Should You Build or Buy Site Search on Shopify?

Written by Deploi EditorialReviewed by Martin Dejnicki, Director of SEO & AI SearchUpdated August 2026Pricing verified July 2026 research corpus (re-verify)

Site search on Shopify splits cleanly by catalog size: native Search & Discovery covers you free to roughly 1,000 SKUs, a mid-tier search app earns its fee in the middle, and above ~10,000 SKUs (where third-party search pays back in 6–9 months per July 2026 research) a self-hosted build starts winning if search leads your UX. Score your catalog before you score your budget.

Your profile — see how the verdict shifts

VerdictWAIT under ~1K SKUs · BUY mid-catalog · BUILD at scale + search-led UX
Buy score
6.9
Build score
6.6
Confidence
HighThe boundary is measurable: SKU count, zero-result rate, and search's revenue share decide it, and the native ceiling is well documented
Reference scenario
$20M–$100M GMV · 10K–30K SKUs · search in the top-3 conversion paths
As of
August 2026

Decision at a Glance

Your profileVerdictWhy
Under ~1,000 SKUsWAITNative Search & Discovery (free) handles synonyms, boosts, and filters at this size. Spend nothing here until the catalog or your zero-result rate grows.
~1,000–10,000 SKUsBUYNative strain shows first in filters and relevance tuning. A mid-tier app fixes both fast; full enterprise search only pays back in 6–9 months above ~10K SKUs (July 2026 research), so buy modestly.
~10,000–100,000 SKUsDEPENDSThird-party search pays back in 6–9 months at this size (July 2026 research). Buy if search is a utility; build once ranking control and search-led browsing become your edge.
100,000+ SKUs, or search-led UXBUILDRecord- and query-based app pricing scales with your catalog while an Elasticsearch stack's cost stays mostly flat, and the ranking logic becomes yours. This is the lane Deploi's Eluma build sits in.

What Site search Actually Drives

OutcomeImpactHow it works
Revenue — directHighSearch users arrive with declared intent, so every relevance gain acts on the visitors most ready to buy; each fixed zero-result query is a recovered path to checkout.
Customer experienceHighTypo tolerance and honest filters decide whether a shopper finds the product in one query or leaves; on a large catalog, search effectively is the navigation.
Data & insightHighQuery logs and zero-result terms are unfiltered customer language: what people want, what they call it, and what you don't stock yet. Who owns that stream is a core stake in this decision.
Revenue — indirectMediumSearch terms feed merchandising, buying, and SEO: zero-result queries surface assortment gaps, and top queries tell you what belongs on the homepage.
Operational efficiencyMediumSynonym and merchandising management either lives in a dashboard your merch team drives or becomes a standing dev ticket; the tooling choice sets the ongoing labor cost.

Spend ceiling: Size the spend to search's share of revenue, not to catalog vanity. If search touches a third of orders, an enterprise tier or a build is proportionate; if it's a utility on 800 SKUs, free native is the ceiling.

What buying enables (top apps)

  • + Live inside a week: typo tolerance, synonyms, filters, and merchandising rules without touching infrastructure
  • + Vendor-run relevance models that keep improving without your engineering time
  • + Merchandising dashboards, searchandising rules, and A/B testing your merch team can drive from day one
  • + Search analytics out of the box: zero-result reports and query trends from install

What building additionally unlocks

  • + A ranking recipe apps don't expose: weighted scoring on your own signals (margin, stock depth, newness) with a dashboard your team tunes; a DTC beauty brand runs exactly this
  • + Flat-cost economics at scale: no per-record or per-query meter as the catalog grows
  • + Query and click data in your own warehouse, feeding AI and personalization work with no export ceiling
  • + Server-rendered results with zero third-party script weight on the storefront

Find Your Verdict in 3 Questions

  1. Is your catalog under roughly 1,000 SKUs?

    Yes: Your verdict: WAIT — free native Search & Discovery covers this size; revisit when the catalog or your zero-result rate grows.

    No: Go to question 2.

  2. Are you past roughly 10,000 SKUs, or is search-led discovery core to your storefront UX?

    Yes: Go to question 3.

    No: Your verdict: BUY — a mid-tier search app fixes filters and relevance fast, and the fee stays proportionate at this catalog size.

  3. Do you have dev capacity and appetite for a permanent relevance-tuning habit?

    Yes: Your verdict: BUILD — own the ranking logic and the query data; the subscription you skip helps fund it.

    No: Your verdict: BUY — take the strongest tier your catalog justifies, and diary a re-decision at each renewal as record pricing grows.

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 →

DimensionBuyBuildWhy
Cost
Acquisition & implementationAn app indexes your catalog in days; an Elasticsearch stack with real-time sync is an estimated 8–14 weeks (Deploi estimate, illustrative).
Recurring feesSearch apps price on records, queries, or both, so the bill grows with catalog and traffic; a build pays mostly flat hosting plus upkeep.
Maintenance & upgradesThe vendor absorbs relevance-model and index maintenance; a build carries ~15–20% of build cost per year (Deploi estimate) for sync, cluster, and tuning upkeep.
Switching & exitProducts re-index from Shopify anywhere, but synonym lists, boosts, and merchandising rules rebuild by hand when you leave an app; a build's schema and ranking logic stay yours.
Risk
Vendor riskThe category consolidates: Klevu and Searchspring merged into Athos Commerce in January 2025, and roadmaps reshuffle after mergers. A build has no vendor to lose.
Security & compliance surfaceSearch queries carry little sensitive data either way; an app adds one more data processor, a build adds infrastructure you patch.
Platform-deprecation exposureApps track Shopify's API cycles for you; a self-hosted sync must absorb version bumps (~every 6 months) and cost-based rate limits itself.
Value
Fit to requirementTop apps cover synonyms, merchandising rules, and A/B testing well; a build adds your exact ranking recipe, like weighted scoring tuned to margin or stock depth.
Time to marketDays versus roughly a quarter: speed stays the app path's honest headline advantage.
Performance & scaleManaged engines are quick, but injected widgets carry script weight (the app-bloat page-speed tax is a documented recurring pattern); a server-rendered build answers without the third-party payload.
Data ownership & AI-readinessQuery logs, zero-result terms, and click signals are merchandising gold. Owned, they feed your ranking and AI work with no export ceiling; rented, they live behind the vendor's dashboard.
Focus & opportunity costRelevance is a permanent tuning habit, not a ship-and-forget build; take it on only when search is a top conversion path.

The App Landscape

AppStatusPricingBest for
AlgoliaLiveAPI-first hosted search and faceting with a strong headless storyUsage-basedSpeed-critical, large catalogs with dev support on hand
Athos CommerceLiveKlevu and Searchspring merged into Athos Commerce (Jan 2025); evaluate the combined roadmap, not the legacy brandsQuote-basedMerchandising-heavy category pages and searchandising rules
Boost AI Search & FilterLiveThe category's best-known filter-tray and search combo on ShopifySKU/session-tieredFilter-rich stores stepping up from native
NostoLiveSearch inside a broader personalization suiteQuote-basedStores already invested in Nosto personalization

The Build Path

  • Elasticsearch + real-time catalog sync: The Eluma pattern: Elasticsearch behind the storefront with webhook-driven sync from Shopify, typo tolerance, and faceted filters. Search stops being a widget and becomes infrastructure you tune.
  • Weighted scoring + merchandising dashboard: A ranking recipe built on your signals (margin, stock depth, newness) with an admin dashboard so merchandisers adjust weights without a deploy. A DTC beauty brand runs exactly this stack today.
  • Metafield-source filters, no platform swap: The smaller move: build the filter layer from metafields on top of your current engine. Deploi has delivered this as a standalone project without changing search platforms.
Effort band
$35,000–$90,000 build (Deploi estimate, illustrative); most scopes land in the $25–75K contact-form band, search-led rebuilds reach $75K+
Typical timeline
8–14 weeks (Deploi estimate, illustrative)
Maintenance, honestly
~15–20% of build cost per year (Deploi estimate): index-mapping updates, cluster patching, sync fixes as Shopify's API versions cycle, and the ongoing relevance tuning that makes search good. There is no per-query bill.
What you own — and what you take on
You own: the ranking logic, the query and zero-result data, the filter architecture, and the dashboard your merchandisers tune. You take on: cluster uptime, sync integrity against Shopify's rate limits, and a permanent relevance-tuning habit.

3-Year Total Cost of Capability

Buy (app path)Build (custom path)
Year 0 (setup)$1,000–$5,000 (onboarding + tuning)$35,000–$90,000
Years 1–3 (recurring)$18,000–$90,000 (subscription)$16,000–$54,000 (maintenance)
3-year total≈$19,000–$95,000≈$51,000–$144,000
Illustrative cumulative cost over 36 months$0$26k$51k$77k$103kMo 0Mo 12Mo 24Mo 36Buy (app path)Build (custom path)
Illustrative cumulative cost at mid-band pricing: the app line stays lower through year 3 for a mid-size catalog. Push record and query volume toward the top band and the lines cross inside two years; that inversion, plus owning the ranking logic and the query data, is the whole build case.
  • All figures illustrative samples for the reference scenario — not quotes, not verified pricing.
  • App path: mid-tier record/query pricing held flat; real bills scale with catalog and traffic, which is conservative for the build case.
  • Build includes Elasticsearch cluster, real-time sync, faceted filters, and a merchandising dashboard; three-year horizon.

What the Sticker Price Hides

On the buy path

  • Record- and query-based pricing grows with your catalog and traffic; the bill you sign is the smallest it will ever be (community-reported pattern)
  • Front-end search widgets add script weight; the app-bloat page-speed tax is a documented recurring pattern (July 2026 research)
  • Synonyms, boosts, and merchandising rules accumulate inside the vendor's dashboard, so leaving means rebuilding your tuning by hand
  • Post-merger roadmaps shift: Klevu and Searchspring became Athos Commerce in January 2025, and consolidation can reshuffle plans and pricing (verify at renewal)

On the build path

  • Real-time sync against Shopify's cost-based API rate limits is the fiddly 20%; THROTTLED errors arriving inside a 200 response are a documented dev trap
  • Relevance is a permanent tuning job, so budget merchandiser hours, not just engineering hours
  • ~15–20% of build cost per year in upkeep (Deploi estimate); skip it and the symptom is stale indexes and drifting results
  • Scope creep from search into a personalization platform doubles the build; hold v1 at ranking plus filters

What Merchants Say

The pricing-creep complaint shape: search apps feel affordable at signup, then record counts and query volume push the plan tier up right as the store starts growing.
community-reported pattern
Search and filter widgets breaking at theme updates and responsive breakpoints is a recurring gripe; the storefront layer is where the buy path gets fragile.
app-store 1–2★ review theme

If You Change Your Mind Later

If you bought and outgrow it

Your catalog re-indexes from Shopify anywhere, so products are never stranded. What you lose is the accumulated tuning: synonym lists, boosts, merchandising rules, and the query-analytics history. Export what your plan allows, document the rule logic as you go, and treat each renewal (especially after a vendor merger, like Athos) as a natural re-decision point.

If you built and want out

An Elasticsearch stack is portable infrastructure: the schema, ranking logic, and query logs move with you to any host, and retreating to an app later is a re-integration job, not a data rescue. The realistic exit cost is decommissioning a cluster and re-pointing the storefront; nothing accumulates inside a vendor you can't take with you.

When This Answer Changes

We're watching for:

  • Shopify raising the native Search & Discovery ceiling: semantic search or richer filters on all plans would push the WAIT boundary well past ~1,000 SKUs (re-verify quarterly)
  • Athos Commerce consolidation settling: post-merger pricing and roadmap clarity could strengthen or weaken the BUY lane (merger closed January 2025)
  • AI-native search entrants resetting mid-tier pricing, which would shift the payback math that currently favors third-party search only above ~10K SKUs (July 2026 research)

Verdict change log:

No changes since first publication (August 2026).

Common Questions

Is Shopify's native search good enough?

Up to roughly 1,000 SKUs, yes: the free Search & Discovery app covers synonyms, boosts, filters, and basic recommendations, and most stores under that line lose nothing by waiting. Past it, strain shows up as zero-result queries and clumsy filtering. Watch your zero-result rate in search analytics; when it climbs while the catalog grows, that's your upgrade signal.

When does building site search beat buying an app?

Two conditions, ideally together: a large catalog (third-party search pays back in 6–9 months above ~10,000 SKUs, per July 2026 research) and a search-led UX where ranking is a competitive edge. At that point record-based app pricing scales against you while a self-hosted Elasticsearch stack stays mostly flat, and the ranking recipe becomes yours. Deploi's Eluma build is exactly this pattern.

What does a custom Shopify search build cost?

An estimated $35,000–$90,000 one-time for the Elasticsearch pattern with real-time sync, typo tolerance, faceted filters, and a merchandising dashboard, plus ~15–20% of that per year in upkeep (Deploi estimate, illustrative). Smaller moves exist too: a metafield-source filter build upgrades filtering without swapping engines. Compare against three years of subscription at your record count, not against the sticker price.

Your Next Steps

If you're going with BUY

  1. Pull 90 days of search analytics first: zero-result rate, top queries, and search-to-purchase conversion
  2. Shortlist by catalog size: mid-tier apps under ~10K SKUs, enterprise tiers above (payback runs 6–9 months there, July 2026 research)
  3. Price the renewal, not the signup: model record and query growth two years out
  4. Test PDP and collection speed before and after the trial install; cap script weight
  5. Diary a re-decision at every renewal and after vendor mergers like Athos

If you're going with BUILD

  1. Baseline search's revenue share and zero-result rate now; that's the ROI denominator
  2. Scope v1 to ranking plus filters: Elasticsearch, real-time sync, typo tolerance, faceted filters
  3. Design the sync for Shopify's cost-based rate limits from day one, including THROTTLED responses
  4. Give merchandisers a weights dashboard so tuning never needs a deploy
  5. Stand up query-log dashboards at launch; owned search data is half the payoff

Official Docs & Sources

Official documentation linked for verification — our verdicts and estimates are our own.

Ready to fix what shoppers can't find?

We work both lanes: honest app selection and integration when buying wins, and the Eluma-style Elasticsearch build when your catalog and UX have outgrown the tiers. Bring your SKU count and 90 days of search analytics; the verdict falls out fast.

Contact us today

Ecommerce development at Deploi

Verdict 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.

No affiliate links. No paid placement. We make money building and integrating solutions — not on referral fees.