Recommendations
On-site recommendations
with a holdout that keeps the number honest.
Three cards on a product page, an add-to-cart in the card, and a reorder strip for the customer who is running low. Every rule is previewed through the real engine before it is saved, and every result is followed to the till.
Questions this answers
- Do the recommendations actually sell, or just get clicked?
- Which rule should show on which page?
- Can it use the discounts I already run?
- What did the widget earn, against a holdout?
The widget
Three cards, an add-to-cart, and a clock.
Recommendations drawn from browse and buy behaviour, with the product’s real stars and buyer counts, a customer quote, and an add-to-cart that lets the shopper pick the size in the card, never sending them back to a product page. The promise is a clock band that counts down; the reorder pick is a Running low? strip.
- Real tabs, one true proof pill per card
- Pick the size in the card
- A Running low? strip with its own Reorder button
- Theme matrix: it fits the store it lands in
Omega-3 1000 mg
₹1,499BOUGHTMagnesium + B6
₹1,299NOT BOUGHTD3 + K2
₹1,899BOUGHTAshwagandha KSM-66
₹1,699NOT BOUGHTCollagen + hyaluronic
₹2,249NOT BOUGHTEight rules
Each one previewed through the real engine before you save it.
Free-shipping gap, pair and code offers, the reorder rung, anchor-named proof, and more: eight rules, one page per rule, each with a live preview of what the engine would actually show. The section opens like Measurement does, with a live count of rules that are on.
- Free-shipping gap, pair offers, code offers, reorder rung
- Anchor-named proof
- N of 8 rules live, one page per rule
- Previewed through the real engine, never a mock
brand search
₹33,480 spent · ₹1,44,720 back
cheapest growth
retargeting · DPA
₹37,350 spent · ₹1,43,460 back
well past the line
magnesium search
₹55,890 spent · ₹1,37,430 back
profitable
pmax · bestsellers
₹1,13,130 spent · ₹2,33,280 back
just above it
prospecting · advantage
₹1,62,180 spent · ₹1,57,320 back
under break-even
generic supplements
₹1,46,610 spent · ₹39,420 back
loses every order
Your discounts
Learned from orders, imported from the export, suggested as rules.
No Shopify token needed. Discounts are observed from the orders that used them and imported from the real Shopify discounts export, 36,000 rows in under a second, then suggested as rules where the data says they would sell.
- Observed from orders
- The real export, imported in under a second
- Suggested as rules
- Categories as reported, so a product type can count as another category
What came in
the shopper typed it
phone
from checkout
address
city, zip, country
click id
captured on landing
What we sent
em
a1f4…9c02
ph
7be1…44d8
ct / zp / country
3d90…b7e1
fbc / gclid
passed through raw
external_id
stable per customer
Match quality
0/10
Followed to the till
What was bought, what it was worth, against the holdout.
A share of visitors never see the widget. The stats page on Commerce shows what the rest bought and what it was worth, beside the holdout, so the number the widget claims is a number it earned.
- Ladder plus holdout
- Stats on Commerce > Recommendations
- Add-to-cart via the store’s own handle
- Proven live on a store before it shipped
Checkout abandoned
cart ₹6,240 · known customer, reachable
45 minutes later
coupon COMEBACK10 sent by email + WhatsApp
Opened, clicked, came back
viewed cart → checkout, same evening
Purchased ₹6,240
credited to the send, not re-credited to the ad
Sent
90
Opened
48
Recovered
12
Revenue
₹0
Inside the module
Inside Recommendations
A widget, a rules page, and a stats page that keeps them honest.
The widget
Three cards, add-to-cart, clock band, reorder strip
Recommendation rules
Eight, previewed through the engine
Discounts
Observed, imported, suggested
Stats
Bought, worth, holdout
Catalog
Categories as reported, product IDs per destination
Install
One flag in the snippet builder
Results are reported against a holdout, never against nothing. If a number here cannot be earned, it is not shown.
The other modules
Attribution
Every order credited to the ad that earned it, with the receipt and a match score.
OpenGrowth intelligence
Command Center, Today’s Actions and What We Found: what to do on Monday, priced on your costs.
OpenAnalytics & Insights
Twenty reports on one date window, every one ending in a decision.
OpenAudiences
Segments from behaviour the store records, sized against what Meta can match before they sync.
OpenPut three honest cards on every product page
Recommendations that sell.Measured against a holdout.
Switch the widget on in the snippet, pick a rule, and read what it earned on the stats page a week later.