SocialHub.AI

AI EDM Marketing · Recommendation engine

Products picked for each member — from what your customers actually buy.

Flash has a built-in recommendation engine. It learns which products go together from your real order history, picks the right ones for each member, and drops them into the email — with nothing to configure.

Recommended for Maya · chosen for her
Trail Runner GTXFrequently bought togetherfrom $129
Merino Sock 3-packFrequently bought togetherfrom $24
Summit Daypack 22LPopular in categories you shopfrom $89

Drawn from what your customers actually buy together

The problem

“Best-sellers for everyone” is not a recommendation.

One list for the whole list

The same hero products go to every member, regardless of what they've bought. Relevance — and revenue — left on the table.

Merchandising by hand

Someone hand-picks products for each send. It doesn't scale, and it's stale the moment the catalog moves.

Bolt-on tools, broken cards

Generic add-ons recommend products that are out of stock, mis-priced, or link nowhere — because they don't read your real catalog.

How it works, end to end

From your order history to a personalized email — automatically.

The engine works in two simple ideas: products real customers buy together, and the best-sellers in the categories a member already shops. Here's the whole path.

01Learn from real ordersEvery night, the engine scans your order history and learns which products customers actually buy together.
02Pick for each memberAt send time, it matches each member's own purchases to those pairings — and leaves out what they already own.
03Fill from their categoriesIf there aren't enough pairings yet, it tops up with the best-sellers in the categories that member already shops.
04Pull live product detailsEach pick is resolved against your catalog — real name, image, current price and the real product link.
05Drop into the emailThe Recommendations block fills with that member's cards. No data feed to wire, no rules to write.

Refreshed every night

The “bought together” patterns are rebuilt nightly from your latest orders, so recommendations track what's selling now.

Computed per member, at send

Each recipient's picks are worked out from their own purchases the moment the email goes out — not a frozen list.

Zero configuration

Add the Recommendations block and it just works. There's no data feed to connect and no rules to write.

Where to use it

The campaigns it's built for.

One block, several jobs. Drop it into the moments where the next product matters most.

Cross-sell — “frequently bought together”

The core of the engine. Recommend the products that real customers pair with what this member already bought — the accessory, the refill, the matching piece.

Post-purchase — “complete your set”

Because picks come from a member's own order history and exclude what they own, a post-purchase email naturally suggests the next thing, not the thing they just bought.

Category expansion — “popular in categories you shop”

Surface the best-sellers in the categories a member already buys from — a gentle nudge to go wider without leaving their interests.

Zero-config “Recommended for you”

Add the block to any campaign — a newsletter, a win-back, a tier upgrade — and every recipient gets their own personalized row, with nothing to set up.

Honest by design

It would rather show nothing than guess.

A recommendation you can't stand behind costs trust. The engine is built to be grounded in real data — and to step aside gracefully when it doesn't have enough.

Grounded, never fabricated

If a member has no purchase history to learn from, the block simply doesn't show — no made-up “you might like” filler.

Never recommends what they own

Products a member already bought are excluded, so the email always points forward.

Only real, sellable products

A pick with no image, no price, or no working link is dropped — members never see a broken card.

Bring your own engine, optionally

The built-in engine is the zero-config default. If you already run an external recommender, you can plug it in instead — your choice, not a requirement.

Part of the platform

One catalog, one member record — powering more than email.

The engine reads the same product catalog and order history that run the rest of Flash — no second copy, no export. The same recommendations are available through the API and to AI assistants, not just in the inbox.

Reads your real catalog

Names, images, prices and product links come straight from the catalog that's already kept in sync from your store.

In your campaign emails today

Live in the Recommendations email block, filled per recipient at send — the path most teams start with.

Open to your AI tools

The same member recommendations are reachable over the API and as a governed tool for AI assistants.

What changes for the business

Personalized product recommendations in every send — earned from real data, with nothing to set up.

Per member

a different set of products for each recipient

Zero config

add the block — no data feed, no rules

Nightly

patterns refreshed from your latest orders

Grounded

real catalog only — no broken or made-up cards

See it recommend from your own catalog.

400M+
50+
12+