Every algorithm ends by ordering what it found. The score blend is the account-wide setting that
decides how that ordering weighs its inputs — and it applies to every widget on the site at once.
Video: Trending products and recommendation sets: what shoppers buy together (3:09) — it plays at the top of Recommendation Algorithms — What Each One Picks.
Settings → Site profile → Recommendations. Changes reach live widgets within minutes; no
republish, no rebuild.
The four weights
| Weight | What it does | Default |
|---|---|---|
| Purchases vs views | How much a purchase outweighs a view when ranking popularity. | ×1 |
| Freshness | How much a recent item is favoured over an older one. | ×1 |
| Click & purchase feedback | How much the engine learns from what visitors actually clicked and bought out of the widgets themselves. | ×1 |
| Visitor’s price range | Per-visitor, not global — see below. | Off |
placeholder. It is where to start, and where to go back to if a change makes things worse.
Presets, by kind of site
A travel site is not a shop, and a publisher is neither. The presets fill the weights with editorial starting points
in one click, and you can tweak from there:
Balanced · Content & publishing · Travel &
booking · B2B · Services
The visitor’s price range
The fourth weight is different in kind from the other three: it is per visitor. It nudges up items
priced between roughly half and double that person’s own average — taken from what they have viewed, carted
or bought.
- It is off by default, and no preset ever switches it on. Turning it on is always
an explicit choice. - Visitors with no priced history are unaffected — there is no average to work from, so nothing is nudged.
The safety floor: minimum users behind a recommendation
Above the blend sits a plainer control, and it decides what the engine is allowed to use at all rather than how it is ordered. Minimum users behind a recommendation is how many distinct visitors must support an item-to-item pair before the engine will serve it.
It is a straight trade: higher is safer and gives fewer recommendations. One person who viewed two unrelated products is not evidence that they go together; twenty people are. Raise it when a widget is showing pairings that look like coincidence, and lower it when a smaller catalogue or a quieter site leaves widgets falling back too often.
It applies to the computed item-to-item pairs — bought-together, viewed-together and their relatives. It has nothing to say about popularity, the catalogue itself, or pairs you uploaded by hand as cross-sell or up-sell sets.
Similar items (AI)
The same screen carries the switch for Similar items (AI) — the nightly pass that reads your catalogue’s own text and computes each item’s nearest neighbours by meaning, so a brand-new item can be recommended before anyone has viewed it.
One switch covers the whole account, products and content alike. It is the same setting you will find in each catalogue’s extra-data settings — two doors into one switch, not two switches. What it costs and what it does well is described under Similar Items (AI).
Per-widget boosts, in context
A single widget can still be tuned on its own, under Fine-tune the algorithm in the recommendation
editor — Boost by profit margin, Boost new items, Reserve a slot for new items.
That card shows the account blend live underneath it (“Account-wide blend now: Purchases ×1 ·
Freshness ×1 · Click & purchase feedback ×1 · Visitor’s price range
Off”), so a per-widget boost is never set blind to the account setting it stacks on.
Related
- Recommendation algorithms — what each one picks before the blend orders it.
- Site profile settings
- The data behind recommendations