incidents totalling 124 minutes, average incident 62 minutes, most recently on August 27, 2026. Cloudflare recorded 99.89% with 2 incidents totalling 142 minutes, most recently on August 12, 2026. Over the same window, Google Ads, Meta Ads, Microsoft Advertising, TikTok Ads, Pinterest Ads and Snapchat Ads each recorded 100% uptime with zero major or critical incidents. These figures come from StatusBird's independent 2-minute monitoring over the 90 days ending September 10, 2026. In other words, when your ads are running and your store is not, the delivery side is usually the healthy half.
What is the correct order to pause spend?
Pause highest-spend and highest-intent traffic first, because that is where the money burns fastest. Branded search and retargeting deliver the most clicks per minute for most stores, so they go first even if the total budget is smaller.
- Branded and non-branded search. These are click-billed and intent-heavy. Every click is a customer who wanted to buy right now.
- Retargeting and dynamic product ads. Same logic, plus these visitors already know your store and will notice the error.
- Shopping and product feed campaigns. Prolonged 5xx responses on product URLs can also trigger feed disapprovals, which cost you well past the end of the outage.
- Prospecting video and awareness. Impression-billed traffic with lower click-through, so it is last in the queue.
| Platform | Where to pause fast |
|---|---|
| Google Ads | Campaigns view, filter to Status: Enabled, select all rows, Edit, Pause |
| Meta Ads | Ads Manager, Campaigns tab, tick the header checkbox, toggle Off |
| Microsoft Advertising | Campaigns grid, select all, Edit, Pause |
| TikTok Ads | Campaign tab, bulk select, Change status to Paused |
| Pinterest Ads and Snapchat Ads | Campaigns list, bulk select, set status to Paused |
Pause at campaign level, not ad or keyword level. Campaign level is one action per campaign, it is reversible in one action, and it leaves your structure intact so you are not rebuilding ad sets at 11pm.
Should you pause or cut budgets to near zero?
For an outage you expect to last under an hour, pausing is the cleaner move and easier to reverse. If your account is sensitive to learning resets on Meta or TikTok and the outage looks short, dropping daily budgets to your platform minimum is a middle path that keeps the campaign live while throttling spend. Pick one approach and write it into your runbook now, because the middle of an incident is the wrong time to debate delivery theory.
Two situations where you should not pause everything:
- Checkout is broken but browsing works. Keep prospecting live, swap the landing page to a collection page with an email or SMS capture offer, and pause only the campaigns that point directly at cart or checkout URLs.
- The failure is regional. If the outage only affects one region, use campaign location settings to exclude that region rather than killing global delivery.
What else keeps spending while the store is down?
Ad platforms are the obvious leak, but they are not the only one. Work down this list while the site is still down:
- Scheduled email and SMS sends. A campaign that lands mid-outage drives a spike of traffic into an error page and burns SMS credits. Pause any send scheduled inside the outage window in Klaviyo, Postscript, Attentive, Omnisend or Mailchimp, and reschedule rather than cancel.
- Affiliate and influencer link drops. Message anyone with a post going live in the next hour and ask them to delay.
- Marketplace listings. If your listings on Amazon Seller Central, eBay or Etsy fulfil independently of your storefront, leave them running. That channel is often the only thing still taking orders.
- Automated rules. Any rule that pauses ads on poor performance runs on a schedule, not instantly, so do not rely on it to catch a 20 minute outage.
How do you restart without wasting the recovery?
Do not unpause on the strength of a status page update. Vendor status pages frequently lag the actual recovery in both directions.
- Place a real test order end to end, including payment authorisation, from a mobile device on cellular.
- Confirm the order reached your order management or fulfilment tool, since checkout can look fine while downstream sync is still backed up.
- Unpause in the reverse order you paused: awareness first, then shopping, then retargeting, then search. This spreads the traffic return and gives you a few minutes to catch a partial recovery before the bulk of spend resumes.
- Restore any reduced budgets to their normal level the following morning rather than the same evening, so you are not compressing a full day of budget into a few hours.
- Annotate the window in your analytics tool and record the outage start and end times to the minute. You will need them for performance reviews and for any vendor credit conversation.
How do you find out fast enough for any of this to matter?
All of the above is worthless if you learn about the outage from a customer email 40 minutes in. With a 62 minute average incident length on the platform side in the last 90 day window, a 40 minute detection delay means you acted for the last third of the event and paid full freight for the first two thirds. August 2026 alone saw 16 major or critical incidents across 9 monitored services and 40.4 hours of combined downtime, which you can see broken out in the outage index.
Set up two separate signals. First, an external check on your own storefront URL, hitting a product page rather than the homepage, so you catch theme and app failures that leave the homepage cached and fine. Second, alerts on the services that can take your storefront down without your involvement, primarily your e-commerce platform and your CDN. Route both to SMS or a phone-notifying Slack or Teams channel, because email alone will not wake anyone. Details on how checks are run and how uptime is calculated are on the methodology page, and the full record for your platform is on its reliability history.
If you want that first signal handled for you, StatusBird checks your storefront and the third-party services behind it every 2 minutes and sends alerts by SMS, email, Slack, Teams or Discord. You can set it up here and point the alerts at whoever holds the ad account logins.