Docs/FAQ

FlowSearch FAQ

Public answers to common setup, search, AI, analytics, billing, and troubleshooting questions.

Common workflows

Get search live

  1. Add your website in the dashboard and confirm the production or staging domain.
  2. Run a crawl so FlowSearch can index your published pages and CMS content.
  3. Install the Search Bar Script on all pages and the Results Page Script only on the search results page.
  4. Publish your website and test from the public production or verified staging URL.

Improve result quality

  1. Use clear page titles, meta descriptions, headings, and page copy.
  2. Review zero-result analytics to identify missing content and terminology gaps.
  3. Adjust include and exclude patterns when pages are missing or irrelevant pages appear.
  4. Re-crawl after publishing major content changes.

Use AI features safely

  1. Enable AI Summary with data-summary="true" on the Results Page Script.
  2. Use paid plans for follow-up conversations and External Knowledge.
  3. Use Business or Enterprise for uploaded document knowledge.
  4. Review source labels: pages, documents, and General AI knowledge when External Knowledge is enabled.

Diagnose issues

  1. Open DevTools and check the Console and Network tabs.
  2. Confirm window.FlowSearch exists on published pages.
  3. Check that data-fs-search, data-fs-input, data-fs-results, and result template attributes are present.
  4. Include console errors, request status codes, website domain, and affected query when contacting support.

Setup and indexing

What is FlowSearch?

FlowSearch adds fast, typo-tolerant search to websites. It crawls your published content, indexes it in Typesense, serves results through the FlowSearch widget and API, and can generate AI summaries from your website content.

How do I install FlowSearch on a website?

Add and verify the website, run a crawl, paste the Search Bar Script into the site-wide head/custom-code area, create a results page, then paste the provided HTML and Results Page Script there.

Why do you recommend /search-results instead of /search?

The default keeps setup consistent and avoids common conflicts with existing search pages. You can use another path if data-results-page points to it.

Does FlowSearch work in a private editor preview?

Only when the preview has a public HTTPS URL returning standard HTML. Private, password-protected, localhost, and editor-only previews are not supported.

Which script goes on every page?

The Search Bar Script, widget.js, goes in site-wide Head Code. It powers search forms, live search, keyboard behavior, and navigation to the results page.

Which script goes only on the results page?

The Results Page Script, results.js, goes only in the Head Code for your search results page. It renders results, pagination, optional AI Summary, sources, and paid follow-up chat.

What do data-fs-* attributes do?

They tell FlowSearch which page elements to enhance. Common attributes include data-fs-search, data-fs-input, data-fs-results, data-fs-result, data-fs-url, data-fs-title, data-fs-snippet, and data-fs-summary.

How does FlowSearch find pages to index?

FlowSearch uses your sitemap when available and applies your include and exclude patterns. It indexes published pages and CMS content from the completed crawl.

Why are some pages missing from search results?

The page may not be published, may be missing from the sitemap, may be excluded by a pattern, may be blocked by crawl behavior, or may need a fresh crawl after recent edits.

How long does indexing take?

It depends on site size, crawl limits, network conditions, and the search cluster. Small sites usually finish quickly, while larger sites and scheduled crawls take longer.

Keys, rate limits, and search performance

What is a Search Key?

A Search Key is the public fs_pk key used by the browser widget and public search endpoints. It is safe to use client-side, but should still be restricted to your allowed origins.

What are origin restrictions?

Origin restrictions limit which domains can use your Search Key. Add the production origin and any verified staging origin used for testing.

Can I regenerate a key?

Yes. Regenerating a key invalidates the old key immediately, so update the scripts on your website before or right after regenerating.

Why am I seeing SEARCH_RATE_LIMIT_EXCEEDED?

The site exceeded the per-minute search request or concurrency limit for its current plan. Reduce live-search frequency, lower per-page requests, or upgrade to a higher-performance plan. Monthly quota issues use MONTHLY_SEARCH_QUOTA_EXCEEDED instead.

Why does search speed differ by plan?

Search speed depends on cluster resources, region proximity, index size, query options, live-search volume, Supabase validation, and serverless cold starts. Higher plans can use faster shared or dedicated search infrastructure.

How do I enable AI Summary?

Add data-summary="true" to the Results Page Script and include a data-fs-summary block on the results page. Initial AI summaries are available on all plans.

Can visitors ask follow-up questions?

Yes, but follow-up AI conversations are available only on paid plans and are capped at 20 user turns. Free sites still receive the initial AI Summary when AI Summary is enabled and monthly AI credits are available.

What is External Knowledge?

External Knowledge lets the AI add brief general knowledge when it directly helps answer the question. Website content and Business+ uploaded content are still prioritized first. This is a paid-plan feature.

AI answers and document knowledge

Why does General AI knowledge appear as a source?

When External Knowledge is enabled, FlowSearch labels General AI knowledge as a source so visitors can see that the answer may include wider model knowledge beyond your website pages.

Can the AI ask visitors questions back?

No. FlowSearch instructs AI Summary responses to answer directly and not end with assistant questions or prompts like 'Would you like...'.

Which plans can upload documents?

Uploaded PDF, Markdown, and text document knowledge is available only on Business and Enterprise plans. Basic and Pro can use follow-up chat and External Knowledge, but not uploaded document knowledge.

How are uploaded documents used?

Business and Enterprise document text is split into searchable chunks, embedded for semantic matching, and selected as relevant context for AI Summary answers. Older documents are upgraded lazily the next time AI Summary needs them, so the first answer after an upload or upgrade can take a little longer. Documents are not included for lower plans, even if old records exist.

What analytics does FlowSearch track?

FlowSearch records search queries, result counts, search timing, sessions, visitors, result clicks, AI summary generation, and AI helpfulness feedback where available.

What are zero-result searches?

Zero-result searches are queries where visitors found no matching content. They are useful content-gap signals and can guide new pages, rewritten copy, synonyms, or navigation improvements.

Where do I manage billing and plans?

Open Dashboard > Settings > Billing & Subscription. Plan changes affect website limits, crawl schedules, search performance, AI features, API access, and document upload availability.

What should I include when contacting support?

Include your website URL, affected query, browser, plan, page URL, console errors, network status codes, and whether the issue happens on staging or production.

What should I check first if FlowSearch is not working?

Confirm the site is published, scripts are in the correct Head Code locations, the Search Key is current, allowed origins include the current domain, the crawl completed, and data-fs attributes exist on the page.

Analytics, billing, and support

How do I contact support?

Use the contact form on this Help & Support page. It saves your request and notifies support@flowsearch.io. You can also email support@flowsearch.io directly.

Need more help?

Logged-in users can open Dashboard Help & Support. Public visitors can also review Troubleshooting or contact FlowSearch.