Adobe Commerce Live Search runs on a Hyva storefront through the Hyva Enterprise Sensei compatibility modules. Live Search is a SaaS engine powered by Adobe Sensei that replaces your Elasticsearch or OpenSearch cluster on Adobe Commerce. It adds semantic autocomplete, machine-learning ranking, and built-in analytics, but it requires an Adobe Commerce license, a Hyva Enterprise license, and version 2.4.4 or higher.
Search is where a lot of revenue quietly leaks. A shopper who searches has high intent, and a weak result set sends them to a competitor. Live Search is Adobe’s answer, and Hyva is the frontend most performance-focused Adobe Commerce merchants run. This guide covers how the two fit together, what the AI ranking genuinely does versus what the marketing implies, and the honest cases where a self-hosted OpenSearch cluster is still the better call.
What Live Search actually is
Live Search is a cloud-hosted search service that Adobe runs for you. Instead of indexing your catalog into an Elasticsearch or OpenSearch cluster you maintain, your catalog syncs to Adobe’s SaaS, and search queries are answered over GraphQL from Adobe’s infrastructure. The engine is powered by Adobe Sensei, Adobe’s machine-learning layer.
That architecture has three consequences worth stating plainly:
- You stop running and tuning a search cluster. No more sizing nodes, patching, or watching heap usage.
- Live Search is available on Adobe Commerce, not Magento Open Source. Open Source stores stay on OpenSearch.
- Live Search cannot run alongside OpenSearch or Elasticsearch on the same store. It is a replacement, not an add-on.
The feature set includes Sensei semantic autocomplete, configurable facets, merchandising rules, dynamic synonyms, and integrated search analytics that show you what shoppers look for and where results fall short.
The AI ranking model, without the hype
The phrase “AI-powered search” gets thrown around loosely, so here is what Sensei ranking does in concrete terms. It observes shopper behavior across your store, click patterns and purchase patterns tied to search queries, and it uses those signals to reorder results so that products people actually click and buy rise toward the top.
In practice this means high-margin and best-selling products surface earlier for a given query, without a merchandiser hand-ranking every term. The model adapts as behavior changes, so a product that starts selling well climbs on its own.
What it does not do is read your mind or index sources outside your Commerce catalog. It cannot rank a blog post, a PDF spec sheet, or an external inventory feed. Its intelligence is bounded by the behavioral data your store generates, which means a low-traffic store gives the model less to learn from than a high-traffic one. That is the single most important expectation to set: Sensei ranking rewards stores with enough search volume to produce a signal.
How Live Search renders on Hyva
Here is the part specific to Hyva. A Hyva theme does not use the Luma frontend that Adobe’s default Live Search widgets assume. So Live Search results, the search-as-you-type popover, and product recommendations need a Hyva-native rendering layer. That layer is Hyva Enterprise.
Hyva Enterprise extends the Hyva theme and Hyva Checkout with compatibility modules for Adobe Commerce enterprise features, and the Sensei module is the one that integrates Live Search, Product Recommendations, and behavioral event tracking into a Hyva storefront. The order of installation matters: configure the Adobe Live Search modules first, then install the Hyva Enterprise Sensei compatibility on top.
The requirements as of 2026:
| Requirement | Minimum | Notes |
|---|---|---|
| Platform | Adobe Commerce 2.4.4+ | 2.4.6 is the tested target; 2.4.4 and 2.4.5 are less validated |
| Hyva Themes | 1.2.9 or 1.3.10+ | Older theme versions are not supported |
| Hyva Enterprise license | Required | Separate from Hyva Commerce and Hyva Checkout licenses |
| Adobe Commerce license | Required | Live Search is not on Magento Open Source |
Getting this stack right is the same kind of integration discipline that a serious Hyva development engagement brings to any Adobe Commerce enterprise feature: install in the right order, verify events fire, and test the storefront rendering before launch.
Live Search versus OpenSearch: an honest comparison
Adobe would prefer you always choose Live Search. Reality is more nuanced. Here is where each wins.
| Factor | Adobe Live Search | Self-hosted OpenSearch |
|---|---|---|
| Infrastructure | None, Adobe hosts it | You run and tune the cluster |
| Ranking intelligence | Sensei behavioral ML, automatic | Manual relevance tuning, full control |
| Customization | Bounded to Commerce catalog | Index anything, custom scoring |
| Platform | Adobe Commerce only | Open Source and Commerce |
| Analytics | Built in | Build your own |
| Cost model | Included with Commerce | Server and ops cost |
| Multi-source data | No | Yes |
Choose Live Search when you are on Adobe Commerce, have enough search traffic for the model to learn, and want to stop running a cluster. Choose OpenSearch when you need to index non-catalog sources, require a specific custom ranking algorithm the Live Search rule engine cannot express, or run Magento Open Source where Live Search is not an option. Both are legitimate; the deciding factors are platform edition, customization depth, and whether you want to own search operations.
Setup sequence that avoids rework
Teams that rush Live Search onto Hyva tend to hit the same avoidable problems. This order prevents them:
- Confirm edition and versions. Verify you are on Adobe Commerce 2.4.4 or higher and a supported Hyva theme version before buying anything.
- Install and configure Adobe Live Search first. Get the catalog syncing and search working on the Adobe side independently of Hyva.
- Add Hyva Enterprise Sensei compatibility. Install the compatibility module so results, autocomplete, and recommendations render in the Hyva theme.
- Verify behavioral events. Confirm that click and purchase events are firing, because without them the ranking model has nothing to learn from.
- Tune merchandising rules and synonyms. Once data flows, set up synonyms for the terms your shoppers actually use and add merchandising rules for priority products.
- Watch the analytics. Use the built-in search analytics to find zero-result queries and fix them with synonyms or catalog changes.
That behavioral-event step is the one most often skipped, and it is the one that determines whether the AI does anything at all. This kind of end-to-end verification is core to how solid Magento development teams ship enterprise features.
Common problems and how to handle them
Three issues come up repeatedly.
The first is thin ranking on low-traffic stores. If your search volume is low, Sensei has little to learn from, and results can feel no smarter than a basic engine. The fix is patience plus strong merchandising rules to guide results while the model gathers data.
The second is zero-result searches for brand or model terms. Sensei does not automatically know that shoppers call a product by a nickname. Add dynamic synonyms for those terms, and check the analytics weekly for new zero-result queries.
The third is rendering gaps after a Hyva theme upgrade. Because Live Search rendering depends on the Hyva Enterprise Sensei module, a theme upgrade can outpace the compatibility module. Check the compatibility tracker before upgrading and test search on staging first.
Migrating from OpenSearch to Live Search
If you already run OpenSearch and want to move, treat it as a planned cutover, not a config toggle. The catalog needs to sync to Adobe’s SaaS, the storefront search components change, and any custom relevance logic you built into OpenSearch does not carry over.
Start by cataloguing what your current search does that is non-standard: custom scoring, boosted attributes, indexed fields beyond the default catalog, and any external data you fold into results. Each of those is a decision point, because Live Search may not express it. Boosted attributes usually map to merchandising rules. Custom external data usually does not map at all, and that is the signal to stay on OpenSearch.
Then run both in parallel on staging. Point staging at Live Search while production stays on OpenSearch, and compare result quality on your top hundred search terms. Pay attention to zero-result queries and to whether your best sellers appear where shoppers expect them. Only cut production over once the staging comparison holds up and behavioral events are confirmed firing.
The rollback plan matters too. Because Live Search and OpenSearch cannot coexist on one store, switching back is itself a cutover. Keep your OpenSearch configuration documented and your cluster available during the first weeks after launch, so a regression does not leave you scrambling.
Measuring whether Live Search is working
Do not judge Live Search on feel. Use the built-in analytics and your own funnel data. Track search conversion rate, the share of searches that return zero results, and the click-through position on result pages. A healthy rollout shows zero-result queries dropping as you add synonyms, and search-driven conversion holding or climbing as the ranking model matures. If those numbers stay flat after a couple of months on a store with real traffic, revisit your merchandising rules and event tracking before blaming the engine.
Frequently asked questions
Does Adobe Commerce Live Search work on a Hyva theme?
Yes, through the Hyva Enterprise Sensei compatibility module. It renders Live Search results, autocomplete, and product recommendations in the Hyva frontend. You need Adobe Commerce 2.4.4 or higher, a supported Hyva theme version, and a Hyva Enterprise license.
Is Live Search available on Magento Open Source?
No. Live Search is an Adobe Commerce feature. Magento Open Source stores continue to use OpenSearch. If you are on Open Source and want AI-style search, you would evaluate third-party search providers instead.
Can I run Live Search and OpenSearch at the same time?
No. Live Search replaces the OpenSearch or Elasticsearch engine for your store. You choose one search backend, not both. Plan the switch as a cutover with testing on staging.
Does Live Search cost extra on top of Adobe Commerce?
Live Search itself is included with Adobe Commerce. Running it on Hyva requires a separate Hyva Enterprise license in addition to your Hyva Commerce or Hyva Checkout licenses. Budget for the Hyva Enterprise license specifically.
Will Live Search improve conversions on a small store?
It depends on search volume. The Sensei ranking model learns from shopper behavior, so a high-traffic store sees more benefit than a low-traffic one. On a small store, strong merchandising rules and good synonyms matter more than the automatic ranking.
Where this fits
Live Search on Hyva is a strong pairing for high-traffic Adobe Commerce merchants who want smart search without running a cluster, provided you accept the Adobe Commerce and Hyva Enterprise license costs and set up behavioral tracking properly. For lower-traffic stores or those needing deep custom ranking, a tuned OpenSearch setup can still be the better engineering choice.
Bemeir is the first US-based Hyva partner and a full Adobe Commerce agency, with a broad technology partner network across search, personalization, and payments. We implement Live Search, Product Recommendations, and OpenSearch, and we also build on Shopify, Shopware, and BigCommerce when a different platform fits better. Learn more about Bemeir and our approach to Adobe Commerce enterprise features.
Useful external references: the Hyva Enterprise Sensei documentation, the Hyva Enterprise getting-started guide, and a broader Magento 2 AI search overview.
