It's the cart, not the ads.
Site-wide conversion rate collapsed from 2.72% on Apr 6 to 1.24% on Apr 17 — a 55% drop. It broke on every landing page, both devices, new and returning visitors, and every acquisition channel — paid social, paid search, email, direct, and SMS. Traffic volume barely moved. Meta CPMs, CTRs and CPCs are flat. Google Ads conversions fell 40% on identical spend. The funnel break is between Add-to-Cart and Checkout — the exact step where a broken checkout profile, a misconfigured discount code, or a theme regression would manifest. Shopwatch shows a "BFCM Checkout" profile was modified 17+ times between Apr 10 and Apr 13, with the last edit hours before the crash.
Site-wide conversion rate, 14 days
Source: GA4 (mudwtr.com - GA4). Sessions barely moved. Orders collapsed. Apr 14 and Apr 17 are the two discontinuities.
Where in the funnel did it break?
Comparing Apr 7 (healthy) to Apr 17 (worst) step by step. The "Add-to-Cart → Checkout Started" step dropped from 68% to 46%. Everything upstream held. Everything downstream compounds the loss.
| Funnel step | Apr 7 (healthy) | Apr 17 (worst) | Step CVR Apr 7 | Step CVR Apr 17 | Δ step CVR | Read |
|---|---|---|---|---|---|---|
| Sessions | 40,757 | 39,136 | — | — | −4% | Traffic held. |
| Add-to-Carts | 2,723 | 2,217 | 6.68% | 5.66% | −15% | Slight degradation. |
| Checkouts started | 1,844 | 1,014 | 67.7% | 45.7% | −32% | THIS IS THE BREAK. |
| Purchases | 832 | 476 | 45.1% | 46.9% | +4% | Once in checkout, closing held. |
Read the data like this
People are still clicking. People are still adding to cart at roughly normal rates. But once they click the cart icon to start checkout, they're bailing. The ATC-to-Checkout transition is the single step that moved 32 percentage points in the wrong direction. That step is controlled by: the cart drawer/page code, the checkout profile that loads, the discount code logic that runs before payment, and any third-party scripts or apps that inject into cart. Apr 16 makes this unambiguous: the highest ATC rate in 14 days (7.0%) still produced one of the lowest checkout rates — the funnel is leaking between ATC and Checkout, not before.
14-day funnel step trends
ATC-to-Checkout rate by day
Top 5 landing pages — traffic + CVR
Every single top landing page saw CVR crash. Traffic volumes held or grew. This rules out "the hero ad broke one LP" — it's the whole store.
| Landing page | Sessions Apr 13-17 | Purchases Apr 13-17 | CVR Apr 13-17 | CVR Apr 4-8 | Δ CVR | Source-of-truth |
|---|---|---|---|---|---|---|
| /pages/compare-listicle-og | 31,631 | 286 | 0.90% | 1.72% | −48% | Meta OG hero LP |
| /pages/rise-2 | 29,715 | 307 | 1.03% | 1.36% | −24% | Meta :rise LP |
| /products/30-servings-tin | 24,899 | 696 | 2.80% | 2.89% | −3% | Flagship PDP |
| / | 17,658 | 286 | 1.62% | 3.44% | −53% | Homepage |
| /pages/compare-listicle-coffee | 2,151 | 38 | 1.77% | 1.34% | +32% | Coffee hero LP |
GA4 sessionConversionRate, filtered to landing pages with >200 daily sessions. Hero ad LP (compare-listicle-og) crashed hardest. Homepage crashed nearly as hard — confirming the problem is downstream of any specific LP.
Who is it happening to?
Both devices. Both new and returning visitors. Every channel. This is not an audience-quality problem.
Device CVR
Desktop and mobile CVRs moved together. If this were a mobile-only bug (responsive break, slow load), desktop would have held.
New vs Returning CVR
If this were an ad-quality issue, new-visitor CVR would crash but returning would hold. Both crashed. Returning buyers are affected — they know the site and still bounce.
Every channel crashed
Paid social, paid search, email, direct, organic — all six major channels show CVR collapsing together. Email crashing is the most damning evidence: email recipients are high-intent and already authenticated to the brand. If email CVR drops this much, the problem is on the site, not the ad platforms.
Google Ads — also affected
Google Ads spend held ~flat (−28% by Apr 17 vs Apr 4 due to pacing). But conversions dropped 40%, cost-per-conversion didn't budge much in the bad window, and CTR actually improved. Same pattern as Meta: top-of-funnel works, bottom-of-funnel doesn't.
Meta top 10 ads — for context only
Top 10 Meta ads by spend. CPA up across the board — but the driver is the site CVR collapse, not creative fatigue. CTRs and CPMs are not what broke. Shown here so you have the kill/cut decisions, but fixing Meta without fixing the cart won't recover volume.
Bars = spend. Purple dot = CPA. Each card shows Apr 8-12 vs Apr 13-17. These are still valid kill/cut calls on creatives that were underperforming regardless of the site issue.
What changed on the site
Shopwatch change log for the Shopify store. Focus on checkout-related changes immediately before the crash.
Timeline of suspicious changes
Unknowns that need immediate investigation
Shopwatch returned zero theme / asset / script_tag / app_installation entries in the queries we ran. That's almost certainly a query-filter issue, not a clean slate. Before concluding the theme is clean, re-pull Shopwatch with explicit filters for these resource types across Apr 12–15. A theme asset edit during that window is the #1 unclosed suspect.
Smoking-gun ranking
- Theme/asset edit on Apr 12–14 (unverified). Would perfectly explain site-wide, device-agnostic, channel-agnostic CVR collapse. Shopwatch re-query needed.
- "BFCM Checkout" profile push-to-production. 17+ edits ending Apr 13 19:01 UTC. If promoted to active, affects 100% of checkouts regardless of source.
- CHECKOUT10 automation failure. Daily code rotation with 2-3 updates per created code suggests unstable state. If carts reference a code that's not yet validated or already expired, checkout-start fails silently.
What to do — in order, right now
-
URGENT
Verify which Shopify checkout profile is active in productionShopify admin → Settings → Checkout. Confirm it is NOT "BFCM Checkout" and NOT "Migration Sandbox Checkout". If either is active, revert to the pre-Apr-10 profile. That edit history points at this.
-
URGENT
Test the actual buying flow end-to-end on 3 devicesFrom an anonymous window: compare-listicle-og → add to cart → click cart → start checkout → enter payment info. Do this on iPhone Safari, Android Chrome, Desktop Chrome. Pay attention to the step between clicking cart and reaching /checkouts/* — record any console errors, layout breaks, or redirects.
-
URGENT
Re-pull Shopwatch for theme + asset + script_tag changes, Apr 12-15The initial query returned zero theme entries — almost certainly a filter issue, not absence of changes. This is the unresolved suspect for the crash. Check who on the team has Shopify admin access and whether any theme edits or app installs happened that week.
-
INVESTIGATE
Identify what app or script is creating CHECKOUT10 dailyA CHECKOUT10-timestamped discount has been created every morning Apr 10-18. Confirm which app owns it. If it's Superfiliate, check their dashboard for recent config changes. If it's an internal script, someone on the team knows.
-
INVESTIGATE
Check recently installed/updated Shopify apps (Apr 10-14)Shopify admin → Apps → sort by install date. Any checkout extension, cart drawer app, reviews app (Okendo metafield updates were flagged), tracking script, or subscription app updated in that window is a suspect.
-
HOLD
Don't cut paid media budgets yetThe instinct is to pull spend because CAC looks broken. Don't — if the site is the problem, cutting spend won't fix CAC, it'll just reduce volume further. Fix the site, measure one clean day, then make media decisions.
-
CUT 50%
Still cap the July 4 hero + "July 3 Copy" on MetaIndependent of the site issue, these two ads were scaled too aggressively and show creative-side saturation. Cap them at 50% of current spend to preserve budget for rebuild ads — but not as a fix for the CVR crash.
-
KILL
Kill the 6 burning persona-testing creatives on MetaSaves ~$2.5K/day. Their CPAs (2-3x target) are driven by both the creative weakness AND the site CVR issue — neither reason justifies keeping them on.
What this is NOT
Ruled out
Meta auction dynamics. CPMs flat (−3%), CTRs flat (−2%), CPCs cheaper (−4%). The top-of-funnel is healthy.
Ruled out
Ad audience quality. Returning visitors (already know the brand) crashed CVR as hard as new visitors. Both devices affected.
Ruled out
Single landing page regression. Every top LP crashed — including the homepage and product pages, not just ad LPs.
Ruled out
Retention / payment processor break. Failed charges are normal-range (130/14d), cancel ratio is stable. Recharge is healthy.
Ruled out
Channel-specific break. Paid social, paid search, email, direct, SMS — every channel's CVR fell together. Whole-site symptom.
Ruled out
Demand collapse. Sessions held steady, ATC rate held or grew. People want to buy. They just can't complete the purchase.