Skip to main content

Fine-tuning recommendations

Find out how to change the calculation method for recommendation types that support fine-tuning at API level.

⏳ Duration: 10 minutes
🤖 Complexity: high

📌 Prerequisites

1. An Aqurate account (get it at aqurate.ai)

2. An active integration of your Shop data (connect your shop here)

3. An active subscription with Aqurate Personalize

📖 Resources

Check out and test our endpoints in the Recommendation section of our docs.

🤔 What can be fine-tuned

Two recommendation types support fine-tuning: Bestsellers (store, category, and item level) and Recommended for you. Everything below is done via API. For prioritizing or hiding specific products instead, use Merchandising.

⭐️ Bestsellers

Each bestseller endpoint takes the same two settings:

Recommendation

Endpoint

Set lookback window

Set method

Popular now

/recs/bestseller

/set-lookback-window/{days}

/set-method/{method}

Popular in category

/recs/item/bestseller-in-same-category

/set-lookback-window/{days}

/set-method/{method}

Popular in same category

/recs/category/bestseller

/set-lookback-window/{days}

/set-method/{method}

  • Lookback window: days of data used to calculate bestsellers. Default: 14.

  • Method: volume, balanced, value, distinct buyers, most viewed. Default: volume.

👉 Recommended for you

/recs/user:

  • /set-max-orders/{max_orders} — excludes products the user already bought. Default: 0.

  • /set-max-views/{max_views} — excludes products the user already viewed. Default: 0.

Did this answer your question?