Your devs flagged 35% fewer sales and a steep drop in store traffic. We compared their store numbers against every Meta and Google ad from August 31 to September 27.
Orders fell from 104 to 71 across the three weeks, about 30% fewer. That matches what your devs saw.
Weekly ad spend moved from about $7,300 to $6,200. Nothing was switched off except one small retargeting campaign on Sep 9.
All our ads send 5,000 to 6,000 visitors a week. Direct traffic fell by about 52,000. The gap is ten times bigger than every ad click combined.
Same scale, same weeks. The purple bars are your store's sessions. The colored bars are every click from Meta and Google ads.
Even if every ad click were counted as "direct," ads could explain at most about 6,000 sessions a week. Paid clicks fell by about 1,300 over these weeks. Direct sessions fell by about 52,600.
A store that converts 0.09% of visitors is almost always counting bots, crawlers, or junk referrals. Fine jewelry stores usually convert between 0.5% and 1.5%. When that traffic stops, sessions fall and the conversion rate climbs, with no change in real shoppers.
This is our read of the numbers, not yet confirmed. A sessions export by landing page, country, and device will settle it (see the ask below).
Weekly spend on Meta and Google. The Meta mix shifted toward your own photo and video content, which has been the strongest performer since August.
Sales each channel reported, by week. Meta returned 6.8x and 7.3x over the last two weeks, close to the 7x bar. The drop sits in Google: the shopping campaign and people searching "Automic Gold" by name.
| Ad-reported sales | Aug 31–Sep 6 | Sep 7–13 | Sep 14–20 | Sep 21–27 |
|---|---|---|---|---|
| Meta ads | $12,662 | $38,392 | $20,573 | $22,846 |
| Google shopping | $19,735 | $18,422 | $10,662 | $7,955 |
| Google brand search | $12,313 | $15,898 | $12,212 | $6,416 |
| Google other search | $0 | $624 | $849 | $783 |
| Brand search clicks | 711 | 684 | 681 | 791 |
More people clicked on brand search last week, but fewer of them bought. That points to softer demand from people who already know Automic Gold, which ads don't create. Meta and Google each count sales by their own rules, so some orders appear in both rows.
A Shopify sessions export for the last four weeks, broken out by landing page, country, and device. Or add us as a store collaborator and we'll pull it ourselves.