EliteCart

Most merchants think of curating recommendations as a manual job: open each product, pick the three or four items that should appear alongside it, and repeat for the whole catalog. That works for a store with twenty products. It collapses at two hundred. Tag based product recommendations solve this by letting you describe what a good suggestion looks like once, in plain attributes you already store on every product, and have the engine apply that description across the catalog for you.

Your Shopify products already carry three fields that are quietly perfect for this: tags, vendor, and product type. Used well, they become a curation layer that steers recommendations without anyone hand-picking a single pairing. This guide covers practical tagging schemes and the include, exclude, and match-the-cart logic that turns them into better suggestions.

Why Tag Based Product Recommendations Beat Manual Pairing

A recommendation engine learns what sells together from your order history. That is a strong starting point, but it has no opinion about your strategy. It does not know that you would rather push your private-label line than a third-party brand, or that "gift" items should stay together, or that a consumable should always suggest a refill. Those are merchandising decisions, and tags are how you encode them.

The advantage over manual pairing is reach. Tag one hundred products as premium and you have just described a recommendation rule that covers all one hundred at once, plus every future product you tag the same way. You maintain the labels, the engine maintains the pairings. When you add a product next week and tag it correctly, it inherits the curation automatically.

This is the same instinct behind merchandising rules for product recommendations: you set the policy, the engine fills in the products.

Tagging Schemes That Translate Into Better Suggestions

Good curation starts with good tagging hygiene. The goal is a small, consistent vocabulary that maps to merchandising intent, not a sprawl of one-off labels. A few schemes that pay off directly in recommendations:

Bundle-friendly. Tag the products that genuinely pair well as add-ons (bundle-friendly) so the engine can prefer them when filling a cross-sell slot. A phone case, a screen protector, and a charging cable all earn the tag; the phone itself does not.
Premium. Tag your higher-margin or flagship items as premium so you can nudge recommendations upmarket when it makes sense, without ever forcing it.
Gift. A gift tag keeps giftable items grouped, which is useful around seasonal peaks when a gift in the cart should suggest more gifts rather than everyday staples.
Consumable. Tag refills, cartridges, pods, and other repeat-purchase items as consumable so a consumable in the cart can reliably surface its companion or refill.

Keep the vocabulary short and write it down. A tag only curates well if it means the same thing on every product, so bundle-friendly should never drift into bundle, bundles, and Bundle Friendly across different products. Inconsistent tags are the single most common reason curation rules quietly miss products.

Vendor and Product Type Do Real Work Too

Tags are flexible, but vendor and product type are often the cleaner lever because they are usually already accurate on every product.

Vendor to keep a brand together. If a customer adds a specific brand's product, recommending that brand's matching accessories keeps the suggestion coherent. Filtering or nudging by vendor does this without any tagging at all.
Vendor to push your private label. If you carry third-party brands alongside your own line, vendor is how you tilt recommendations toward your private-label products, where your margins are best.
Product type to stay in the right department. Product type keeps a suggestion inside the right category, so a "Coffee" product does not start recommending "Mugs" unless you want it to.

Include, Exclude, and Match-the-Cart Logic

Once your products are labeled, the curation itself comes down to three kinds of rule. EliteCart's Fine-tune EliteAI™ Ultra controls express all three on top of an engine trained on your own store's orders and catalog.

Hard Filters: Include and Exclude

A filter is an absolute rule. Include narrows the recommendation pool to products that carry a given tag, vendor, or product type. Exclude removes them entirely. A product that fails an include rule, or matches an exclude rule, simply cannot appear, no matter how strong its purchase signal is.

This is the right tool when a boundary is non-negotiable:

Include only bundle-friendly items in an add-on slot, so the cross-sell never offers a flagship product as an afterthought.
Exclude a clearance tag so you never recommend something about to sell out.
Include a single vendor to keep a brand-line product page recommending only that brand.

If you want a deeper tour of keeping the wrong products out, how to control which products get recommended walks through filters and exclusions in detail.

Match the Cart Item's Tag

The most useful piece of tag logic is also the least obvious: recommend products that share a tag with whatever is in the cart. Instead of naming a specific tag, you tell the engine to recommend items carrying the same tag as the cart item. Add a summer product and it suggests other summer products; add a gift and it suggests other gifts. One rule adapts to every product in the catalog, which is curation without enumeration.

This pairs naturally with collection-scoped recommendations when you want suggestions to stay inside a relevant slice of the catalog.

Soft Boosts: Nudge Without Removing

Filters are binary. Boosts are a dial. A boost nudges recommendations toward a tag, vendor, or product type on a five-way scale without removing anything else from consideration. This is how you express a preference rather than a rule.

Use a boost when you want to lean a direction but keep variety:

Lightly favor your premium items so they surface more often, while everyday products still appear.
Nudge toward your own vendor line so private-label products get a visibility edge over third-party brands.
Gently prefer bundle-friendly items so add-on slots skew toward genuine companions.

The practical rule of thumb: reach for a filter when something must or must not appear, and a boost when you simply want more or less of it.

If/Then Conditions for Context-Aware Curation

Curation gets sharper when a rule only applies in the right situation. An If/Then condition runs a rule only when the cart item matches an attribute. For example, only push a brand's accessories when the cart item is from that brand, or only include gift items when the cart item is itself a gift.

This keeps your rules from over-reaching. A blanket "boost premium" applies everywhere, which is not always what you want. "If the cart item is premium, then boost other premium items" applies the nudge exactly where it belongs and stays out of the way otherwise.

You can also choose what the engine pairs from. The default base suggests frequently-bought-together products, while a cross-category base surfaces complements from adjacent categories. Tag and vendor rules layer on top of either, so the same curation vocabulary works whether you are recommending refills or reaching across departments.

Putting It Together

A tagged catalog turns into a tuned engine in a handful of moves. Train a fine-tuned version on your own orders and catalog, then assign it where it matters: the cart, the product page, the two-step cart, or checkout modules. You can keep up to three live versions, so a gift-leaning configuration can run during a seasonal push while your everyday setup serves the rest of the year. The EliteAI fine-tuning update covers how versions and per-surface assignment work.

For the exact controls, the Filters and Boosts reference lists every option, and the Fine-Tuning setup guide walks through building your first version.

The work that matters most happens before any of this, though: clean, consistent tags and accurate vendor data on your products. Tagging hygiene is the one investment that makes every recommendation rule easier, and it is data you control entirely.

Start with your tags. Pick three or four labels that describe real merchandising intent, apply them consistently across your catalog, and let include, exclude, and match-the-cart logic turn that vocabulary into curated recommendations that scale with your store.

Using Product Tags and Vendors to Curate Recommendations