Skip to main content
🔍

Search Checklist for AI-Built Apps

Find content across your app

When you vibe code search with tools like Cursor, Lovable, Bolt, v0, or Claude Code, the generated code often works in development but misses critical production requirements. This checklist helps you catch what AI missed before you ship.

Danger Zone

moderate risk

Bad search is invisible until people start looking for things — then it becomes the reason they leave

Search feels simple when you're testing with 10 items. Type a word, see matching results. But good search needs to handle typos ("iphone" should find "iPhone"), understand that "laptop" and "computer" are related, rank results so the best ones come first, stay fast when you have 100,000 items, and work across multiple fields at once. The difference between basic matching and actual search is bigger than it looks.

Failure scenario

You launch a marketplace with 5,000 products. Search works fine in testing. Six months in, someone searches for "wireles headphons" and gets zero results even though you have 200 wireless headphones. They try "bluetooth earbuds" — still nothing because your product titles say "headphones." Meanwhile your competitor's search just works. You never hear from that person again.

Common mistakes

  • Exact matching only — one typo returns zero results instead of guessing what they meant
  • Searching only titles or names, missing descriptions, tags, or other relevant fields
  • No ranking logic — a barely-related match shows up before the perfect match
  • Search gets slower and slower as you add more items until it times out
  • Searching for "new york" requires the exact phrase in that exact order instead of finding "york, new" or documents with both words

Time to break: 3-6 months as your catalog grows and search expectations rise

How are you building this?

Showing what to check when using a managed service

Audit Prompts

Copy these into your AI coding assistant to check your implementation.

Does search handle typos and variations?
performance
Test our search with common mistakes and variations. Try misspellings ("wireles headphons"), different word orders ("york new" vs "new york"), singular vs plural ("shoe" vs "shoes"), and related terms ("laptop" when the item says "computer"). Does it still find the right things? Are results ranked sensibly — best matches first?

People don't type perfectly. If one typo breaks your search, you're losing customers who would have bought if they could find what they're looking for.

Is search actually fast?
performance
Test search speed under realistic conditions. Search for common terms and see how long results take to appear. Try it with your current data size and imagine 10x more items — does your search service have limits on how many records you can search? Are you using the service's features to pre-index content, or searching the raw database every time?

People expect search results instantly — within 100 milliseconds. If it takes even one second, it feels broken.

Are you searching the right things?
performance
Check what fields are actually being searched. Are we only searching titles/names, or also descriptions, tags, categories, and other relevant content? When someone searches for a product feature or benefit, does it find items that have that in the description? Can filters (like price range or category) be combined with search?

Searching only titles is like only reading the first sentence of every page. You're missing most of the information that would help someone find what they need.

Will search break as you grow?
reliability
Check our search service plan limits. How many records can we search? How many searches per month? What happens if we hit those limits — does search just stop working? Are we close to any limits now? Does our plan include features like typo tolerance and ranking, or do we need to upgrade for those?

Search services have tiers. You might be on a free plan that works great for 1,000 items but silently breaks or gets expensive at 10,000.

Checklist

0/8 completed

Smart Move

It depends

It depends on what you're searching and how many items. For a directory with 50 entries, database search is fine. For an e-commerce site with 5,000 products, a search service like Algolia pays for itself in customers who can actually find things. The breaking point is around 1,000-5,000 items or when people start expecting instant, typo-tolerant search.

Algolia

The gold standard — handles typos, ranking, instant results, and provides ready-made UI components

10,000 searches/month and 10,000 records free

Typesense

Open source alternative you can self-host or use their cloud — similar features to Algolia

Self-host free, cloud starts at $0.03/hour

Meilisearch

Another solid open-source option with good typo tolerance and fast results

Self-host free, cloud plans available

Tradeoffs

Search services cost money at scale (Algolia can get expensive past the free tier). Self-hosting is cheaper but means you manage infrastructure. Database full-text search is free but limited. Start simple, upgrade when you feel the pain.

Did you know?

43% of website visitors go straight to the search bar, and those searchers are 2-3x more likely to convert than people who browse — but only if search actually works.

Source: Forrester Research, E-commerce Search Study

Related Checks

📊Analytics

Track what users do

🤖AI Features

Add LLM/AI capabilities

🩺Monitoring

Error tracking and observability