1. Attribution

Can you check where the today’s purchases are coming?

<aside> 💡

Yiqi’s thoughts: what I find impressive is in the last paragraph, where AI is pointing out that there are 12 orders that clicked on a Meta ad in the last 90 days but they were attributed because Meta only uses 7d click attribution window. This really helps understand long-term impact of Meta ads.

</aside>

Here's the click/UTM attribution**:** 108 orders, $56,453.22 total. Of those:

Bucket (first-touch from landing PageView in 7d)   Orders   Share
─────────────────────────────────────────────────  ──────   ─────
utm_source=facebook  (paid)                          15     13.9%
utm_source=google    (cpc / product_sync)            11     10.2%
utm_source=klaviyo   (email flow / campaign)          9      8.3%
utm_source=ig        (paid social)                    4      3.7%
click_id only — fbclid (no utm)                       4      3.7%
click_id only — gclid (no utm)                        1      0.9%
direct / no signal in 7d landing PV                  64     59.3%
─────────────────────────────────────────────────  ──────   ─────
TOTAL                                               108    100.0%

Paid traffic that did convert — Meta is the largest source: 23 orders total (15 utm=facebook + 4 utm=ig + 4 fbclid-only). Google adds 12 (11 utm=google + 1 gclid-only). Klaviyo flows + campaigns drove 9.Top campaigns (utm_campaign on first-touch URL):

2025-p-max-google-c                           8   Google P-Max (SE)
1***************27                            6   Meta campaign ID
1***************28                            5   Meta campaign ID
1***************28                            3   Meta campaign ID
sag_organic                                   3   Organic social
email #1 - abcdef request (Klaviyo)           3
abandonedment cart                            2   Klaviyo flow

Caveats on the 64 "direct/no signal" orders:• 29 of them still have _user_data._fbc set on the Purchase CAPI payload, meaning that user clicked a Meta ad at some point in the last 90 days (just not within our 7-day landing PageView lookback). So Meta's true share is probably closer to 23 + ~12 = ~35 orders if we extend the window.

Can you pull purchases from 05/25 - 06/07 that comes from Meta ads campaign 12024672290253 and 12024591100253**?**

<aside> 💡

Yiqi’s thoughts: A lot of the brands don't know why Meta has certain number of attributions. You can just ask AI, and this solves all the problems.

</aside>

Meta Purchase attribution, 05/25 → 06/07 (fbp-matched from Athena PageView landings → CloudWatch CAPI Purchase events):

Campaign            Orders   Total
──────────────────  ──────   ────────
12024672****290253       9   $  756.54
12024591****100253      21   $ 1489.97
──────────────────  ──────   ────────
Combined                30   $ 2246.51

Order-level breakdown: Campaign 12024672****290253 — 9 orders:

Date (UTC)           Order ID              Value  Email
2026-06-04 07:53:19  8391857144108    $168.95 USD  s**********@gmile.com
2026-06-04 20:26:40  8392947761452    $43.95 USD  q*******@yahoo.com
2026-06-05 02:01:41  8393438495020    $96.95 USD  r*********@gmail.com
2026-06-05 05:19:45  8393684517164    $48.95 USD  v*********@gmail.com
2026-06-05 20:56:18  8394937925932     $47.9 USD  t*******@gmail.com
2026-06-06 18:11:03  8396422021420    $107.94 USD  a****@sonlightpower.org
2026-06-07 04:54:29  8397238763820       $78 USD  r***********@gmail.com
2026-06-07 05:11:24  8397254132012    $81.95 USD  s**********@live.com
2026-06-07 16:01:45  8397877149996    $81.95 USD  g********@gmail.com

Campaign 12024591****100253 — 21 orders:

Date (UTC)           Order ID              Value  Email
2026-05-27 12:21:35  8378109460780      $165 USD  j***********@live.com
2026-05-28 15:14:46  8380012036396       $78 USD  t*********@yahoo.co
2026-05-29 08:38:49  8381270294828    $87.95 USD  p*****@aol.com
2026-05-29 10:49:09  8381385441580       $78 USD  t******@hotmail.com
2026-05-29 13:27:51  8381556326700    $43.95 USD  j********@yahoo.com
2026-05-30 04:39:35  8382775525676    $81.95 USD  f**********@yahoo.com
2026-05-31 18:32:05  8385385857324    $146.93 USD  s*****************@gmail.com
2026-05-31 21:34:54  8385651769644     $47.9 USD  t***********@aol.com
2026-06-01 01:19:55  8385974337836     $9.95 USD  r*********@aol.com
2026-06-01 02:13:36  8386036957484    $47.95 USD  d*****@cox.net
2026-06-01 12:13:58  8386633302316       $78 USD  t**********@yahoo.com
2026-06-01 16:30:52  8386995454252    $47.95 USD  k*************@yahoo.com
2026-06-02 20:27:53  8389282365740       $78 USD  d********@gmail.com
2026-06-04 07:56:11  8391862092076    $107.94 USD  p************@yahoo.com
2026-06-04 12:23:45  8392108933420    $43.95 USD  a****@bellsouth.net
2026-06-04 19:45:57  8392887927084    $43.95 USD  w*******@yahoo.com
2026-06-04 23:22:46  8393193521452     $9.95 USD  s**************@gmail.com
2026-06-05 01:50:19  8393414508844    $81.95 USD  m********@gmail.com
2026-06-05 05:19:44  8393684418860    $50.75 USD  v*********@aol.com
2026-06-05 23:03:16  8395118182700    $81.95 USD  x******@yahoo.co
2026-06-06 00:59:43  8395245519148       $78 USD  f**********@aol.com

2. Debugging

My conversion rate is down about 50% compared to last month. Can you take a look?

<aside> 💡

Yiqi’s thoughts: This is a very interesting use case. This brand has a conversion rate drop, and AI is able to solve that problem within like 10 minutes. They discovered it's a payments issue in the US.

</aside>

Funnel for past 7d
```
PV                      36,874
VC                       4,054   11.0% of PV
ATC                      4,560
IC                       1,545   33.9% of ATC
AddPaymentInfo             152    9.8% of IC  ← the leak
Purchase                    66   43.4% of APIcap, 4.3% of IC
```
Where the leak is: InitiateCheckout → AddPaymentInfo

The IC→API cap conversion is 9.8% — a healthy e-commerce funnel is usually 50–70% at this step. That means ~90% of buyers who reach checkout never complete the payment step. That single drop accounts for almost all of the "completed checkout" decline.

By the time the funnel reaches AddPaymentInfo, the conversion to Purchase is healthy (~41%), so the issue is at the checkout page itself, not after payment.

Common causes for the IC→API cap drop, ordered by frequency
1. Shipping cost surprise — added shipping at checkout makes the order > customer's threshold
2. Payment-method problem — a popular method (Apple Pay, PayPal, Klarna, etc.) was disabled / errors silently
3. Currency / FX confusion — multi-currency stores sometimes show product prices in one currency and checkout in another
4. Coupon code failing — discount code doesn't apply at the payment step
5. Form / field validation bug — recent theme or checkout-extension change broke something silently
6. Geographic restriction — country selector blocks customers at the address step
```
country   IC      APIcap   Purchase   IC→APIcap   IC→Purchase   APIcap→Purchase
cz        178     21       9-31*       11.8%       5.1-17%       ~50-70%
sk        113     9        5-20*        8.0%       4.4-18%       ~50-70%
pl         97     9        3-8*         9.3%       3.1-8%        ~30-50%
hu         53     2        1-5*         3.8%       1.9-9%        ~50%
us         21     5        0           23.8%       0.0%          0% ⚠️
th          7     0        0            0.0%       0.0%           -
ro          4     0        0            0.0%       0.0%           -
ca/de/in/vn/pk/ie/es  each 1 IC, 0 APIcap, 0 Purchase
```
Note: the Purchase column has a range because the earlier pull was undercounted (5000-row CW limit). For example CZ has 9 events that came up in the 5000-row sample and 31 events in the full stats query. The IC→Purchase % range uses both bounds.

Read-out (with the corrected, fuller numbers)

Real per-country completion rates (using the full Purchase count = 66):
```
country   IC      Purchase   IC→Purchase
cz        178     31         17.4%  ← actually solid
sk        113     20         17.7%  ← solid
pl         97     8           8.2%  ← below average
hu         53     5           9.4%  ← below average
us        21      0           0.0%  ⚠️ broken
others    ~10     2          ~20%   (small N)
```
That's actually a much healthier picture than I painted earlier. CZ and SK are around 17–18% completion (industry benchmark range), so the home market funnel is mostly healthy. The weak spots are:
1. US: 21 IC → 5 APIcap → 0 Purchase. Reaches payment but never converts. US-specific payment-method issue.
2. PL: 8.2%, HU: 9.4% — below CZ/SK by half. Worth comparing checkout language, payment options, shipping cost vs CZ/SK.
3. Small markets (RO, TH, etc.) starting checkout but never paying — shipping zones likely excluding them.
Is there any traffic from Pinterest in the last 7 day?

<aside> 💡

Yiqi’s thoughts: So this brand is not sure that if Pinterest is actually working. This data analytics tells you exactly how much and what is driven by Pinterest.

</aside>

Headline: 631 Meta CAPI events (336 PageView + 295 ViewContent) with Pinterest signal, ~237 distinct users. 427 of them came from inside the Pinterest mobile app itself (UA Pinterest/iOS / Pinterest/Android).Daily volume (ramping up):
```
2026-06-05   19
2026-06-06   92
2026-06-07  220
2026-06-08  300
```
(Days before 06-05 had zero Pinterest-tagged events.)Source signals detected on the landing URLs:• utm_source=Pinterest&utm_medium=PaidSocial&utm_campaign=pall|pcata• Pinterest enhanced-match epik= query param• In-app Pinterest browser user-agentSample landing URL:
```
<https://comfy.sg/products/etna-small-vanity-table?utm_source=Pinterest&utm_medium=PaidSocial>
&utm_campaign=pall&utm_content=2680090595161&epik=dj0yJnU9MlhlMHRm…
```
Funnel so far (Pinterest-attributed only):• PageView: 336• ViewContent: 295• AddToCart: 3• Purchase: 0So Pinterest is driving real upper-funnel traffic but it hasn't converted yet in this 7d window.

My revenue is down ~30% this week. Can you figure out why?

<aside> 💡

Yiqi's thoughts: This is the question every brand asks and the hardest one to answer. Normally an analyst pulls funnel data, then country data, then device data, then payment data, days of work. The AI slices every dimension at once and lands on the root cause in minutes.

</aside>

Headline: Revenue this week $38,420 vs $54,890 last week (-30.0%). Orders 412 → 296. AOV is flat ($133 vs $130), so this is a conversion problem, not a basket-size problem.

Step 1 — Funnel comparison (this week vs last week)

Step               Last wk    This wk    Δ        Step conv (last → this)
─────────────────  ────────   ────────   ──────   ───────────────────────
PageView            41,200     40,850    -0.8%
ViewContent          5,140      5,020    -2.3%    12.5% → 12.3%
AddToCart            3,890      3,760    -3.3%
InitiateCheckout     1,310      1,250    -4.6%    33.7% → 33.2%
AddPaymentInfo         790        510   -35.4%    60.3% → 40.8%  ← the leak
Purchase               412        296   -28.2%    52.2% → 58.0%

Traffic and upper funnel are flat. The drop is concentrated at InitiateCheckout → AddPaymentInfo.

Step 2 — Segment the leak

By country:

Country   IC last wk → this wk   IC→APIcap last → this
US               680 → 655            64% → 38%  ⚠️
CA               195 → 188            58% → 55%
UK               160 → 152            61% → 59%
AU                95 →  90            57% → 56%

The leak is US-only. Other markets are unchanged.

By device (US only):

Device      IC→APIcap last → this
iOS mobile        66% → 21%  ⚠️
Android           60% → 58%
Desktop           63% → 61%

By payment method (US iOS checkouts):

Method       Share of completed payments, last → this
Apple Pay              48% → 3%  ⚠️
Shop Pay               24% → 41%
Credit card            22% → 46%
PayPal                  6% → 10%

Root cause: Apple Pay on US iOS checkouts effectively stopped converting on Tuesday. Some buyers fell back to card/Shop Pay, but most abandoned. Checking your Shopify payment settings, the Apple Pay domain verification expired after the theme update on 06/02 — Apple Pay button renders but the payment sheet fails silently.

Estimated recovery: restoring US iOS payment completion to its prior rate ≈ +105 orders / +$13,900 per week.

3. Data Analytics

check the all the data sent to Meta in the last 24 hours

Meta event           Total      Hourly distribution (UTC)
PageView              47        25/18:8, 25/19:3, 25/20:2, 25/22:2, 25/23:4,
                                  26/00:6, 26/03:4, 26/08:10, 26/10:2, 26/13:4, 26/16:2
ViewContent            2        26/00:2 only

That's it — no AddToCart, no InitiateCheckout, no Purchase, no Lead. Just 47 PageView + 2 ViewContent, all from the browser pixel side.

Read on this:

Shop is 48 minutes shy of 24h old — virtually all the data we'd expect is just PV/VC from someone (probably the merchant or a tester) clicking around. Hour 26/08 had a 10-PageView burst that's likely the merchant doing a tour; the 2 ViewContent at 26/00 are 2 distinct PDP visits.
Pixel 1626615351740654 is verified and CAPI is wired up correctly — events that did fire dispatched cleanly.
Nothing's broken. If you were expecting more event types, the shop just doesn't have organic traffic yet.

Beta Testing

We are opening up 50 Beta Testing slots to our existing Aimerce customers. We will give each of you 500 credits (~50 back&forth conversations using Claude model API), in exchanges for feedback, use cases and bug reports.