How to Use Product Metafields for Smarter Upsell Recommendations

Most Shopify stores hold far more product knowledge than their upsells ever use. Which bulb fits which lamp. Which serum suits which skin type. Which roast belongs to which brewing method. That knowledge usually lives in product metafields, and using product metafields for upsell recommendations is one of the simplest ways to make every offer feel hand-picked.
This guide explains why metafields beat tags for recommendation logic, the difference between a hard rule and a soft preference, and six concrete examples you can copy for your own store.
Why Product Metafields Make Better Upsell Recommendations
Recommendation engines learn from purchase history: what gets bought together, gets recommended together. That works well for bestsellers and badly for everything else. A new product has no history. A long-tail item may only sell a few times a month. And purchase data cannot know that a product is incompatible, only that nobody bought it.
Metafields fill that gap. They are structured fields you define once in Shopify under Settings → Custom data and then fill in per product: a fitting code, a material, a product line, a skin type. Unlike tags, they are typed and consistent, so "GU10" is always "GU10" rather than "gu-10" on one product and "GU10 bulb" on another.
That makes them ideal recommendation inputs:
- They work from day one. A new product with a filled-in metafield can be recommended correctly before it has sold once.
- They encode facts, not guesses. Compatibility, material and size are things you know, not things an algorithm has to infer.
- You already maintain them. Many stores fill metafields for filters, product pages or feeds. Reusing them costs nothing.
Hard Rules vs. Soft Preferences
There are two ways to use a metafield in your recommendation logic, and choosing the right one matters.
A hard rule (filter) decides what is allowed to show. "Only recommend bulbs with the same fitting as the lamp in the cart." Anything that fails the rule is removed. Use this when a wrong recommendation is genuinely wrong, like an accessory that physically won't fit.
A soft preference (boost) decides what shows first. "Prefer products from the same collection line, but fill the rest with the best general picks." Nothing is removed, matches just rank higher. Use this when a match is better but a non-match is still a fine offer.
A good rule of thumb: if a shopper would return the product, filter. If they would simply find it less exciting, boost.
Soft preferences also work in reverse. You can dampen matches to push them down the list, which is how you stop a cart from recommending three near-identical versions of the product the shopper already chose.
6 Examples of Metafield-Driven Upsells
Each example names the metafield, the store type it suits, and whether it works best as a hard rule or a soft preference.
1. Compatibility codes for lighting and electronics
Metafield: fitting or connector type, such as E27, GU10, USB-C.
A shopper adds a pendant lamp. The best upsell is a bulb, but only one that fits. Store the fitting on both lamp and bulb, then only recommend bulbs with the same value. This is a hard rule: a bulb that doesn't fit is a return waiting to happen.
The same pattern covers phone cases by model, printer ink by printer series, or filters by vacuum model. With a list metafield, a case that fits three phone models can carry all three values.
2. Product lines for fashion and beauty
Metafield: product line or collection name, such as Linen Essentials or Hydra Range.
Someone buying the cleanser from a skincare line is a strong candidate for the matching toner and moisturiser. Boost products from the same line so they appear first, but keep other bestsellers in the mix for shoppers who want to explore. A soft preference fits here, since a product from another line is still a reasonable offer.
3. Skin type, hair type or fit profile
Metafield: skin type (dry, oily, sensitive), hair type, or fit (slim, relaxed).
Recommending a mattifying primer to someone who just added a rich cream for dry skin feels careless. Boost products that share the profile value, or filter strictly if a mismatch could cause a bad experience. Profile metafields also make your product pages and filters better, so the work pays off twice.
4. Variety for food, coffee and candles
Metafield: roast level, flavour family, or scent family.
Here the goal is the opposite of matching. A shopper buying a dark roast often wants to try something different, not a second dark roast. Dampen products with the same roast, or prefer products whose value differs from the cart item. The cart stops repeating itself and starts suggesting discovery, which tends to land better for consumables people buy repeatedly.
5. Seasonal or campaign values
Metafield: season or campaign, such as Winter 2026 or Gift Guide.
During a campaign, boost products carrying the campaign value so they lead your recommendations without hard-coding a manual product list. When the campaign ends, change the strength back to neutral or clear the value. This pairs well with collection-scoped product recommendations when your seasonal range already lives in its own collection.
6. Material care and refills
Metafield: material (leather, merino, cast iron) or refill system.
A leather bag sells leather care. A cast iron pan sells seasoning oil. A refillable diffuser sells its own refills. Store the material or refill system on both sides and filter, so a shopper never sees suede spray next to a merino jumper. These upsells are cheap, useful and highly relevant, exactly the kind shoppers add without a second thought.
Setting Up Metafields That Work for Recommendations
A few habits keep metafield-based rules reliable:
- Create a definition first. In Shopify, add the metafield under Settings → Custom data → Products. Definitions make the field selectable in apps and keep values consistent.
- Use single line text or a list of it. Short, exact values like
GU10match cleanly. Use a list when one product fits several values. - Agree on a value format. Decide on capitalisation and spelling once, and write it down for anyone who adds products.
- Fill both sides. A compatibility rule only works if the trigger product and the recommended product both carry the value.
- Start with your best-selling categories. Fill metafields where most carts start, then expand.
Putting It Into Practice with EliteCart
In EliteCart, metafields plug straight into fine-tuned versions of EliteAI™ Ultra. Under Upsells → AI upsells → Fine-tune EliteAI™ Ultra, you can use a metafield as a filter to require a match, or as a boost to prefer, or dampen, matching products on a five-step scale. Boosts also shape the recommendations EliteAI fills in from your catalog for products with little order history, so new products follow your rules immediately.
For every option with worked examples, see the filters and boosts help article.
Start with one metafield. Pick the product fact that most often makes or breaks an add-on in your store, whether that is a fit, a line or a material, and turn it into a recommendation rule. Your catalog already knows what belongs together. Let your upsells use it.