Should You Build or Buy Filters & Faceted Navigation on Shopify?
Filters and faceted navigation reward a CUSTOMIZE play for most mid-market Shopify catalogs: model facet attributes as metafields first, because native Search & Discovery renders them free to roughly 1,000 SKUs, and a third-party engine only pays back in 6–9 months once you pass about 10,000 (July 2026 research). Apps rent the render; the metafield schema is the asset. No engine can facet an attribute you never captured.
Your profile — see how the verdict shifts
- Confidence
- High — Every render path (native, app, or custom stack) facets the same metafields, so the data-first verdict holds across vendors and catalog sizes
- Reference scenario
- $20M–$100M GMV · 5,000–20,000 SKUs · agency dev bench
- As of
- August 2026
Decision at a Glance
| Your profile | Verdict | Why |
|---|---|---|
| Under 1,000 SKUs | WAIT | Native Search & Discovery is free and holds up at this size (July 2026 research). Do the metafield hygiene in-house; skip the subscription. |
| 1,000 – 10,000 SKUs | CUSTOMIZE | The sweet spot for the thesis: model facets as metafields and let native or a light filter app render them. The data work moves the needle; the app choice barely does. |
| 10,000 – 50,000 SKUs | CUSTOMIZE | Above ~10K SKUs a third-party engine pays back in 6–9 months (July 2026 research), but it still facets only the attributes you've modeled. Buy the engine; keep owning the schema. |
| 50,000+ SKUs / marketplace-grade catalog | DEPENDS | Enterprise filter tiers and dedicated-stack economics converge here. Build an Elasticsearch-class stack when discovery is the product; buy the enterprise tier when it's a feature. |
What Filters & faceted navigation Actually Drives
| Outcome | Impact | How it works |
|---|---|---|
| Revenue — direct | High | Filter users self-qualify: narrowing a 400-product collection to the 12 that fit turns browsing sessions into add-to-carts, which is where large-catalog conversion actually happens. |
| Customer experience | High | Facets make a big catalog feel small: size, material, compatibility, and ingredient filters answer the questions a store associate would ask first. |
| Data & insight | High | Filter selections are declared preference data, and the metafield schema behind them is the same structured attribute set product feeds and AI shopping surfaces read. |
| Operational efficiency | Medium | One attribute model feeds filters, site search, feeds, and PDP spec tables at once, so merchandisers stop re-tagging the catalog per tool. |
| Revenue — indirect | Medium | Faceted collection views surface long-tail inventory that never earns a homepage slot, spreading demand deeper into the catalog. |
Spend ceiling: Size the spend to the attribute model, not the filter tray. A rendered facet is worth little when the data behind it is missing or dirty, and the schema plus backfill is the only part that ports across every engine you'll ever use. The tray is a commodity; the model compounds.
What buying enables (top apps)
- + Polished filter trays with counts, swatches, and price sliders, live within days
- + Merchandising controls: pin, boost, demote, and rule-based ordering inside filtered views
- + Bundled site search with typo tolerance and synonym management in the same subscription
- + Filter-usage and zero-results analytics out of the box
What building additionally unlocks
- + A facet schema modeled on your catalog's real buying logic (fit, compatibility, ingredients) that any engine renders without re-tagging
- + Owned structured attributes that double as feed, PDP spec, and AI-shopping-surface data, outliving any filter vendor
- + Script-free, server-rendered filtering on native storefront surfaces, with no app-bloat speed tax
- + A credible path to a dedicated search stack at marketplace scale (the Eluma pattern) without redoing the data
Find Your Verdict in 3 Questions
Do your products carry structured metafields for the attributes shoppers filter by, like size, material, compatibility, or ingredients?
Yes: Go to question 2.
No: Your verdict: CUSTOMIZE — start with the metafield model; no filter app can facet attributes that don't exist.
Is your catalog above roughly 1,000 SKUs, or do you need merchandising rules and search analytics native doesn't offer?
Yes: Go to question 3.
No: Your verdict: WAIT — native Search & Discovery renders metafield facets free at your size (July 2026 research); revisit as the catalog grows.
Is discovery the product itself: a 50,000-SKU-plus or marketplace-grade catalog where search is the primary UX?
Yes: Your verdict: BUILD — a dedicated Elasticsearch-class stack with faceted filters is defensible at this scale; it's the Eluma pattern.
No: Your verdict: CUSTOMIZE — buy the engine and keep the schema; above ~10K SKUs a filter app reading your metafields pays back in 6–9 months (July 2026 research).
The TCC Scorecard — 12 Dimensions
TCC — Total Cost of Capability: what it actually costs to have this capability over three years, whichever way you get it. Each dimension is scored 0–5 for both paths. How we score →
| Dimension | Buy | Build | Why |
|---|---|---|---|
| Cost | |||
| Acquisition & implementation | A filter app installs in days, though facet mapping is real setup work; the schema, backfill, and theme filter UI run an estimated 4–8 weeks (Deploi estimate, illustrative). | ||
| Recurring fees | Filter and search apps price by SKU count or sessions and climb with catalog growth; metafields cost nothing monthly and native rendering is free (included). | ||
| Maintenance & upgrades | App widgets breaking at theme updates and breakpoints is a documented community pain; a theme-native filter UI updates with the theme you already maintain (~$3K–$6K/yr, Deploi estimate, illustrative). | ||
| Switching & exit | Facet config, synonyms, and merchandising rules live in the vendor's dashboard and rarely export cleanly; metafields stay in Shopify no matter what renders them. | ||
| Risk | |||
| Vendor risk | This category consolidates: Klevu and Searchspring merged into Athos Commerce (Jan 2025), and roadmaps moved under new owners. A metafield schema has no vendor to lose. | ||
| Security & compliance surface | Filter apps read catalog and browsing data, not payments; a modest surface either way, and the customize path keeps query behavior first-party. | ||
| Platform-deprecation exposure | Storefront filtering and metafields are first-class primitives Shopify keeps investing in; apps absorb the ~6-month API version cycles for you, while a theme UI rides native surfaces. | ||
| Value | |||
| Fit to requirement | Apps ship polished trays and merchandising controls, and they're genuinely good; but no engine can facet an attribute you never modeled, so fit tracks the schema. | ||
| Time to market | An app renders filters this week; the data model is the long pole on every path, and honest backfill takes weeks, not days. | ||
| Performance & scale | Injected filter widgets add script weight, part of the documented app-bloat page-speed tax; native filtering renders with the theme. At 50K+ SKUs a dedicated stack outruns both. | ||
| Data ownership & AI-readiness | The decisive dimension: structured attributes are what product feeds, AI shopping surfaces, and future engines read. Owned metafields compound; app-side facet config is rented and stays behind at exit. | ||
| Focus & opportunity cost | The schema work isn't optional under any path, so building here means doing unavoidable work once, properly; only a dedicated stack becomes a real roadmap commitment. | ||
The App Landscape
| App | Status | Pricing | Best for |
|---|---|---|---|
| Boost AI Search & Filter | Live — The category's best-known filter-tray and search combo on Shopify | SKU/session-tiered | Fast, polished facet UX with merchandising rules on top of your metafields |
| Shopify Search & Discovery | Native — First-party, free. Shopify's free first-party app; renders metafield-based storefront filters | Free (included) | Catalogs to roughly 1,000 SKUs that have done the metafield work (July 2026 research) |
| Algolia | Live — API-first hosted search and faceting with a strong headless story | Usage-based | High-volume or headless storefronts with dev capacity to use the APIs |
| Athos Commerce | Live — Klevu and Searchspring merged into Athos Commerce (Jan 2025); evaluate the combined roadmap, not the legacy brands | Quote-based | Merchandising-heavy catalogs, especially existing Klevu or Searchspring customers |
The Build Path
- Metafield facet schema + native rendering: Define facet definitions per category (size, material, compatibility, ingredients), backfill with bulk operations, and let the free Search & Discovery app render them as storefront filters. Owned data, native render, no subscription.
- Same schema + commercial engine on top: Keep the metafield model as the source of truth and add a filter app for merchandising rules, synonyms, and analytics. Deploi has delivered metafield-source search filters this way for a mid-market brand without a platform swap.
- Dedicated search stack (Elasticsearch-class): Real-time sync, typo tolerance, and faceted filters on infrastructure you run: the pattern behind Eluma, Deploi's self-hosted Shopify search build. It earns its keep at marketplace scale, not before.
- Effort band
- $12,000–$35,000 for schema design, backfill, and theme filter UI — Deploi estimate (illustrative); most projects land in the $10–25K contact-form band, dedicated-stack builds in $25–75K
- Typical timeline
- 4–8 weeks for the schema and theme filter work; add 2–4 weeks when a commercial engine rides on top (Deploi estimate, illustrative)
- Maintenance, honestly
- ~$3,000–$6,000/yr (Deploi estimate, illustrative) for schema evolution and theme compatibility; as with any custom work, budget ~15–20% of build cost per year in upkeep (Deploi estimate). The data itself is low-touch: it updates with your products.
- What you own — and what you take on
- You own: the facet schema, every attribute value, and a data model that also feeds site search, product feeds, PDP spec tables, and AI shopping surfaces. You take on: schema governance, because someone has to keep attribute hygiene honest as new products land.
3-Year Total Cost of Capability
| Buy (app path) | Build (custom path) | |
|---|---|---|
| Year 0 (setup) | $1,000–$4,000 (install + facet config) | $12,000–$35,000 (schema, backfill, theme filter UI) |
| Years 1–3 (recurring) | $16,000–$34,000 (subscription, held flat) | $7,000–$18,000 (upkeep + schema evolution) |
| 3-year total | ≈$17,000–$38,000 | ≈$19,000–$53,000 |
- † All figures illustrative samples for the reference scenario — not quotes, not verified pricing.
- † App path: mid-tier SKU-based pricing held flat across the horizon (real filter-app pricing steps up with catalog growth — conservative for the customize case).
- † Customize path: metafield schema, backfill, and theme filter UI with native rendering; any commercial-engine subscription is excluded from the build column.
What the Sticker Price Hides
On the buy path
- — SKU- and session-tiered pricing steps up as catalog and traffic grow; the tier you priced isn't the tier you'll pay (community-reported pattern)
- — Filter widgets breaking at theme updates and breakpoints is a recurring community complaint; budget QA time per theme release
- — Vendor consolidation: Klevu and Searchspring became Athos Commerce (Jan 2025), and roadmaps or pricing get renegotiated over your head
- — Facet config, synonyms, and merchandising rules rarely export; switching engines means rebuilding them from scratch
On the build path
- — Backfilling attributes across thousands of SKUs is the unglamorous 60% of the project: content work, not code
- — Schema governance drifts without an owner; new products land without facet values and filters quietly degrade
- — A dedicated search stack is infrastructure you now operate; reserve it for catalogs that earn it
- — ~$3,000–$6,000/yr upkeep (Deploi estimate, illustrative)
What Merchants Say
The recurring shape in filter-app complaints: filters render fast but show junk facets, because the underlying product data was never structured. The app gets blamed for the catalog's gaps.
Merchants report filter and search widgets breaking after theme updates or at specific breakpoints, turning every theme release into a regression test.
If You Change Your Mind Later
If you bought and outgrow it
Your metafields stay in Shopify, so leaving a filter app costs config, not data: facet mappings, synonyms, and merchandising rules get rebuilt in the next engine, a few weeks of work rather than a migration. If you skipped the metafield work and leaned on the app's own tagging, the exit is far worse, because the structure leaves with the vendor.
If you built and want out
Nothing strands. The schema is plain Shopify metafields that any future engine reads: native, a commercial app, or a dedicated stack. You can change the render layer three times without touching the data, and that portability is the argument for doing the data work first, whatever you decide about engines later.
When This Answer Changes
We're watching for:
- ▸ Shopify raising Search & Discovery's practical SKU ceiling or adding merchandising rules; the free first-party app keeps absorbing the category's floor
- ▸ Further vendor consolidation after Klevu and Searchspring merged into Athos Commerce (Jan 2025); re-check vendor health before signing any annual contract
- ▸ AI shopping surfaces leaning harder on structured product attributes, which raises the value of the owned-schema path (2026 research corpus theme)
Verdict change log:
No changes since first publication (August 2026).
Common Questions
Are Shopify's native collection filters good enough?
Yes, up to a point. The free Search & Discovery app renders metafield-based filters cleanly and holds up to roughly 1,000 SKUs (July 2026 research). The ceiling isn't rendering; it's the attribute model, because native filters only look as good as the metafields behind them. Past about 10,000 SKUs, a third-party engine typically pays back in 6–9 months (July 2026 research).
Why do our collection filters show useless or missing options?
Because filters can only facet attributes your products actually carry. When size, material, or compatibility live in unstructured descriptions instead of metafields, every engine (native or paid) renders thin, inconsistent facets. The fix is data modeling: define facet definitions per category, backfill values across the catalog, then let your chosen engine read them. That's why swapping filter apps so rarely fixes filter quality.
Do we need to replace our search app to get better filters?
Usually not. Deploi has delivered metafield-source search filters for a mid-market Shopify brand without a platform swap; the schema and backfill did the heavy lifting while the existing engine stayed in place. Start with the data model and re-evaluate the engine afterward. You'll either find the current one was fine or make a cleaner, cheaper switch with fully portable attributes.
Your Next Steps
If you're going with CUSTOMIZE(matches your selected profile)
- Audit your top 10 collections: list the attributes shoppers narrow by and check which exist as clean metafields
- Define facet definitions per category (size, material, compatibility, ingredients) and set allowed values
- Backfill with Shopify bulk operations; assign an owner for attribute hygiene on new products
- Render through native Search & Discovery first; add a commercial engine only for merchandising rules, synonyms, or analytics you can name
- Track zero-results searches and filter usage from day one; they're your schema backlog
If you're going with WAIT
- Stay on the free Search & Discovery app and configure metafield filters; the render layer is covered at your size
- Do the same schema work the customize path prescribes, since it's what you'd feed any future engine anyway
- Cap app installs: skip the filter subscription until facets are data-limited, not render-limited
- Diary a re-decision at roughly 1,000 SKUs or when merchandising rules become a named need (July 2026 research ceiling)
Official Docs & Sources
- Storefront search (covers the Shopify Search & Discovery app) — Shopify Help Center
- Storefront API reference — shopify.dev
Official documentation linked for verification — our verdicts and estimates are our own.
Related Decisions
Athos Commerce vs. Boost AI Search & Filter: Which Engine?
Boost wins the self-serve mid-market zone; Athos Commerce wins for quote-led merchandising depth. The metafield facet schema comes first either way.
Boost AI Search & Filter vs. Native Filters: Worth the Fee?
A metafield-first customize play beats defaulting to Boost: native renders clean facets free to ~1,000 SKUs, and Boost pays back above ~10K or for filter-led UX.
Should You Build or Buy Site Search on Shopify?
Site search on Shopify splits by catalog size: native to ~1,000 SKUs, buy in the middle, build at big-catalog, search-led scale.
Should You Build or Buy Merchandising Rules on Shopify?
Merchandising rules are a BUY once a merch team trades collections weekly.
Should You Build or Buy a Mega Menu on Shopify?
A theme-level mega menu wins for mid-market Shopify stores with any dev capacity.
Ready to fix filters at the data layer?
Deploi has delivered metafield-source filters without a platform swap and built a self-hosted faceted search stack (Eluma). We'll scope the schema before anyone talks subscriptions.
Contact us todayVerdict scored for the reference scenario above. Estimates are not quotes; app pricing carries its verification date and gets re-verified quarterly. Full scoring anchors: see the TCC methodology.
Read how we score these decisions (the TCC Framework). No affiliate links, no paid placement — no app vendor pays to appear here.