Examples & Best Practices

Example use cases for brand operators and agencies.

The Eraya MCP server gives AI assistants access to your A/B test results, Bayesian statistics, store orders, product analytics, and business KPIs — all queryable in plain English.

The examples below are organized by role: managing your own store as a brand operator, or managing multiple stores as an agency.

For Brand Operators

1. Weekly Test Pulse Check

Use for a quick status instead of reviewing dashboards

Give me a quick status on all my running tests:
- Which have reached statistical significance?
- Which are trending positive vs negative?
- Any that should be ended early (clear winner or clear loser)?
- Any that need more time to reach significance?

What you'll get: Status of all active experiments, Bayesian win probabilities, significance flags, and recommendations for early stopping or extension.

2. Revenue Impact of Completed Tests

Use to quantify the value of your testing program

Look at all my completed tests from the past 6 months.
1. Which winning variations had the highest revenue lift vs. control?
2. Which had the highest conversion rate improvement?
3. Are there any statistically significant winners I haven't implemented yet?
4. Rank my top 3 opportunities by estimated monthly revenue impact.

What you'll get: Ranked list of winners by revenue lift, credible intervals for each estimate, and a prioritized implementation list.

3. Price Sensitivity Analysis

Use to understand what your customers will and won't pay

Based on my price tests:
1. What have I learned about my customers' price sensitivity?
2. Are there products or categories that are more price-elastic?
3. Do new vs. returning customers respond differently to price changes?
4. What's the optimal price point based on test data — maximizing revenue per visitor or conversion rate?

What you'll get: Price sensitivity insights from test data, segment-specific price response differences, and data-backed pricing recommendations.

4. Shipping Strategy Analysis

Use to optimize your shipping offer

Analyse all my shipping tests:
1. Which shipping offer (free shipping threshold, flat rate, etc.) drove the highest conversion rate?
2. Was there a difference in AOV between variations?
3. Did the winning variation hold up for both new and returning visitors?

What you'll get: Conversion rate and AOV comparison across shipping variations, segment breakdowns, and a data-backed recommendation for your shipping strategy.

5. Profit Trap Detection

Use to catch tests where more conversions = less profit

Review my completed tests and identify any "profit traps" — tests where conversion rate improved but revenue per visitor or AOV decreased.

Flag any tests where we might be winning the battle but losing the war.

What you'll get: Tests where conversion went up but revenue went down, warnings about implementing harmful "winners", and profit-focused recommendations.

6. Customer Segment Intelligence

Use to understand how different customer types respond

Across my completed tests, compare how these segments respond:
- Mobile vs. Desktop visitors
- New vs. Returning customers

Are there segments consistently outperforming? Should I run segment-specific tests? Where's the biggest opportunity gap?

What you'll get: Conversion rates and revenue lift by device type and visitor type, and strategic direction for segment-specific testing.

7. Store Health Check

Use for a quick pulse on overall performance

How did we perform last month vs. the month before?
Summarise the key changes — sessions, conversion rate, revenue, and AOV.
Flag anything that moved meaningfully in either direction.

What you'll get: Period-over-period KPI comparison with automatic flagging of notable changes.

8. Product Performance Deep Dive

Use to find underperforming products

Using product analytics for the past 30 days:
1. Which products have the highest add-to-cart rate? Which have the lowest?
2. Which products are driving the most revenue?
3. Are there products with high page views but low conversion — candidates for testing?
4. Suggest 3 products that would benefit most from a price or content test.

What you'll get: Product-level funnel data, add-to-cart rates, revenue rankings, and a prioritised list of test candidates.

9. Testing Strategy Health Check

Use to evaluate the effectiveness of your testing program

Review my testing history and tell me:
1. How many tests have I completed in the past 3 months?
2. What's my win rate (% of tests with a statistically significant winner)?
3. What test types am I running most? What am I not testing?
4. Am I building on previous learnings or testing in isolation?
5. Suggest 3 follow-up tests based on what I've already learned.

What you'll get: Test count, win rate, test type distribution, and a strategically grounded test roadmap.

10. AOV and Discount Trend

Use to catch margin compression early

Is our average order value trending up or down over the past 8 weeks?
How has units per order changed in the same period?
Are we discounting more than we used to, and is it affecting revenue per order?

What you'll get: AOV and units-per-order trend data, discount usage changes, and revenue-per-order movement that can signal margin pressure before it shows up in reports.

For Agencies

1. Monthly Client Performance Review

Use when preparing for monthly client calls

Give me a comprehensive analysis of all tests completed this month for [STORE NAME]. Include:
1. Which tests reached statistical significance and their winners
2. The estimated revenue impact of implementing winners
3. Key insights to share with the client
4. Any tests that should be ended early or extended

What you'll get: List of completed tests with win/loss status, revenue lift percentages, statistical confidence levels, and actionable recommendations for client discussion.

2. Cross-Test Pattern Analysis

Use to find winning strategies that work consistently

Analyse all completed tests for [STORE NAME] over the past 3 months.
What patterns appear across multiple winning tests?
What concepts — urgency, trust, social proof, simplification, pricing anchoring — consistently win or lose for this customer base?

What you'll get: Patterns across winning variations, concepts that resonate with the store's audience, and strategic direction for future testing.

3. Quarterly Test Roadmap

Use for strategic planning with clients

Based on [STORE NAME]'s test history:
1. What test categories have delivered the best ROI?
2. What areas haven't been tested yet (blind spots)?
3. Based on winners, what specific follow-up tests should we run?
4. Prioritize 5 test ideas for next quarter by potential impact.

What you'll get: Analysis of highest-performing test categories, untested areas, and a prioritised test roadmap for the next quarter.

4. Cross-Store Performance Scan

Use to quickly identify which clients need attention

Switch through each of my stores and check KPIs for the past 30 days.
Which stores have seen the biggest drop in conversion rate or revenue?
Rank them by urgency so I know where to focus this week.

What you'll get: A prioritised list of stores showing performance deterioration, so you can triage proactively rather than waiting for clients to flag issues.

Tip: Use list_stores to see all your stores, then switch_store to move between them during the session.

Best Practices

Start Every Session with the Schema

Call get_eraya_schema at the beginning of a session. It gives the AI the full picture of Eraya's data model — test types, order property conventions, and which tool to use for each task — resulting in much more accurate responses.

First, call get_eraya_schema to understand the data, then show me all active tests.

Use `get_experiment_statistics` for Winner Determination

get_experiment_statistics contains pre-computed Bayesian win probabilities and credible intervals — the most reliable signal for deciding a winner. get_experiment_results shows live session funnel data but doesn't include statistical significance.

Use get_experiment_statistics for: "Is variation B the winner?"
Use get_experiment_results for: "What's the add-to-cart rate by device type?"

Select Your Store First

If your account has access to multiple Shopify stores, always select the correct store at the start of a session.

Call list_stores, then switch to [STORE NAME].

Add a System Prompt or Custom Instruction to your AI client with your preferred store name — the AI will switch to it automatically at the start of every session.

Narrow Date Ranges for Faster Results

When analysing orders or product analytics, specify a date range to keep response sizes manageable and results relevant.

Show me orders from April 1 to April 30, 2026.

Combine Tools for Deeper Analysis

The most powerful queries combine multiple tools. For example:

Get the statistics for my top price test, then pull the raw orders for the winning variation and tell me the average order value and most purchased products.

This chains get_experiment_statistics → get_experiment_orders → product-level analysis in a single conversation.

Security

Never share your MCP access token — it grants read access to your store data.
Tokens are valid for 365 days. If you suspect a token is compromised, re-authorise via https://api.eraya.ai/mcp-oauth/authorize to issue a new one.
MCP access is verified on every request. Downgrading from the Pro plan immediately revokes access.

PreviousAvailable Tools NextClaude

Last updated 13 days ago