Personyze Wiki Personyze Wiki docs
Open Personyze
Product Recommendations

Use Case: Product Recommendations, End to End

From empty account to a revenue-attributing recommendations widget: catalog, interaction tracking, algorithm, placement, QA and measurement — in order, with links into every detailed guide.

Updated 1 day ago 3 min read
A
by Admin


The goal: a recommendations widget on your site that learns from what visitors view, add to cart and buy — and measurably lifts revenue. This recipe walks the whole path in order, with a link into the detailed guide at every step. Budget roughly an hour for a first working setup.

The path at a glance

  1. Product catalog in
  2. Interaction tracking on
  3. Campaign + widget
  4. Algorithm, filters, fallbacks
  5. Placement & look
  6. QA, publish, measure

1. Get your product catalog in

The catalog is what Personyze recommends from — titles, prices, images, stock, and any field you’ll want to filter or display. Pick one route:

2. Turn on interaction tracking

Three events feed the engine — product viewed, added to cart, purchased. Views and carts drive the ranking; purchase also attributes revenue. Set it up per product interaction tracking — or let the Shopify / WooCommerce integrations wire it, or reuse your GA4 dataLayer through the GTM template’s ecommerce checkbox.

No history yet? Until enough data accumulates, Personyze falls back to contextual recommendations so the widget is never empty — and you can upload past transactions to expedite learning.

3. Create the campaign and widget

Run the Product Recommendations wizard — it walks Catalog → Interactions → Content → Target → QA in order and checks readiness as you go. Pick a widget template in the gallery (sliders, grids, cards — each documented under Product Recommendation Templates).

Template gallery
Pick the widget; every template family has its own detailed article. Click to enlarge.

4. Choose the algorithm — and sharpen it

Start simple and personal: recommendations for this visitor on the homepage, frequently bought together on product pages, items left in cart on return visits. The full catalogue with when-to-use guidance: algorithms overview and algorithm types. Then:

  • Filters — in-stock only, price range, same category as viewed, any catalog field.
  • Fallbacks — what fills the widget when the primary algorithm runs short.

5. Placement and look

Inline in a placeholder (product page, cart, homepage rail) or floating. Card fields map to your catalog columns; every setting has audience-based ⚡ Variations and there’s an optional quick add-to-cart button. Details: the action guide.

6. QA, publish, measure

  • QA step: preview as a visitor, confirm readiness checks are green, then publish to staging → live.
  • The Performance step tracks impressions, CTR and attributed revenue; A/B test algorithms or designs against each other and let conversions decide.

Variations on this recipe