OpenSearch Relevance Testing

Automated OpenSearch Regression Detection. No Manual Checks Required.

OpenSearch's Search Relevance Workbench only runs when someone opens it. Between checks, your ranking degrades — and nobody knows until a customer complains or a revenue report flags it. ReleGuard runs automatically, 24/7.

No credit card required
5 minute setup
Read-only access only

What a Caught Regression Looks Like

Instantly See What Changed — and What Broke

ReleGuard — Test Results
"running shoes" → Nike Air Max 90
MustBeInTopN · Position: 2 · Expected: ≤ 5
PASS
"wireless headphones" → Sony WH-1000XM5
ProductMustAppear · Found at position 1
PASS
"winter jacket" → Patagonia Down Sweater
MustBeInTopN · Position: 14 · Expected: ≤ 3 · REGRESSION
FAIL
"laptop stand" → NoZeroResults
NoZeroResults · Returned 23 results
PASS
Alert sent via Email and Slack — "winter jacket" regression detected, product dropped from position 3 → 14
The Automation Gap

OpenSearch's Built-In Tools
Require a Human Every Time

OpenSearch ships with the Search Relevance Workbench — a genuinely useful tool for comparing rankings side by side. But it only runs when someone opens it and clicks. There's no scheduling, no alerting, and no regression history. The moment you stop watching, you're flying blind.

Capability
Search Relevance Workbench
ReleGuard
Compare search results
Scheduled automated testing
Automatic regression alerts
Regression history & diffs
Runs without human intervention
CI/CD integration via API

The Search Relevance Workbench is a useful exploration tool. ReleGuard handles continuous regression detection — they serve different purposes.

Search Pipeline Changes Break Silently

You update a search pipeline processor — renormalize scores, add a filter, change a phase — and the pipeline returns 200 OK. But the result ordering shifts. Products that ranked #1 now rank #8. Nothing errors. Nothing alerts. Your customer sees wrong results.

Neural Search Model Updates Shift Rankings

You swap an embedding model or update ml-commons settings. Semantic similarity scores change across the board. Queries that matched perfectly now surface irrelevant results. The model works — it just doesn't rank the way it used to.

ISM Policy Transitions Corrupt Search Behavior

An Index State Management policy rolls an index to a new alias, changes replica counts, or triggers a force merge. The transition completes cleanly — but search routing changes and results shift. ISM logs show success. Your search quality silently degrades.

Plugin and Version Upgrades Alter Scoring

An OpenSearch version upgrade or plugin update changes default analyzer behavior, scoring algorithms, or query parsing. Your cluster health stays green. Your tests still pass at the infrastructure level. But your product rankings have shifted — and nobody notices for days.

Why Automation Matters

Manual Checks Miss Most Regressions.
Automated Ones Don't.

Nobody opens the Relevance Workbench after every deployment. Nobody runs a check at 2am when a catalog pipeline fails. That's exactly when regressions happen — and exactly when they go undetected longest.

3–14
Days undetected

How long a typical silent search regression runs before it surfaces as a customer complaint, support ticket, or revenue anomaly.

4–6 weeks
To build it yourself

The engineering cost to wire up scheduled OpenSearch relevance tests with alerting, history, and maintenance — that nobody owns six months later.

<1hr
Detection time with ReleGuard

Hourly tests on Professional and above. Or trigger on-demand from your CI/CD pipeline after every deploy, on any plan.

Discovery timelines are general estimates. Your actual exposure depends on deployment frequency and traffic volume.

Don't wait for a customer complaint to discover a regression.

Continuous Monitoring in 5 Minutes

Set up once. Runs forever. Alerts you the moment your OpenSearch results drift.

1

Connect Your OpenSearch Domain

Enter your OpenSearch endpoint and credentials — AWS-hosted or self-managed. ReleGuard only runs search queries, never modifies your indices or settings.

2

Define Relevance Tests

Write assertions like "product X must appear in top 5 for query Y" or "this query must return at least 10 results." These run automatically — no human required.

3

Get Instant Regression Alerts

When a test fails, you get an alert via Email, Slack, or Webhooks with exactly what changed and what broke — before a customer opens a support ticket or revenue takes a hit.

Real-World Protection

What ReleGuard Catches
for OpenSearch Teams

A custom scoring script update shifts rankings on your top revenue queries

Detects position changes on monitored products without anyone opening the Workbench

An AWS OpenSearch Service version upgrade changes default scoring behavior

Catches ranking drift by comparing results against pre-upgrade baselines automatically

A data pipeline update drops a field and zero-result queries spike on affected searches

NoZeroResults conditions fire immediately when monitored queries return empty results

Index mapping changes cause brand name queries to stop matching full-text fields

ProductMustAppear conditions validate that expected products still surface for critical queries

A synonym configuration push creates unexpected result cross-contamination

ProductMustNotAppear conditions catch when irrelevant products begin surfacing unexpectedly

Ready to catch regressions before your customers do?

The Build-or-Buy Question

You Could Build This Yourself.
Here's Why Teams Don't.

Build It Yourself

  • 4–6 weeks of senior engineering time
  • Build scheduling, alerting, history storage
  • Ongoing maintenance nobody owns
  • Silently stops working 6 months later

Use ReleGuard

  • Running in under 5 minutes
  • Scheduling, alerting, and history built-in
  • Alerts via Email, Slack, and Webhooks
  • Starts at $49/mo — less than 1 hour of engineer time

Built by a Search Engineer Who's Been There

ReleGuard was built by a Tech Lead who has led search implementation and optimization projects at companies ranging from $25M to $500M+ in revenue. After watching one too many silent search regressions cost teams days of debugging and thousands in lost conversions, the choice was clear: build the monitoring tool that should have existed all along.

10+ years in search engineering
Led search at $25M–$500M+ companies
Built from real production pain
FAQ

Frequently Asked Questions

Q Does ReleGuard replace the OpenSearch Search Relevance Workbench?

A

No — they're complementary. The Search Relevance Workbench is excellent for manual exploration and comparing ranking strategies interactively. ReleGuard handles continuous automated regression detection: it runs on a schedule, maintains historical baselines, and alerts you when results change. Use the Workbench for development; use ReleGuard for production monitoring.

Q Does it work with AWS OpenSearch Service (formerly Amazon Elasticsearch Service)?

A

Yes. ReleGuard connects to any OpenSearch HTTP endpoint. For AWS OpenSearch Service domains, you'll connect using your domain endpoint and the appropriate credentials. Both public and VPC-accessible domains are supported depending on your network configuration.

Q Does ReleGuard write to my OpenSearch index or modify my domain configuration?

A

No. ReleGuard only executes read-only search queries. It never writes documents, modifies index mappings, changes settings, or touches your domain configuration in any way. The credentials you provide need only search read permissions.

Q Can I trigger tests from our CI/CD pipeline after deployments?

A

Yes. ReleGuard provides a REST API for triggering test runs programmatically. You can call it from your deployment pipeline to run your OpenSearch relevance test suite immediately after each deploy — and fail the pipeline if regressions are detected.

Q How quickly will I know about a regression?

A

On the Professional plan and above, tests run hourly — so you'll know within the hour. On the Starter plan, tests run daily. You can also trigger tests manually or via API at any time on any plan.

Stop Relying on Manual Checks
to Catch OpenSearch Regressions

The Search Relevance Workbench is useful for exploration. But it won't catch the 3am catalog pipeline that silently broke your top queries — and costs you a week of conversions before anyone notices. ReleGuard will.

No credit card required
Read-only access only
Cancel anytime

Also Available For