1. Write three buying tasks
Choose one collection and write three realistic tasks in plain language. For example: find a tent that fits a family’s sleeping arrangement, compare chairs that fit a storage space, or shortlist a fridge within a stated size range. These are example tasks, not universal filter requirements.
Try completing each task using the existing store. Note when you have to open several product pages just to discover a basic difference. The missing comparison may be a filter candidate, but it could also belong on the product card. A filter is useful when it narrows the range; a card detail helps someone evaluate the remaining products.
2. Separate categories, filters and specifications
A category describes the range someone is browsing. A filter narrows that range. A specification explains an individual product. Treating all three as interchangeable can create menus with hundreds of entries and filters full of values that only occur once.
Create a small planning table before changing the theme. Record the customer’s question, the proposed attribute, its source and the products that lack a value. Use a field only when your team can explain and maintain its meaning.
| Buyer question | Candidate attribute | Data check |
|---|---|---|
| How many people is it designed for? | Stated sleeping capacity | Keep capacity separate from floor dimensions |
| Will it fit in storage? | Packed size information | Use consistent units and distinguish each dimension |
| What does it need to operate? | Power-source category | Verify terminology and required accessories |
3. Check the data and theme support
Shopify Search & Discovery supports filters based on standard attributes and supported custom data, including product options and metafields. The storefront also needs compatible theme support. Check Shopify’s current documentation before choosing a source or designing around platform limits.
Then inspect a sample of your own catalogue. “2 person”, “Two-person” and “2P” may mean the same thing to the team but produce an inconsistent experience. Agree a preferred vocabulary, preserve useful source data and correct the customer-facing values. Avoid inventing a value for products whose specification has not been confirmed.
4. Test the complete filtering task on mobile
Start with no filters, apply two useful choices, inspect a product and return to the results. Check whether selected values remain understandable, whether clearing one value behaves as expected and whether the result count gives useful feedback. An empty result should have a clear route back.
Also try the less tidy cases: a missing attribute, a sold-out variant and a collection with only a few items. Check that keyboard users can reach the controls and that opening a mobile filter panel does not hide the apply or close action. Record the expected behaviour rather than accepting whatever the app happens to do.
5. Evaluate discovery and maintain the fields
Compare how people complete the original tasks before and after the change. Where suitable analytics exist, review filter use and onward product visits alongside stock and traffic changes. A higher click count alone is not proof of a better buying experience.
Give the attributes an owner in the product-creation process. A good filter degrades if new products arrive without the necessary data. Decide separately which collection pages should attract search traffic; do not assume every combination of filter values deserves its own indexed landing page.
Want a second pair of eyes?
Tell us about your products, your current store and what needs to work better. We’ll discuss a suitable scope before you commit to paid work.
Discuss your product discovery