Almost every app in this category sells both, which is why merchants treat search and filtering as one feature with two names. They are not. They serve opposite shopper mindsets, they run on different data, they break in different ways, and they are measured with different numbers.
The practical consequence: a merchant convinced they have a search problem will often buy a solution to a filtering problem, or the reverse, and stay frustrated afterwards.
The distinction in one line
Search is the shopper telling you what they want. Filtering is you showing the shopper what is available.
Search starts with the customer's words. They arrive with something specific in mind, type it, and expect you to interpret it. The burden is on your store to understand language.
Filtering starts with your catalogue. The shopper does not know exactly what they want, so they narrow a set by attributes you have defined. The burden is on your product data to be complete and consistent.
One is interpretation. The other is structure.
How search works on a Shopify store
Search runs on text matching plus a set of behaviours designed to forgive imprecision.
Shopify searches product title, description body, product type, tags, vendor, variant title, SKU and barcode. Ranking considers how often the term appears, which field it appears in with titles outranking descriptions, how long the matching field is, and popularity signals including recent sales.
Layered on top are typo tolerance, singular and plural matching, automatic prefix matching on the last word, and — on Grow, Advanced and Plus plans with under 200,000 products — semantic understanding that interprets related concepts rather than literal words.
There are also two distinct search surfaces that behave differently. The results page, and predictive search, the autocomplete dropdown that appears as someone types. They are not the same feature, and semantic understanding does not apply to the dropdown.
How filtering works on a Shopify store
Filtering runs on structured attributes rather than language.
Filters come from six standard sources — Availability, Category, Price, Product type, Tags and Vendor — plus custom filters built from product options, metafields and metaobjects. You can have 25 filters in total, each source usable only once.
The logic is fixed and worth understanding: filters combine with AND between different filters, and OR between values within one filter. Choosing red and green in a colour filter widens results. Choosing red in colour and 8 in size narrows them.
Filters also have hard ceilings. A collection over 5,000 products displays no filters at all, and any single filter displays a maximum of 100 values on the storefront.
Notice what is absent from all of that: interpretation. A filter does not guess. It either has the attribute recorded on the product or it does not.
The cleanest proof they are different things
Here is the fact that settles it. On Shopify, metafields can be filtered but not searched.
You can build a filter on a "material" metafield and shoppers can narrow to cotton. But if a shopper types "cotton" into the search box, the native search does not look at metafield values at all.
Same data. Same store. Available to one system, invisible to the other.
If you remember nothing else from this article, remember that, because it explains a scenario merchants find baffling: the product is right there, the filter finds it, and search returns nothing.
They fail in different ways
Search fails loudly. The shopper gets a blank page, an obviously wrong result, or nothing. You can count these — the zero-result report tells you exactly which queries failed. Search failure is visible and diagnosable.
Filtering fails silently. Nobody sees an error. A shopper picks "cotton" and sees eleven of your nineteen cotton products, because eight of them never had the material recorded. They buy one of the eleven, or leave. Nothing in your analytics flags it. The filter worked perfectly; the data behind it did not.
This asymmetry matters for where you spend attention. Search problems announce themselves. Filter problems require you to go looking, usually by auditing attribute completeness across the catalogue.
[YOUR STORY: an example of the silent filter failure — a catalogue where incomplete attributes were hiding products from shoppers who filtered. This is the concept most merchants have never considered, so a real example makes it stick.]
They are measured differently
For search: zero-result rate, top search terms, whether searches convert better than non-search sessions. Search users are usually your highest-intent traffic, so their conversion rate is worth isolating.
For filtering: which filters get used, which combinations return nothing, and how complete your attribute data is. Attribute completeness is the leading indicator, and it is the one you can act on before customers are affected.
A useful habit: read the zero-result report monthly for search, and audit attribute coverage quarterly for filters. Different cadences because they degrade at different speeds.
Which one is actually your problem?
A quick diagnostic.
It is a search problem if: shoppers get no results for products you stock, results are irrelevant, customers use different words than your product titles, or the same terms keep appearing in your zero-result report.
It is a filtering problem if: shoppers land on a large collection and cannot narrow it, filter options are missing or duplicated, filtering returns fewer products than you know you have, or your collections are so large that filters have stopped displaying entirely.
It is a data problem if: both are broken. This is the most common answer, and it is worth sitting with, because neither a search app nor a filter app fixes a catalogue where nobody recorded the material, the brand is spelled three ways, and half the metafields are empty.
Filters are generated from your product data, and search ranks on your product text. Both are downstream of catalogue quality. That is why the honest first step in either case is an audit rather than a purchase.
So why are they always sold together?
Two legitimate reasons and one commercial one.
They share the same underlying index, so a vendor building one can usually build the other. They also appear together in the customer journey — a shopper searches, lands on results, then filters those results — so the handoff between them needs to work.
The commercial reason is that bundling makes for a longer feature list. Which is fine, as long as you evaluate the bundle against your actual problem rather than being impressed by its size. If your issue is that shoppers cannot narrow a 900-product collection, a sophisticated semantic search engine is not what you needed, and you will be paying for it every month.
Common questions
Do I need both search and filters?
Most stores benefit from both, but with different urgency. If your catalogue is small and browsable, filtering matters more than search. If it is large and shoppers arrive knowing what they want, search matters more.
Can Shopify search find products by metafield?
No. Metafields can power filters, but the native search does not search their values.
Is Search & Discovery a search app or a filter app?
Both. Shopify's free app handles synonyms and product boosts on the search side, and filter creation, grouping and sorting on the filter side.
Why do my filters show fewer products than I have?
Almost always incomplete product data. A product only appears under a filter value if that value is recorded on it.
Which should I fix first?
Whichever is failing measurably. Check your zero-result report for search, and audit attribute completeness for filters. Fix the one with evidence behind it.
The useful takeaway
Before you evaluate any app, decide which of the two you are actually trying to fix, and confirm it is not really a product data problem wearing a costume.
That single distinction will save you money, because the app that fixes one of these often does very little for the other — and the fee is the same either way.
For the filtering side, how to add filters without paying extra covers the free setup in full. For the search side, native search versus a third-party app covers what Shopify does natively and where it genuinely stops.
