Checkout intermittently stops working mid-day, GMC products vanishing without reason, Meta pixel misconnection causing months of structural CV decline, P-MAX CPA spiking 15x overnight — 2 months of platform-level anomalies I can't explain

Title: Checkout intermittently stops working mid-day, GMC products vanishing without reason, Meta pixel misconnection causing months of structural CV decline, P-MAX CPA spiking 15x overnight — 2 months of platform-level anomalies I can't explain


Hello everyone,

I run a Shopify store in Japan (health & wellness niche). Since mid-February 2026, I’ve been experiencing multiple baffling issues that seem to be interconnected. None of them appear to be caused by my own configuration mistakes — I believe they are rooted in the behavior of Shopify, Google, and Meta’s platforms and APIs. I’m looking for anyone who has experienced similar issues or has technical insights to share.


[Issue 1] Checkout intermittently stops functioning entirely, mid-day (mid-February to present — ongoing)

This is the most severe and longest-running problem.

The pattern is consistent: orders come in normally from morning to early afternoon. Then, at some point, conversions drop to absolute zero — while sessions continue to climb. Shopify’s Live View shows 0 active carts, 0 checkouts in progress, and 0 purchases for hours at a time. When I test the checkout myself, everything works perfectly.

This “mid-afternoon shutdown” has been especially pronounced over the past three days (March 12–14), but looking back at our records, similar shutdowns occurred on February 17, 19, 20, 28, and March 2, 3, and 10 — nearly every week since mid-February.

The most telling moment came on March 3, when an actual customer emailed us to say: “The connection was so bad that the screen just went blank before I could complete my purchase, but I managed to buy it eventually.” Since my own device never reproduces the problem, I had no way of knowing that real customers were experiencing a near-unusable site. This is not an algorithm or bidding issue — it means real users are physically unable to reach checkout on a recurring basis.

For reference, on February 21, 2026, Shopify had a documented checkout incident (00:55–22:49 JST) during which customers could not complete payment. A significant discrepancy between ad-reported conversions and actual Shopify orders that day was confirmed with Shopify support and recorded. I believe the intermittent shutdowns since then share the same underlying nature, even if less severe in scale.

What I’ve already tried / ruled out:

  • Uninstalled a bundle product app

  • Uninstalled a smart address validation app

  • Disabled a third-party payment gateway and reorganized payment options

  • Removed all GTM tags from Shopify theme Liquid

  • Switched email marketing apps (the newly reinstalled app is currently the primary suspect)

  • Removed embeds from the official Shopify Subscription app

  • Cleared Chrome site data to eliminate false-positive self-test results

None of these resolved the issue completely. Microsoft Clarity is installed, but since the problem never reproduces on my own device, I haven’t been able to identify a definitive pattern in the recordings.

For reference, our store is running on the Prestige theme.


[Issue 2] Shopify-synced products vanishing entirely from Google Merchant Center

On the afternoon of March 9, 2026, I added SKUs to my products for the first time. Between March 10–14, products began disappearing from Google Merchant Center (GMC) — not as “disapproved,” but completely removed from the list with no trace. Up to 7 products were affected.

The most likely technical explanation is that adding new SKUs changed the product IDs that Shopify sends to Google (shopify_JP_productID_variantID format), causing the API to issue a delete command for the old IDs and attempt to re-register under new IDs — but only the deletion completed successfully. Additionally, the “continue selling when out of stock” setting appears to have caused the availability attribute to be sent as preorder, resulting in products being shown as “coming soon” in Shopping ads and effectively paused.

The critical point is that adding SKUs is a routine product management action. There was no intention to trigger any changes in GMC. The fact that Shopify’s backend can silently send unintended delete commands to GMC is something a merchant cannot predict or control.

I’ve since turned off “continue selling when out of stock” for all products, re-added affected products through the Google & YouTube channel, and stopped Google’s automatic crawl data source. As of March 14, some products are still missing or stuck in “coming soon” status. I’m now considering implementing a supplementary feed via Google Sheets as a permanent fix.


[Issue 3] Meta keeps sending “your campaign has been approved” notification emails repeatedly

Normally, Meta sends an approval notification once when a new campaign or ad is published. However, we repeatedly received approval notifications for existing campaigns that had not been modified.

It was later discovered that one of our main campaigns had been sending conversion signals to a “dead pixel” — a previously used pixel ID that was no longer active — for approximately 60 days. We discovered this on March 7, 2026, and attempted to switch to the correct pixel, but the structure of the existing campaign made it impossible to complete the fix, so we ultimately had to pause it and create a new campaign from scratch.

I believe the recurring approval notifications and the dead pixel misconnection are not unrelated. My hypothesis is that Meta’s system was internally detecting some kind of anomaly but had no way to communicate it accurately to the merchant — so what arrived on our end was simply a repeated “approved” notification, with no indication of the underlying problem.


[Issue 4] Meta ad conversions have been declining structurally over months — this is not a learning or optimization issue

This is a multi-month trend that cannot be explained by bidding strategy or temporary learning fluctuations.

Between November 2025 and February 2026 — roughly three months — we increased our ad spend while conversion volume stayed flat or declined, and our overall ad efficiency fell to less than one-third of what it had been at the start of that period.

This deterioration overlaps almost exactly with the period during which the dead pixel was receiving incorrect signals. In other words, Meta’s algorithm may have spent approximately 60 days learning that users who did not actually purchase were “close to converting.” The result appears to be a fundamentally corrupted targeting model — one that kept spending budget while systematically failing to reach genuine buyers.

I want to be clear: this is not a case of poor bid settings or inefficient budget allocation. This appears to be structural damage caused by a platform-level pixel management failure that silently corrupted the algorithm’s learning data over months, without the merchant having any way to detect it. We still don’t know how long it will take, or how much additional spend will be required, for the algorithm to recover after switching to the correct pixel.

Additionally, on March 7, 2026, we discovered that the Shopify Facebook & Instagram app had silently disconnected from our Facebook account and required manual reconnection. We don’t know when this disconnection occurred.


[Issue 5] Google P-MAX CPA suddenly spiked to more than 15x normal levels, forcing an emergency pause

Our Google P-MAX campaign had been performing within a reasonable range, with actual CPA consistently in the ¥2,000–¥3,000 range. Then, in the early hours of March 5, 2026, we found that the campaign’s performance over the preceding 7 days had deteriorated to extreme levels. P-MAX CPA had reached nearly ¥45,000 — more than 15 times its normal level. We paused the campaign immediately to stop further losses.

When we contacted Google support, we were told that the deterioration was caused by P-MAX shifting a large volume of delivery to YouTube. What became clear from that conversation was that merchants have no way to exclude or limit YouTube delivery within P-MAX. If the algorithm decides to concentrate spend on YouTube and CPA spikes as a result, there is no mechanism on the merchant’s side to prevent it.


[A note on support contacts — to avoid redundant suggestions]

We have contacted all three relevant parties — Shopify, Google Ads, and our third-party payment provider — multiple times regarding the issues described above. In each case, the response was essentially the same: “We see no issues in our own logs. The cause likely lies with another party’s service — please contact them.” To date, no concrete resolution has been provided by any of the three.


Summary and questions:

Looking at all five issues together, what stands out is a common characteristic: the merchant has no way to notice the problem until significant damage has already occurred. The checkout failure is invisible on my own device. The GMC product disappearances were triggered by a routine management action. The Meta pixel misconnection went undetected for 60 days. The Shopify app disconnection had no known start date. The P-MAX CPA explosion had no advance warning.

  1. Has anyone experienced a “checkout invisible wall” where Live View shows zero activity but sessions remain high? What was the cause and how did you resolve it?

  2. Beyond Clarity and Live View, are there any Shopify tools or logs that can show in real time exactly where in the checkout funnel customers are dropping off?

  3. Has anyone experienced GMC products disappearing completely (not just disapproved) after adding SKUs? Did a supplementary feed solve it permanently?

  4. Is there a reliable way to prevent Shopify’s Content API from sending unintended delete commands to GMC?

  5. After running on a wrong pixel for several months and switching to the correct one, how long did it take for Meta’s algorithm to recover and return to normal performance?

  6. When Meta repeatedly sends “campaign approved” notifications without any changes being made, does that indicate Meta’s system is running some kind of internal reprocessing?

  7. Is there a recommended way to monitor the Shopify Facebook & Instagram app connection proactively, so silent disconnections can be detected early?

  8. Has anyone experienced P-MAX suddenly shifting heavy delivery to YouTube, causing a CPA spike? Is there any practical way for merchants to limit or exclude YouTube delivery within P-MAX?

  9. When reporting an abnormal P-MAX CPA spike to Google support, has anyone actually received a meaningful response or resolution?

Nearly two months of weekly anomalies, across three different platforms. Any experience, insight, or even a partial answer to one of these questions would be deeply appreciated.

Thank you.

1 Like

That’s a brutal combo of issues hitting all at once. The checkout thing is the most urgent since you’re literally losing sales right now. Are you on Shopify Payments or a third-party gateway, and have you checked if the outages coincide with any specific traffic spikes or countries? The Meta pixel disconnect causing CV drops for months is its own nightmare but the checkout stopping mid-day sounds like it could be a gateway timeout or some kind of rate limiting issue.

Thank you — I really appreciate your reply.

At the moment, our store is in a mixed payment situation. We had been using a third-party gateway before, but we have been simplifying things and are now relying mainly on Shopify Payments, while some other payment methods are still present.

I have not fully completed the traffic correlation analysis yet, but I did notice some unusual direct traffic from Council Bluffs, Iowa, US, in Shopify reports. However, when I compared it against the actual outage-like hours, that traffic was relatively limited and did not spike enough to explain the drop in add-to-cart sessions, checkout-start sessions, and completed purchases by itself. So at this point I see it as suspicious background noise, but probably not the primary cause.

One important thing I found in Shopify analytics is that there are certain hours where session counts remain fairly normal, but add-to-cart sessions, checkout-start sessions, and completed purchases all drop to zero at the same time. In other words, it is not only completed checkout that disappears — there are also periods where cart additions themselves become unnaturally zero despite ongoing traffic.

That is why I’m no longer sure this is only a payment gateway problem. It may also involve something earlier in the storefront funnel for some users, such as app / JavaScript interference, storefront-side timeout behavior, or some other issue before the final payment step.

At the moment, the app I personally suspect the most is Omnisend, because it was one of the more recent storefront-related changes on our side. We also had a separate Omnisend button URL issue recently, which makes me wonder whether some Omnisend-related script or embed could be interfering intermittently. At the same time, it does not look like a complete Omnisend failure either — some sign-up form submissions and at least one sign-up coupon purchase still happened — so if Omnisend is involved, it may be an intermittent storefront-side conflict rather than a total outage.

What makes this especially difficult is that the problem does not reliably reproduce on my own device. On my side, checkout often appears normal, but at least one real customer reported that the page went blank before they could complete the purchase. So I do think some real users are hitting an invisible wall at certain times of day.

Your point about traffic spikes and countries was very helpful. I’m continuing to compare the affected hours against traffic source, country, and device data. If you have seen cases where merchants saw normal behavior on their own devices, but real customers intermittently lost the ability to add to cart or proceed toward checkout — especially due to app or script interference — I would really appreciate any further insight.

That mixed payment setup could be causing conflicts if the third-party methods are still active during checkout. The unusual direct traffic you mentioned might be bot traffic or click fraud hitting the checkout at the same time, worth filtering by user agent or session duration to see if those visits look real.

Hi @Shonanroad :waving_hand: This is a great writeup useful for public visibility , yet these are the peer to peer forums.

Respect others time.
SEARCH first it’s the rules and just plain saves everyone time and money for common issues, especially when muddying things up by stacking problems.

One post == one problem, don’t glom multiple issues together it only makes everything harder when your expecting help from strangers for free and now a days just encourages useless bot replies that will tell you what you want to hear but not actually fix anything.

e.g. http://www.catb.org/~esr/faqs/smart-questions.html

Checkout issues

NO ONE here can fix server side issues related to checkout.
NO ONE here can get a third party platform like google to magically fix things.
NO ONE.

For actual bugs contact shopify support DIRECTLY: https://help.shopify.com/ → click “chat with human” button in lower right(as of 2026-03)

Most any answers here on the forums for a post like this will just be a form of troubleshooting 101
like turn X & Y off and on again.
i.e. disable EVERYTHING then toggle through the combinations while also on clean device setups with the same customer regions(VPN if need be).

Unless your conflating cart with checkout and omitting critical detail as a result of not making a post focuses on a singular topic.

Then stop using your own device, get clean devices of different configurations and try it.

Thank you for your feedback. I should also note that I am new to the Shopify Community and may not yet be fully familiar with how things work here. I apologize for any unintentional missteps in how I posted.

That said, I’d like to respond to a few of your points.

First, regarding “search before posting”: I did search, extensively, before writing my post. I was unable to find anyone reporting the same combination of phenomena. That is precisely why I posted here. If the purpose of a peer-to-peer forum is to help people solve problems that existing resources haven’t answered, then this post is exactly what the forum is for.

Second, regarding “one post, one problem” being a rule: I reviewed the official Shopify Community Guidelines. That rule does not exist there. The only posting-related requirement in Section 2 is to search for existing topics before posting to avoid duplicates — which I did. The document you linked (catb.org) is a personal guide from the 1990s hacker culture, and is not an official rule of this community. Beyond that, even if such a principle existed, it would only make sense when the root cause of a problem is already known. At the stage where I cannot yet determine whether these phenomena share a common underlying cause, splitting them into separate posts would mean assuming a conclusion I haven’t reached. Describing them together is not carelessness — it is an honest account of what I can and cannot yet determine.

Third, regarding “contact Shopify support directly”: did you read my original post? I clearly stated that I have already contacted Shopify support, Google Ads support, and the relevant third-party app support teams — repeatedly. In every case, I was told that their own logs showed no issue on their end, and was directed to contact someone else. That cycle of referrals is the dead end I am trying to move beyond.

I fully agree that everyone’s time is valuable — yours, mine, and that of anyone reading this thread. Which is why I’ll say this respectfully: if you don’t have a technical idea or insight that might point toward a solution, please don’t feel obligated to reply.

For anyone who does have relevant experience: I am genuinely looking for technical perspectives on what might connect these phenomena, or evidence that they are truly independent. I intend to post separately about any issues I can confidently identify as isolated and unrelated — but I am not yet in a position to make that determination for all of them.

Thank you, OttyAI — this is exactly the kind of technical perspective I was hoping for.

On the payment setup: I have been gradually simplifying it. I removed several third-party methods (AU Pay, D Pay, Merpay) that were running through an intermediary payment gateway, and replaced them with direct integrations (SoftBank Payment for PayPay, and Paidy’s own direct connection). So the setup is cleaner than it was, though I cannot rule out that residual conflicts remain.

On the bot traffic and click fraud point: this is something I had not seriously considered, and it’s worth investigating. What I observed repeatedly was that sessions continued to rise steadily throughout the afternoon — sometimes to record highs — while purchases dropped to zero for hours at a time. On at least one occasion, Shopify’s Live View showed zero active carts, zero checkouts in progress, and zero purchases over an extended period, despite the session count climbing. My own device showed no issues when I tested the site directly.

If a significant portion of that afternoon traffic were bots or click fraud hitting the checkout flow, that could explain why sessions looked healthy while actual purchase behavior collapsed. I will look into filtering by user agent and session duration as you suggested. Do you have a recommended method for doing this within Shopify, or would this require an external analytics tool?

Good call on moving to direct integrations, that should help isolate whether the payment stack was part of the problem.