No backend engineers. No complex setup. RayoSearch connects to your existing Azure AI Search index and delivers a fully working search widget with one script tag.
Works on any website platform
Follow these steps to go from zero to live search in under 15 minutes.
Azure AI Search (formerly Cognitive Search) is the backbone of RayoSearch. If you already have an index, skip to step 2. If not, create a free Azure account and set up a new search service — the free tier handles up to 50 MB of data and 3 indexes at no cost.
Your data can come from almost anywhere: a SQL database, CosmosDB, Azure Blob Storage, or a JSON file you upload directly. Azure's indexer can pull data automatically on a schedule, so your search results stay fresh without any manual work.
What you'll need:
Sign in to RayoSearch, create a new site, and paste your Azure endpoint URL, index name, and API key. RayoSearch validates the connection instantly and reads your index schema — no manual field configuration needed at this stage.
The connection is read-only. Your API key only needs query permissions — RayoSearch never writes to your index. Your data stays in your Azure subscription.
Tip: Use a read-only query key, not your admin key. In the Azure portal, go to your Search service → Settings → Keys → Manage query keys.
Map your fields (title, image URL, description, price) using dropdowns and see a live preview update in real time. Choose a display mode — modal popup, inline, or full-page results — and set colors to match your brand.
Enable filters for any facetable field in your index: category, price range, tags, date, or any custom attribute. Filters appear automatically in the widget with live result counts — no code required.
Copy your generated script tag from the dashboard and paste it before the closing </body> tag on your page. Search is live instantly — no build step, no npm, no server restart needed.
<script
src="https://cdn.rayosearch.com/widget.js"
data-site-id="site_abc123"
async></script>
</body>That's it. Your search widget is live. Users can open it with ⌘K or by clicking a search icon you've added anywhere on the page.
There are four common approaches to adding search to a website. Here's how they compare honestly.
| Solution | Setup Time | Cost at Scale | AI / Semantic | Managed Hosting |
|---|---|---|---|---|
| Self-built search | Weeks | Low | ✕ | ✕ |
| Algolia | Hours | High ($1k+/mo) | Partial | ✓ |
| Elasticsearch | Days | Medium | Partial | Self-managed |
| Azure AI Search + RayoSearch | 15 min | Low | Full (vector) | ✓ |
A basic self-built search is faster than it sounds — until you try to add typo tolerance, relevance ranking, semantic understanding, analytics, and mobile performance. Each feature adds weeks. Most teams spend 2–4 weeks on a version they regret six months later.
Search is also deceptively hard to get right. Users expect Google-quality results. They'll give you one try before they stop using search entirely.
Azure AI Search is enterprise-grade, with built-in vector search, semantic ranking, typo tolerance, and scalable pricing based on index size — not query volume. That means costs don't spike when you run a campaign or go viral.
RayoSearch sits on top as the frontend layer: the visual builder, the widget, and the analytics. You get AI search quality with a 15-minute setup time.
Most site search implementations fail not because of the technology — but because of these avoidable mistakes.
Burying your search bar inside a navigation menu reduces search usage by up to 50%. Users won't hunt for it. Place search prominently in your header — ideally as a persistent bar or a clearly visible icon that opens a modal. The ⌘K pattern works especially well for SaaS products and documentation sites.
10–15% of all search queries contain a spelling error. If your search does exact-match only, those users get zero results and bounce. A search solution with fuzzy matching and typo tolerance should handle "recieve," "mispelling," and "addres" automatically — without any configuration on your part.
A blank page that says "No results found" is a dead end. When a query returns nothing, show alternatives: popular searches, trending content, a suggestion to browse by category, or a "Did you mean?" prompt. Zero-result pages have the highest abandonment rate of any search outcome — make them useful instead.
You can't improve what you don't measure. Without analytics, you have no idea what your users are looking for, which queries fail, which results get clicked, or where users drop off. Search analytics data is some of the most actionable insight you can collect — it tells you exactly what your audience wants.
Showing the newest results first makes sense for a news feed. It makes no sense for a documentation site, product catalog, or knowledge base. Relevance ranking — which considers field weighting, semantic similarity, click data, and query context — dramatically improves result quality. Date should be a filter option, not the default sort.
Four metrics tell you almost everything you need to know about search health. Track these from day one.
| Metric | What it measures | Target | If below target |
|---|---|---|---|
| Click-through rate (CTR) | % of searches where a user clicks a result | > 60% | Results are irrelevant — improve ranking or field mapping |
| Zero-result rate | % of queries that return no results | < 5% | Add synonyms, improve typo tolerance, or expand your index |
| Reformulation rate | % of users who refine their query without clicking | < 20% | Results aren't matching intent — review top failing queries |
| Abandonment rate | % of sessions where user searched but left without clicking | < 35% | Both CTR and zero-result issues — audit your top queries |
Every search, click, zero-result event, and session is captured and shown in your analytics dashboard. No extra tracking code, no third-party tools, no configuration. You can see your top queries, your worst-performing queries, and click-through rates for any time range — from the moment search goes live.
Enterprise search quality. No backend engineers. No months of development. Just search that works.
Every other enterprise search solution requires server-side code to proxy requests, handle auth, and serve results. RayoSearch handles all of this from the widget itself — your API credentials are scoped to the widget and never exposed.
Shopify, WordPress, Webflow, Wix, Squarespace, Next.js, Gatsby, or a custom static site — if you can add a script tag before </body>, you can add RayoSearch. No platform-specific plugins or app installs.
Azure AI Search vector embeddings understand what users mean — not just what they type. "Comfortable shoes for standing all day" finds ergonomic footwear. Natural language queries that used to fail now convert.
Know exactly what your users search for, which results get clicked, and which queries return nothing — all in your dashboard from day one. No extra tracking code, no third-party tools, no additional configuration.
Debounced inputs, cached responses, and a globally distributed CDN deliver results in under 100ms. Users see results as they type — the experience feels local, not like a network request.
No CSS overrides. No JavaScript. The visual builder gives you full control over colors, layout, result templates, and field display — with a live preview that updates as you configure. Looks exactly right before you ever embed it.
The setup is the same. The payoff looks different depending on your site type.
Docs sites live and die by search. Users land on your docs with a specific problem — if they can't find the answer in 30 seconds, they open a support ticket or churn. Semantic search understands "how do I connect to the API" even when your docs say "API authentication setup."
Shoppers who use site search convert at 2–3× the rate of browsers — they know what they want. Faceted filters by size, color, price, and category let them narrow results without scrolling through hundreds of products. Rich product cards with images and pricing reduce friction from search to add-to-cart.
Content sites with hundreds or thousands of articles need search to stay navigable. Users searching for a specific topic should find the most relevant article — not the most recent one. RayoSearch's relevance ranking surfaces the right content based on meaning, not just keyword overlap.
Everything you need to know before adding search to your website.
Free during beta. No credit card required. Two fields and you're on the list.
"We added RayoSearch to our docs site in about 20 minutes. Our support ticket volume dropped 30% in the first two weeks — users were finally finding answers on their own."
Live in under 15 minutes. Free during beta. Works on any platform.
We're onboarding a small group of early users. Beta members get full access, hands-on onboarding support, and a direct line to shape the product roadmap.
We'll review your application and reach out within 2–3 business days.