One interactive home base for spotting anomalies in Shopify traffic. Each chart overlays total sessions (blue) with the zero-second / instant-bounce portion (amber) and the conversion rate that results (green). Hover any chart for the exact weekly numbers; click a legend label to isolate a line. Weekly, last two years. Every sessions series reconciles to Shopify's reported totals.
Mirror image of quality: as zero-second share crept up and the traffic mix shifted, blended CVR drifted from ~2.5% to ~1.8%.
Stable bands for two years: YouTube ~80%, Facebook/Instagram/Direct ~73%, Google ~57%. No channel shows an attack-style step-change.
Google (search) leads ~4%. Facebook CVR cratered to ~0.7% in Sep–Oct 2025 then recovered; Instagram is sliding toward ~0.6% now.
| Channel | Sessions (2yr) | Zero-sec % | last 4wk | CVR (wtd) | CVR last4 |
|---|---|---|---|---|---|
| 4,058,486 | 74% | 69% | 1.84% | 1.93% | |
| Direct | 9,801,209 | 73% | 69% | 2.52% | 1.78% |
| 1,467,199 | 73% | 75% | 1.87% | 0.89% | |
| Google (organic search) | 2,200,022 | 57% | 52% | 4.12% | 3.78% |
| YouTube | 635,772 | 80% | 74% | 1.06% | 1.17% |
"Direct" (9.7M) absorbs untagged + in-app-browser traffic, so its volume is inflated. TikTok excluded — referral collapsed to ~zero after Nov 2025.
collections/shop & the 30-servings PDP convert highest (~3.7–4.7%); the paid landers (rise-2, compare-listicle, rise-2-coffee) sit ~1.6–1.9% — and slipping.
| Landing page | Sessions (2yr) | Zero-sec % | last 4wk | CVR (wtd) | CVR last4 |
|---|---|---|---|---|---|
| /pages/rise-2 | 3,629,866 | 69% | 68% | 1.79% | 1.37% |
| /products/30-servings-tin | 2,120,256 | 73% | 70% | 3.75% | 3.49% |
| / (homepage) | 2,842,297 | 66% | 54% | 2.31% | 2.39% |
| /pages/compare-listicle-og | 360,313 | 72% | 77% | 1.89% | 1.60% |
| /collections/shop | 418,102 | 53% | 45% | 4.72% | 4.38% |
| /pages/rise-2-coffee | 117,844 | 75% | 79% | 1.63% | 0.51% |
| Landing page \ channel | YouTube | Direct | |||
|---|---|---|---|---|---|
| Rise 2 LP | 59% 169k | 65% 127k | 54% 35k | 68% 24k | 73% 76k |
| Compare listicle | 68% 69k | 78% 64k | 68% 23k | — | 82% 9k |
| 30-serving PDP | 81% 10k | 86% 3k | 51% 53k | 89% 6k | 82% 62k |
| Homepage | 50% 6k | 86% 4k | 41% 24k | — | 55% 82k |
| Shop collection | — | — | 31% 6k | — | 51% 19k |
| Rise 2 coffee LP | 70% 8k | 77% 5k | — | — | — |
| Landing page \ channel | YouTube | Direct | |||
|---|---|---|---|---|---|
| Rise 2 LP | 2.17% 169k | 1.38% 127k | 3.23% 35k | 1.46% 24k | 1.14% 76k |
| Compare listicle | 2.13% 69k | 1.10% 64k | 3.44% 23k | — | 1.45% 9k |
| 30-serving PDP | 1.07% 10k | 1.20% 3k | 5.51% 53k | 0.41% 6k | 2.12% 62k |
| Homepage | 2.66% 6k | 0.25% 4k | 2.77% 24k | — | 2.25% 82k |
| Shop collection | — | — | 4.92% 6k | — | 3.46% 19k |
| Rise 2 coffee LP | 1.37% 8k | 0.90% 5k | — | — | — |
Shopify ShopifyQL · sessions · landing_page_path × referrer_name · pulled via Cowork (in-session bridge was down). "Direct" = untagged / in-app-browser traffic. Empty cells = combination below the top-60 reporting cut (very low volume). Conversion rate is Shopify session-based.
| Relationship (weekly, 2 yrs) | Pearson r | Read | weeks |
|---|---|---|---|
| Instagram zero-second share → Instagram CVR | -0.60 | negative, moderate | 106 |
| Site zero-second share → site CVR | -0.31 | negative, weak | 106 |
| Instagram zero-second share → blended CAC | +0.22 | positive, weak | 106 |
| Site zero-second share → blended CAC | -0.15 | negative, negligible | 106 |
| Site CVR → blended CAC (reference) | -0.60 | negative, strong | 106 |
r ranges −1 to +1. Near 0 = no linear relationship; ±0.2 weak, ±0.4 moderate, ±0.6 strong. CAC = Daily Stand “Total Blended CAC (DTC)” = total spend ÷ Shopify new customers, aggregated to weeks. Rates plotted only where weekly sessions ≥ 200.
What you’re seeing: the cloud slopes clearly down and left-to-right — weeks when more of Instagram’s traffic bounced instantly are the same weeks IG converted worse. At r = −0.59 this is the strongest relationship on the page: the dots hug the trend line fairly tightly, so IG zero-second share is a genuine read on IG traffic quality. The low-bounce weeks (left) cluster near 2–3% CVR; the high-bounce weeks (right) sag toward ~1%.
What you’re seeing: same downward tilt, but looser — r = −0.31. The dots are more scattered around the line because site-wide bounce blends every channel together, which dilutes the signal any single source carries. The relationship is real but weak: zero-second share nudges sitewide CVR, it doesn’t dictate it.
What you’re seeing: a near-flat, faintly-rising cloud — r = +0.20. This is the chart that tests your hunch directly, and the answer is “barely.” The dots are a shapeless blob: plenty of high-IG-bounce weeks have perfectly normal CAC, and plenty of low-bounce weeks have high CAC. If IG junk traffic were driving CAC, this would slope up hard; it doesn’t.
What you’re seeing: no relationship — r = −0.13, essentially a flat trend line through random scatter. Site-wide bounce share tells you nothing useful about what CAC will be that week. CAC is set by spend levels and channel mix, not by how many sessions bounced.
What you’re seeing: the same scatter story, on a timeline. The red IG-bounce line wanders in a fairly tight band the whole period, while the black CAC line climbs through 2025 — they’re not tracking each other. Eyeball the big CAC swings (e.g. the spring-2025 run-up) and you’ll see IG bounce didn’t move with them. That’s the +0.20 correlation made visible: a weak, mostly-coincidental drift, not a lockstep relationship.
Analyst caveat: blended CAC is driven mostly by paid-spend levels, channel mix and seasonality; Instagram’s zero-second share is one channel’s on-site behavior. The weak positive link likely reflects a shared cause (scaling cold prospecting brings both more instant-bounce IG clicks and higher CAC) rather than zero-second sessions causing CAC. Correlation ≠ causation.
| Week | Spend | CTR | CPM | Ad sets spending |
|---|---|---|---|---|
| Jun 1–7 | $114,396 | 1.47% | $27 | 33 |
| Jun 8–14 | $114,157 | 1.52% | $33 | 24 |
| Jun 15–21 | $129,341 | 1.62% | $36 | 30 |
| Jun 22–28 | $122,362 | 1.51% | $38 | 27 |
| Jun 29 – Jul 5 | $103,875 | 2.08% | $48 | 18 |
What you're seeing: CTR 1.47% → 2.08% (+41%) and CPM $27 → $48 (+77%) in five weeks, while the number of ad sets actually taking spend fell from 33 to 18. Spend concentrated into the proven listicle ad set (steady ~$57k/wk) and two rise-2 broad ad sets whose CTR jumped. Fewer, hotter pockets of delivery at higher auction prices — the signature of Meta narrowing who it shows the ads to, not of a creative or settings change.
| Metric (account-level) | Week of May 4 | June avg | Week of Jun 29 – Jul 5 |
|---|---|---|---|
| Meta CTR (all) | 2.01% | 1.57% | 1.72% |
| Meta link CTR | 1.21% | 0.89% | 1.04% |
| Meta CPM | $48 | $36 | $43 |
| Blended CAC (Daily Stand) | $92 | $130 | $107 |
CAC in the Jul 4 week is partly the sale itself (discount pulls conversions forward), so don't read the whole CAC drop as delivery efficiency. But the CTR/CPM shift began Jun 26–28 — before the sale launched — matching the delivery-concentration read.
What you're seeing: the two lines mirror each other more than they track each other — CTR peaks (early 2025, May 2026) line up with the lower zero-second stretches. If rising CTR meant junk clicks, amber would climb with purple; it does the opposite.
What you're seeing: over two years the weekly correlation is ~0 (r = -0.08) because spend levels, mix and seasonality dominate. But zoom into 2026: the May high-CTR stretch was the $92–$108 CAC stretch, June's CTR slump was the $128–$132 stretch, and the current CTR recovery arrived with CAC back near $100. Within a stable spend regime, the CTR line has been a decent leading read on efficiency this year — worth watching, not worth betting the budget on yet.
What you're seeing: a clear downward slope. High-CTR weeks (right side) cluster at the low end of zero-second share. This is the single most direct answer to "are the extra clicks junk?" available in your data: historically, no — high-CTR periods send cleaner traffic.
What you're seeing: the same pattern, slightly stronger. Expensive-auction weeks are clean-traffic weeks — you're paying more per impression precisely when Meta is buying better humans. Cheap reach (left side) is where instant-bounce share creeps up.
| Channel | 0-sec % · Jun 22–28 | 0-sec % · Jun 29 – Jul 5 | prior 4wk | 2-yr norm | CVR · Jul 4 wk | CVR prior 4wk |
|---|---|---|---|---|---|---|
| 71.3% | 70.6% | 65.5% | ~73% | 2.27% | 1.86% | |
| 78.7% | 77.8% | 70.4% | ~73% | 1.15% | 1.06% | |
| Google (ref.) | 49.9% | 49.9% | 53.1% | ~57% | 4.81% | 3.57% |
| FB + IG combined | 75.7% | 74.0% | 67.7% | 73.6% | — | — |
| Site-wide (all traffic) | 68.9% | 68.0% | 65.1% | ~71% | 2.41% | 2.02% |
Verdict: no junk-click signature. FB+IG zero-second share in the anomaly window (74–76%) sits right on the two-year norm (73.6%) — elevated only against June's unusually clean lows, and nowhere near an attack pattern (bot floods print +15–30pp step-changes; Instagram's 78% is ~3pp above its own 12-week high). More decisive: the traffic converted better, not worse — Facebook CVR 2.27% vs 1.86% baseline, Instagram 1.15% vs 1.06% — and the click→session yield ran above normal (118–134 sessions per 100 link clicks). Junk clicks don't buy. The sale discount helped CVR, but bought traffic that bounces normally and buys more is the opposite of click inflation. Combined with the two-year negative CTR↔zero-sec relationship above, the clean read stands: Meta concentrated delivery on fewer, higher-intent pockets, and the efficiency is real.
Analyst caveats: (1) account-level weekly CTR is not the same cut as the per-ad CTR→CAC analysis (creative-cac-correlation), which found per-ad CTR strongly predicts per-ad CAC — both can be true. (2) The Jul 4-week CAC drop is confounded by the sale discount; the pre-sale CTR/CPM shift (Jun 26–28) is the part that's genuinely unexplained-but-benign on current evidence. (3) Correlation ≠ causation throughout.