Tradly Recommendations

Show each user
what they'll
actually want.

Personalised recommendations across every surface — homepage, product pages, email, and search. Powered by real behavioural signals. No manual curation, no ML team required.

Start free →Read the docs

+28%

avg basket size

3×

search conversion

Day 1

works from first session

Signals used

Learns from everything users do.

Every interaction is a signal. Tradly collects and weights them automatically — you don't need to define what matters, the model figures it out.

View history

Pages and listings a user visited, weighted by time spent

Search queries

What they searched for, even if they didn't buy

Purchase history

Past orders inform future preferences strongly

Explicit ratings

Stars, likes, saves — if your UI captures them

Similar users

Collaborative filtering: users like you also bought

Category affinity

Derived from behaviour, no explicit input needed

Placement options

Every surface, personalised.

Turn on recommendations per placement independently. Start with the homepage shelf and add more when ready.

Homepage shelf

"Recommended for you" — personalised from first session

Product page

"Others also bought" and "Similar items" panels

Dynamic recommendation blocks in triggered and marketing emails

Search re-ranking

Surface most-relevant results first per user

Cart / checkout

Upsell and cross-sell at highest intent moment

Empty states

Fill zero-result searches with predicted preferences

Algorithm modes

Hybrid by default. Configurable when you need it.

The default hybrid mode works well for most platforms. Switch modes per placement if your product has specific needs.

Collaborative filtering

Finds patterns across all users. Works well at scale — no manual curation needed.

Content-based

Matches item attributes to a user's preference history. Works from day one for new users.

Popularity fallback

Used for cold-start users. Shows trending items until personal signals build up.

Default

Hybrid (default)

Blends all three modes automatically, weighted by data confidence per user.

API access

Bring recommendations anywhere.

Query recommendations via REST API to embed them in custom layouts, native apps, or headless storefronts.

HTTPGET recommendations for a user

GET /api/v1/recommendations?user_id=usr_7x2kp&limit=12&placement=homepage

Authorization: Bearer <your-api-key>

// Response
{
  "placement": "homepage",
  "user_id": "usr_7x2kp",
  "confidence": "high",
  "items": [
    { "listing_id": "lst_001", "score": 0.94, "reason": "collaborative" },
    { "listing_id": "lst_089", "score": 0.91, "reason": "content_based" },
    { "listing_id": "lst_234", "score": 0.87, "reason": "hybrid" },
    ...
  ]
}