CollectionsIQ · now on the Shopify App Store
Merchandise what actually keeps the money.
CollectionsIQ builds and maintains Shopify collections from your real performance — return-adjusted net ROAS, custom metrics, per-platform and blended ad data — and keeps them current every day. Promote the products that earn after refunds and ad spend; quietly retire the ones that only look like winners.
Read-only on Google & Meta · daily auto-sync · no customer data stored.
What gets clicked isn't what keeps you in profit.
Same product, two truths. Switch between what your ad dashboard reports and what CollectionsIQ actually sees after returns.
High CTR, lots of “purchases”, a strong gross ROAS. It looks like a hero — so it gets pinned to your homepage and shopping feeds. 38% of units come back. After refunds, its kept revenue barely clears ad spend. CollectionsIQ quietly demotes it and promotes the steady 3.1× net performer instead.
How: CollectionsIQ reads your Shopify orders & refunds into a dual-date ledger, classifies genuine product returns (not shipping or pre-fulfillment edits), and derives net ROAS two ways — settled (refunds that have actually landed) and predicted (a forecast return rate from matured sales cohorts). Both feed the rule engine. See how it works →
Every signal that matters, in one place.
Ad performance
ROAS, CPM, CTR, CPC, CVR — per platform or blended across Google & Meta. Computed from summed components, never averaged ratios.
Shopify sales
Units, revenue, orders, AOV — and refunds & returned items as first-class data, straight from your store.
Return-adjusted
True ROAS, net ROAS (settled & predicted), return rate, margin after ad spend. Performance after the money actually stays.
Your own metrics
Define custom metrics with a formula over the raw components — e.g. (gross_sales − discounts − refunds − spend) / spend — and use them in rules like any built-in.
Real merchandising & performance plays.
Click through a few of the ways stores use CollectionsIQ. Every rule previews the exact products it selects before you publish.
Scale the products that earn after returns
Stop rewarding gross-ROAS mirages. Promote only products whose kept revenue clears a healthy multiple of ad spend — a homepage and “Best of” feed that actually compounds margin.
AND min ad spend ≥ $50
→ “Top Performers” fills with true earners, refreshed daily.
Pull the serial returners out of prime placement
High-return products quietly tank blended ROAS and clog reverse logistics. Build a “Needs review” collection — or just exclude them from hero feeds — the moment their predicted net return turns unprofitable.
AND units sold ≥ 25 (enough to trust the rate)
→ Margin-leakers stop getting free homepage real estate.
Promote fast, demote honestly — without flapping
Add winners quickly on a forecast return rate, but only remove them on settled reality — and require several runs of agreement before any change. Membership stays stable, so your feed doesn't churn and reset ad-platform learning.
Net ROAS predicted > 2.0
// retain (honest)
Net ROAS settled > 1.4
add after 3 runs · remove after 5
→ Collections that move with the data, not the noise.
Proven sellers that are also cheap to advertise
Mix sales and ad signals with nested AND/OR groups. Curate a collection of products that are either bestsellers or efficient to reach — but always profitable — and never anything tagged clearance.
AND ( top 20 by units (30d) OR blended CPM < $10 )
AND NOT in collection “Clearance”
→ Profitable demand, automatically curated.
Credit the click, not the cart-riders
Dynamic ads often get credit for whatever else lands in the basket. Score a product on the returns and revenue of its own ad clicks — “ad-subject only” — so a hero isn't propped up by co-purchased items it didn't earn.
scope: ad-subject only
→ Honest per-product performance, basket-riders excluded.
Spot where a product really works
Blended numbers hide platform skew. Require a product to clear a ROAS bar on every platform it runs on, or build platform-specific collections to feed channel-specific campaigns.
AND min spend ≥ $50 per platform
→ “Reliable on Meta & Google” vs “Google-only winners”.
Works with any feed management app that supports smart collections.
CollectionsIQ outputs standard Shopify collections — so they slot straight into the feed tool you already use. Any feed app that can target or split by collection (DataFeedWatch, Feedonomics, Simprosys, Sales & Orders, and the like) can build performance-segmented feeds from them: a "Top net-ROAS" feed for your scaling campaigns, an "Underperformers" exclusion, a per-platform split — all kept current by CollectionsIQ's daily sync.
- No new feed tool to learn — point your existing app at the collection.
- Performance segments (net ROAS, returns, bestsellers) become feed segments.
- Membership updates daily; your feeds follow automatically.
The flow
Performance → collection → feed → campaign.
→ your feed app splits by that collection
→ Google / Meta campaign targets the split
→ your best products get the budget.
An honest, auditable rule engine.
Define any metric as a formula over raw components; use it in rules like a built-in. Live-validated, division-guarded.
Gross, net-settled (real refunds) or net-predicted (forecast) — per condition, on the ROAS metric.
Arbitrary boolean groups, so complex merchandising logic reads exactly as you'd say it out loud.
Dwell-in / dwell-out streaks keep membership stable — no daily flapping, no wasted ad re-learning.
Full-basket or ad-subject-only, so a product is judged on what its own ads actually sell.
Minimum spend / impressions / units, and matured-cohort-only return rates — a fluke can't sneak in.
Blend Google + Meta or require each to pass; compute ratios from summed components, never averaged.
Live and demo data are kept strictly separate — explore safely, never mix synthetic numbers into real rules.
Aggregate per-product numbers + encrypted tokens only. No customer personal data is stored.
One click to start, customize anytime.
Products returning ≥ 4× across platforms — your scale-up shortlist.
Blended CPM under $20 across Meta + Google — efficient reach.
Spending $100+ but returning under 1× — pull from prime placement.
High CTR and cheap clicks over 14 days — momentum to ride.
ROAS ≥ 2× on every platform it runs on — reliable sellers.
Top sellers by units, scoped to any collection, type, or tag.
Set it up in minutes.
1. Connect
Install on Shopify; connect Google & Meta (read-only). Orders & refunds flow in for net-ROAS.
2. Pick a rule
Built-in or custom — and preview the exact products it selects before you commit.
3. Publish
CollectionsIQ creates the collection and publishes it to your storefront (and your feeds).
4. Stays current
A durable daily pipeline re-evaluates and reconciles membership. Hands-off.
Priced on active collections. Start free.
Free
- 2 collections
- Up to 10 products each
- Daily sync
Starter
- 8 collections
- Unlimited products
- Google + Meta + sales rules
Growth
- 25 collections
- Custom metrics + net ROAS
- Nested rules & hysteresis
Pro
- Unlimited collections
- Priority support
- Everything in Growth
Common questions
What is "net ROAS" and how is it different from gross?
Gross ROAS divides ad-reported conversion value by spend. Net ROAS subtracts returns: settled uses refunds that have actually landed (honest, lagging — used to demote), while predicted discounts revenue by a forecast return rate derived from matured sales cohorts (fast — used to promote). Merchandising on net ROAS surfaces the products that keep the money, not just the ones that get clicked.
Does it modify my Google or Meta campaigns?
No. CollectionsIQ has read-only access to your ad accounts. It reads performance to decide merchandising; it never changes campaigns, budgets, or bids.
Does it work with my feed management app?
Yes. CollectionsIQ produces standard Shopify collections, so any feed tool that can target or split by collection can build performance-segmented feeds from them — no new feed app required.
How are blended ROAS / CPM calculated?
From summed base components across the platforms you choose — e.g. blended CPM = total spend ÷ total impressions × 1000. We never average per-platform ratios, which would distort the result.
Won't my collections flip around every day?
Only if you want them to. Turn on hysteresis and a product must pass for several runs before it's added and fail for several before it's removed — so membership stays stable and your feeds don't churn.
What data do you store?
Aggregate per-product metrics, a refunds ledger (no customer identities), and encrypted access tokens — no customer personal data. See our privacy policy.