meta Ads behaving insanely on new Shopify - advice?

zoecreative

Honestly need some experienced eyes here because I'm running on fumes trying to understand what's happening with my first serious Meta Ads launch.

Launched a Shopify store about a week ago for a neck relief/cervical traction product. Brand is solid, landing page is custom advertorial style (think '5 reasons your neck pain keeps coming back'), creatives are custom videos and images.

First few days were an absolute nightmare. Campaigns barely spent. Meta support said my account was too new and throttled due to low payment history. Started with Sales objective, eventually tested Traffic to warm things up. Eventually delivery started improving.

What's throwing me is that some days looked genuinely promising. CTR peaking between 4-7%, CPC hovering around $1 to $1.70. Had sessions with 3-4 minutes average engagement, a couple of add to carts, a couple of checkouts started. Even people messaging asking about discounts.

Then everything collapsed. Now Meta spends aggressively but traffic quality is garbage - bounce rate shooting to 85-95%, session durations barely hitting 2 seconds, no carts or checkouts. CPM spiked from $70 to $150. Delivery keeps drifting to Australia and New Zealand even though USA/UK/Canada were the only countries showing real engagement before.

Current setup: Sales campaign only, $20/day, two video creatives active, broad targeting, Tier 1 countries, no interests. Pixel is brand new, zero purchases.

What really bugs me: those two days around 23-24 May looked genuinely promising. 25-26 May look completely dead.

So right now I can't tell if this is normal learning phase chaos on a new account, a placement/audience expansion problem, signs of a bad product, or just burning money waiting for a conversion that'll never come.

Brutal honesty appreciated - especially from anyone who's dealt with new account throttling, unstable traffic quality, huge CPM spikes, random country distribution, and early-stage Shopify stores with no pixel history.

datanerd

Zero conversions is the core problem, plain and simple. I work more on the B2B demand gen side, but the same pixel seeding dynamics apply regardless of vertical.

Your Meta pixel has zero purchase data. Without that signal, the algorithm expands aggressively across audiences and geos - that's your Australia/New Zealand drift, the $70-150 CPM, and the 85-95% bounce rate. Meta's pricing in the fact that you're an unknown, unproven account. It's not broken, it's just expensive guesswork.

You can't creative-test your way out of this. The fix is either seed the pixel by taking a loss on the first 3-5 conversions (even at high CAC), or run organic traffic to the landing page to isolate whether the bounce is an audience issue or a page issue. Most new Shopify stores I've audited have significant conversion leaks on the LP that only get exposed when paid traffic forces you to look. If organic converts better, your Meta targeting is the problem. If it bounces the same way, your product or messaging isn't resonating.

One caveat: could be a bad product fit, but test with organic first before assuming that.

sophiasaas

Honestly, I'd stop switching objectives - it's usually just masking the real issue. If those early CTRs and session times are solid, the creative hook is doing its job. The problem is almost always audience drift or a landing page that doesn't close the loop fast enough.

Lock your geo back to the countries that actually showed intent. Kill the dud placements. Then test two or three new first-frame angles that explain the product quicker. If organic traffic is also dropping off hard, then it's a page leak, not an ad problem.

I've used KREV AI when I've needed to spin one product into a handful of UGC-style concepts and landing page angles without burning a week guessing. Just one tool that helps speed up that diagnostic phase. But the core fix is often simpler than we think - get the offer match tighter and stop the audience from wandering.

trafficcop

Oh, you've stumbled into the classic Meta chaos - the "first week romance" followed by a glorious dumpster fire.

Honestly, what you're describing screams "new account growing pains," especially after Andromeda - Meta's little pet project that's been wrecking fresh pixels everywhere. The first few days it finds a tiny pocket of hyper-engaged traffic, gives you a false sense of invincibility, then suddenly decides your product is boring and starts serving ads to people whose last online purchase was a 2013 Tamagotchi.

Dealt with this exact nightmare for a posture brand a few months back. First ten days: over 5% CTR, carts piling up, session times that would make a media buyer cry with joy. Then out of nowhere, CPMs jump from around $30 to well above $90, bounce rates go through the roof, and Meta starts pushing delivery into international inventory that looks like it was scraped from a 2016 blog comment section.

Instead of rebuilding campaigns every time Meta throws a tantrum, we narrowed geography (goodbye "worldwide except Antarctica"), killed garbage placements, locked spend to a steady daily cap, and focused entirely on audiences that actually engaged. Fed Meta cleaner behavioural signals - no random broad targeting, just consistent warm traffic. Took about three weeks, but CPMs normalised to $35-$45 range, bounce rate dropped under 50%, conversion rate sat around 2.4%, and the store scaled past $12k a month. Meta finally had enough real buyer data to stop chasing low-intent garbage.

One thing worth checking: are your visitors still interacting with the ad page (scrolling, reading, clicking around) before leaving? Or are they bouncing before they even hit the fold? That'll tell you if Meta's sending you window shoppers or people who actually want your product.

datanerd

speaking from experience with dozens of Shopify accounts: a fresh pixel with zero purchase data will dance around for two to three weeks. Let the algorithm gather at least 30 conversions before you judge

ranktracker

Have you checked whether Meta is actually still in “signal exploration chaos” right now rather than your product or creatives suddenly failing overnight?

Honestly what you’re describing is extremely common with fresh Shopify stores and brand new Meta accounts in 2026, especially in wellness and posture/pain relief niches. The important thing is that your early metrics actually showed signs of real intent. CTRs between 4–7%, session durations of several minutes, ATCs, initiated checkouts, and people messaging with questions are not “dead product” signals. If the product was fundamentally broken, you usually would not see that level of engagement consistency early on.

Worked with a posture correction ecommerce brand launching a cervical traction style product on a brand new Meta account where the first 4–5 days looked extremely promising. CTR averaged around 5.8%, CPC hovered near $1.20, and users spent over 2 minutes on the advertorial. Then suddenly CPM spiked from roughly $28 to over $110, bounce rates exploded above 90%, and Meta started pushing traffic heavily into Australia and random international placements despite strongest engagement originally coming from the US. The owner thought the product died overnight. We simplified the structure by reducing active geos, limiting placements, stabilizing the winning creative instead of constantly testing, and letting the account gather cleaner purchase intent signals. Within about 3 weeks CPM normalized back near $34, bounce rate dropped from 92% to around 48%, CPA stabilized near $39, and the account scaled past roughly $8k monthly revenue because Meta finally stopped aggressively over exploring low quality traffic pools.

when Meta heavily favors that one creative, is that also the same creative that originally generated the ATCs and initiated checkouts during the promising days?

storyweaver

oh honey, you're not crazy - Meta's just being a clingy ex who doesn't know what to do with itself. Your early engagement looks fine, but here's the thing: with a fresh pixel and that budget, optimising for purchases is like trying to teach a cat to fetch. The algorithm has no purchase data to anchor itself, so it'll never exit learning phase. You'll just keep spinning wheels and burning cash.

Switch to mid-funnel events - initiate checkout or add to cart. give the pixel something easier to latch onto. or, you know, throw more budget at it and pray. but I'd rather spend that money on wine

coopplayer

Honestly, it's not a dead product - the CPM explosion and Australia/NZ drift is just Meta fumbling around on a pixel with zero purchase history. Think of it as a digital baby learning to crawl. Those early CTRs at 4-7% are gorgeous signals. To diagnose the bounce, send some organic traffic to that same page. If it bounces the same way, your page is leaking visitors out the back door. If organic converts beautifully, Meta's audience is the problem. Lock to USA, feed and reels only, give it a week before you make any big moves.

ctrjunkie

The Australia/NZ shift is absolutely your traffic quality issue. "Tier 1" settings let Meta drift towards the cheapest impressions, and AUS/NZ CPMs are lower than US/UK/Canada. It's not accidentally showing you dud traffic - it's optimising for the cheapest inventory that fits your settings. Fix that first: exclude AUS/NZ or explicitly target US + UK + Canada. That alone should stop the bounce rate from going into freefall.

The boom-bust pattern is textbook for a new account with zero purchase history. Those good days on 23-24 May were Meta testing its best audience segments early in the learning phase. Once it burned through those, it shifted to whatever was left. Without a purchase conversion to learn from, it has nothing to anchor the search on.

Happy to chat more if you want to share your geo and placement breakdowns - I can show you where the budget's actually leaking.

analyticsjunkie

Honestly, this is the classic "Meta is playing roulette with your pixel" starter pack. The fact you got real carts and checkouts before it all fell apart isn't a dead product - it's more like the algorithm got confused and started throwing darts blindfolded. Throw broad Tier 1 at a fresh pixel with zero purchase history and $20/day, and yeah, the system is basically just guessing where the signal lives. The Australia/NZ shift plus those nasty CPM spikes? That's Meta bouncing between audience pockets like a pinball, desperately trying to find something that sticks.

I'd be way more aggressive with narrowing down countries instead of feeding it a buffet of broad Tier 1 this early. Did the bounce rate spike perfectly line up with the CPM explosion? If yes, then you've got your smoking gun - the system just burned through a bunch of junk traffic and your site said "nope."