Snapshot CRO
// behavioral analytics + a/b testing for Shopify. one app, one workflow, zero session recordings of people typing their address.
Most CRO stacks are three subscriptions duct-taped together: an analytics tool that tells you something is wrong, a heatmap tool that shows you where it might be wrong, and a testing tool that lets you guess at a fix. Snapshot CRO does the whole loop in one place: diagnose problem pages → uncover roadblocks and friction → a/b test solutions. Each step feeds the next, so your tests come from evidence instead of vibes.
- runs on Shopify's web pixel. no theme code edits, no render-blocking script tag.
- counts and aggregates only. no session replays, no keystroke logging, no stored customer PII. (see the privacy policy — it's short because we collect little.)
- built for real stores with real traffic numbers, i.e. not 500k sessions/month. snapshot methodology gets low-traffic stores to usable answers.
tl;dr
// what this page explains, so you know what you're getting into:
- step 0 — every session is scored real / zombie / bot, and the fake ~third of your traffic is excluded before analysis.
- diagnose problem pages — the store audit ranks pages by traffic, conversion & revenue and flags the ones leaking money.
- uncover roadblocks & friction — page snapshots count clicks by zone, exit paths & on-site searches, so the test hypothesis writes itself.
- a/b test solutions — bayesian results with revenue tracking, accurate even on low-traffic stores.
- pricing & install — free tier exists; paid tiers at $49.99 and $99.99/mo.
step 0: filter the fake traffic
Before any diagnosis, every session gets scored: real, zombie, or bot. Bots are datacenter and automation traffic. Zombies are "sessions" that load one page, touch nothing, and leave — technically human-shaped, analytically useless. Your platform analytics counts all of them as visitors, which is one reason your conversion rate looks worse than it is and your tests never reach significance.
| class | sessions | share | |
|---|---|---|---|
| real | 8,412 | 65.5% | ████████████████████ |
| zombie | 2,905 | 22.6% | ████████████████████ |
| bot | 1,530 | 11.9% | ████████████████████ |
→ 34.5% of "traffic" excluded from analysis. you're welcome.
^ recreation of the app's session scoring view, sample data.
1. diagnose problem pages
The Store Audit ranks every page by traffic, conversion, and revenue, then flags the mismatches — pages that get plenty of visitors and quietly do nothing with them. That delta between "should convert" and "does convert" is your leak. You don't need to audit 200 pages; you need to audit the three that are bleeding.
| page | sessions | cvr | revenue | leak |
|---|---|---|---|---|
| /products/trail-runner-v2 | 3,204 | 0.8% | $4,110 | -$9,340 |
| /collections/sale | 2,118 | 1.1% | $3,890 | -$5,120 |
| /products/daily-trainer | 1,867 | 3.4% | $11,240 | — |
| /collections/new | 1,411 | 2.9% | $7,205 | — |
→ leak = expected revenue at store-median cvr minus actual. two pages, $14.5k/mo on the table.
^ recreation of the store audit view, sample data.
2. uncover roadblocks and friction
Take a Page Snapshot of a flagged page and the app audits what actually happens on it: clicks by zone, exit paths, scroll depth, and what people type into search when the page fails them. No pixel-blob heatmap you squint at like a horoscope — counted interactions per zone, ranked.
| zone | clicks/session | signal | |
|---|---|---|---|
| size selector | 4.7 | ████████████ | rage territory |
| image gallery | 2.1 | ████████████ | normal |
| size chart link | 1.8 | ████████████ | opens pdf. on mobile. in 2026. |
| add to cart | 0.3 | ████████████ | the point of the page |
top exit path: size chart pdf → gone (41%)
top on-site search from this page: "trail runner sizing" (188×)
→ hypothesis writes itself: sizing info is the roadblock. put it on the page.
^ recreation of the page snapshot view, sample data.
3. a/b test solutions
Now test the fix — in the same app, against the same scored traffic. Results are Bayesian: instead of a p-value you explain to no one, you get the probability that the variant beats control, plus revenue per session so a "winner" that sells more $12 socks and fewer $140 shoes gets caught. Snapshot-based testing means low-traffic stores get honest answers in weeks, not quarters.
| variant | sessions | cvr | rev/session | p(beat control) |
|---|---|---|---|---|
| A — control (pdf link) | 1,506 | 0.9% | $1.28 | — |
| B — inline size chart | 1,494 | 2.2% | $3.04 | 97.4% |
→ ship B. close the loop. open the next leak from step 1.
^ recreation of the a/b test results view, sample data.
pricing
- free — 1 store snapshot, 3 page snapshots, 1,000 visitors/snapshot. enough to find your first leak.
- whisper, $49.99/mo — 5 store snapshots, 50 page snapshots, 5 a/b tests, 5,000 visitors.
- signal, $99.99/mo — 20 store snapshots, 250 page snapshots, 25 a/b tests, 25,000 visitors.
install
It's on the Shopify App Store. Installs via web pixel, uninstalls cleanly, and the data practices are documented in plain language in the privacy policy below.