Re-analysis of all 7 tracking problems identified on May 19 — what's fixed, what's worse, what hasn't changed
| Window | Spend | Purchases | CPA | Add Payment | Purchase:Payment | vs Yesterday |
|---|---|---|---|---|---|---|
| 90 Days | $1,209,063 | 8,164 | $148.10 | 3,296 | 2.48x | ~same |
| 60 Days | $863,058 | 6,641 | $129.96 | 2,636 | 2.52x | new window |
| 30 Days | $503,915 | 3,785 | $133.13 | 1,352 | 2.80x | ~same |
| 14 Days | $184,316 | 853 | $216.08 | 348 | 2.45x | ↑ from $209 |
| 7 Days | $96,220 | 442 | $217.69 | 171 | 2.58x | ↑ from $211 |
The 7-day CPA ($218) is now 64% higher than the 60-day CPA ($130). Yesterday this gap was 60%. The account is spending more per purchase with each passing week. The 7-day window went from $93K spend / 442 purchases to $96K / 442 — spend increased $3K but purchase count didn't budge.
Yesterday: Purchase-to-Payment ratio worsening (2.31x → 2.22x → 2.80x). Add Payment Info missing ~60% of events.
Today: Ratio unchanged across all windows. Still 2.45x–2.80x depending on timeframe. The 14-day ratio (2.45x) is slightly better than the 30-day (2.80x) but this is within normal noise.
Bottom line: 12,044 people initiate checkout → only 1,352 Add Payment events fire (11.2%) → yet 3,785 complete a purchase. This is physically impossible in a normal checkout flow. The Add Payment Info pixel event is broken.
Yesterday: ViewContent jumped from ~59K to ~117K (+98%) between Feb-Mar and Mar-Apr windows with only 6% spend increase.
Today: The Apr-May window shows 113,695 ViewContent events — essentially flat vs the Mar-Apr window (117,531). The doubling from February is baked in and stable. Whatever was changed in March (pixel reinstall, added to more pages, or Shopify app) is still in effect but not getting worse.
| Period | ViewContent | Spend | VC/Spend | Change |
|---|---|---|---|---|
| Feb 20 – Mar 19 | 56,005 | $328,672 | 0.17 | baseline |
| Mar 20 – Apr 19 | 117,531 | $369,250 | 0.32 | +110% ⚠️ |
| Apr 20 – May 20 | 113,695 | $503,915 | 0.23 | -3% (stable) |
Yesterday: CPA dropped from $231 to $128 (45% decrease) on only 6% more spend — flagged as potentially inflated by pixel changes.
Today: The $128-$133 CPA range has held for 2 months. At this point, the lower CPA appears to be the "new normal" driven by the pixel changes in March. The ViewContent doubling likely brought in more attribution signal, making the CPA appear better. The question is whether the underlying conversion quality is real (Shopify cross-check needed).
| Period | Purchases | Spend | CPA | Change |
|---|---|---|---|---|
| Feb 20 – Mar 19 | 1,472 | $328,672 | $223.28 | baseline |
| Mar 20 – Apr 19 | 2,889 | $369,250 | $127.81 | -43% |
| Apr 20 – May 20 | 3,785 | $503,915 | $133.13 | +4% |
Yesterday: Identified May 8-9 as a clean inflection point where ROAS collapsed from 0.86x to 0.31x. Confidence: 65-70%.
Today — Revised Assessment: With the full daily data, the picture is more nuanced. It's NOT a clean one-time break. The data shows wild oscillation — purchases swing from 66 to 161 to 90 to 211 to 59 day-to-day with NO correlation to spend. This is characteristic of inconsistent pixel firing, not a single tracking event.
| Date | Spend | Purchases | CPA | Add Payment | P/API Ratio | Signal |
|---|---|---|---|---|---|---|
| May 5 | $9,986 | 46 | $217 | 26 | 1.8x | ✓ Ratio OK |
| May 6 | $10,994 | 58 | $190 | 24 | 2.4x | — |
| May 7 | $20,147 | 161 | $125 | 43 | 3.7x | Best day |
| May 8 | $18,026 | 66 | $273 | 26 | 2.5x | ⚠️ 2.2x CPA jump |
| May 9 | $17,597 | 161 | $109 | 35 | 4.6x | Recovered |
| May 10 | $14,574 | 90 | $162 | 20 | 4.5x | — |
| May 11 | $15,949 | 143 | $112 | 34 | 4.2x | — |
| May 12 | $16,553 | 113 | $146 | 17 | 6.6x | ⚠️ Ratio spike |
| May 13 | $17,756 | 211 | $84 | 45 | 4.7x | Best CPA |
| May 14 | $20,005 | 119 | $168 | 12 | 9.9x | ⚠️ Ratio collapse |
| May 15 | $13,880 | 93 | $149 | 9 | 10.3x | ⚠️ WORST ratio |
| May 16 | $16,155 | 137 | $118 | 27 | 5.1x | — |
| May 17 | $12,580 | 59 | $213 | 12 | 4.9x | ⚠️ Low volume |
| May 18 | $14,413 | 94 | $153 | 24 | 3.9x | — |
| May 19 | $14,413 | 104 | $139 | 30 | 3.5x | ↓ Improving |
The Purchase-to-Add-Payment ratio swings from 1.8x (May 5) to 10.3x (May 15) day to day. On May 14-15, the ratio hit 9.9x and 10.3x — meaning for every 1 Add Payment Info event, the pixel recorded nearly 10 purchases. This is NOT normal attribution delay. This confirms the Add Payment Info event is firing intermittently, not consistently broken. Some days it works (~1.8x normal), some days it barely fires at all (10x abnormal). This points to a race condition in the checkout flow — the event fires before the page loads on some browser/device combos but not others.
Yesterday: Only 39 Search events in the last 30 days. Flagged as essentially non-functional.
Today — WORSE: The last 14 days show ZERO Search events. The last 7 days also show ZERO. The Search pixel event has completely stopped firing around May 6. This is a new deterioration — it was at least trickling before.
| Window | Search Events | Status |
|---|---|---|
| 90 Days | 55 | Minimal |
| 30 Days | 39 | Minimal |
| 14 Days | 0 | DEAD |
| 7 Days | 0 | DEAD |
The Search event was already barely functioning, but now it has stopped entirely. If Vegamour's site has a search bar, the pixel event for onsite search is no longer wired up. This means Meta has zero data on what products people are searching for — a valuable optimization signal lost.
Yesterday: All 5 purchase event types (purchase, web_in_store_purchase, omni_purchase, onsite_web_app_purchase, offsite_conversion.fb_pixel_purchase) showed identical counts. No duplication.
Today: Still identical across all types. 90-day: all showing 8,164. 7-day: all showing 442. Purchase deduplication is working correctly. This is the one thing that's clean.
Yesterday: Ratio growing from 2.31x (Feb-Mar) to 2.80x (Apr-May). The payment tracking gap is getting wider.
Today: The structural gap remains. But the daily data reveals something new — the ratio isn't just growing steadily, it's wildly volatile. It ranges from 1.8x on good days to 10.3x on bad days. This suggests the issue is intermittent, likely tied to specific browser/device/page load conditions rather than a permanent misconfiguration.
| Window | Ratio (Yesterday) | Ratio (Today) | Change |
|---|---|---|---|
| Feb – Mar | 2.31x | 2.31x | same |
| Mar – Apr | 2.22x | 2.22x | same |
| Apr – May | 2.80x | 2.80x | same |
| 14 Days | — | 2.45x | slight improve |
| 7 Days | — | 2.58x | middle range |
Campaign: "LSG | Prospecting Retargeting | Second Chance ASC+ | Conversion-Purchase | CBO | ROAS Goal 0.8"
Budget: $1,500/day | Created: May 18 7:54 AM | Paused: May 18 10:24 AM (2.5 hours later)
Note: This is good — someone is testing. But the "ROAS Goal 0.8" in the campaign name is concerning. We flagged in yesterday's audit that a 0.8x ROAS target literally tells Meta that losing 20% on every dollar is acceptable. The fact that it was quickly paused suggests they recognized the issue.
May 19 Performance (before pause took full effect): $1,452 spent / 4 purchases / $363 CPA / 1 Add Payment event (4:1 ratio confirms tracking issues persist on new campaigns too)
As of today, only 2-3 campaigns are actively spending:
Most NRM (previous agency) campaigns have been paused. The LSG campaigns appear to be from a new team managing the account. The transition from NRM → LSG naming convention happened around early May.
| Issue | Yesterday | Today | Status |
|---|---|---|---|
| #1 Inverted Funnel | 2.80x ratio | 2.80x ratio | Not Fixed |
| #2 ViewContent 2x | +98% jump | Held at +110% | Stabilized |
| #3 CPA Anomaly | 45% drop suspicious | Stable at $128-133 | Stabilized |
| #4 May 8-9 Break | 65-70% confident | Oscillation, not clean break | Revised |
| #5 Search Events | 39 in 30 days | 0 in 14 days | Got Worse |
| #6 No Duplication | Clean | Clean | Still Clean ✓ |
| #7 Payment Ratio | 2.31x → 2.80x | 1.8x–10.3x daily swing | Not Fixed |
Look at Event Match Quality score. If below 6/10, CAPI is broken or misconfigured. This is the single fastest diagnostic.
Pull Shopify daily order count for May 1-19. Compare against Meta's daily purchase count. If they diverge, we know exactly what Meta is over- or under-reporting.
If server-side events stopped sending, that alone explains the wild daily variance in Add Payment Info (works when browser pixel fires, fails when CAPI should cover).
The Search pixel event has completely died. Check if the on-site search bar still triggers the Search event. Likely needs to be re-implemented.
The 1.8x to 10.3x daily swing means this event fires intermittently. Check if it's a page load race condition on the checkout page. The event script may fire before Shopify's checkout page fully renders on slower connections.
Check Shopify app installs, theme changes, and pixel configuration changes around March 18-22. Something caused ViewContent to double overnight.