🚰 Feedback

You can now run real price A/B tests on Shopify—no coding, no duplicate products. Set up pricing experiments in minutes using the Convert app UI and test actual checkout prices across: Product pages (all variants) Collection pages Search results Cart & checkout (Shopify Plus required) Create flexible pricing rules by product, collection, tags, or patterns, stack multiple discounts, and let Convert handle the rest automatically. Perfect for testing discounts, price increases, bundles, or promotional strategies—with real revenue impact, not just display changes. Learn more: https://support.convert.com/hc/en-us/articles/shopify-pricing-a/b-testing-quick-start-guide

You can now export experiment results as Tabular CSVs—built specifically for clean imports into tools like Airtable, Notion, Coda, and Monday. CSV Aggregated Tabular: One row per variation/goal with totals for the selected date range CSV Daily Tabular: Daily time-series data with a flat, predictable schema Both formats use header-first tables, consistent column names, and machine-friendly data types, so you can import them directly—no cleanup required. Perfect for recurring reports, dashboards, and automated workflows. Learn more: https://support.convert.com/hc/en-us/articles/export-download-in-convert-understanding-the-different-export-types-

You can now work together in the Visual Editor without accidental overwrites. When teammates edit the same experience, the Visual Editor detects changes and prompts you to Reload or Overwrite & Save. A short safety window helps prevent conflicts, and Change History lets you see who changed what. Overwrite carefully—this replaces the latest saved version. No impact on live traffic. Learn more: https://support.convert.com/hc/en-us/articles/edit-locking-version-conflict-prevention-in-the-visual-editor#overview

You can now connect your Convert optimization data directly to Claude Desktop, Cursor, or any MCP-compatible AI assistant—no middleware, no complicated setup. With the Convert MCP Server, you get: > Zero-config setup (run it via npx) > Full REST API v2 coverage (organized into clean namespaces) > Built-in Convert knowledge (5,279+ indexed Help Center chunks) > Configurable access levels: reporting (default), readOnly, or all (read/write) Once connected, you can ask in plain English: “Show all active experiments and performance” “Summarize top winners from the last 30 days” “Which audiences are performing best?” Start with reporting access and upgrade permissions only if you need deeper configuration or write actions. Learn more here: https://support.convert.com/hc/en-us/articles/convert-mcp-server

Auto-Allocation powered by Multi-Armed Bandits (MAB) lets Convert automatically send more traffic to better-performing variations while your test is running. Instead of fixed splits, MAB adapts in real time—helping you learn faster, reduce wasted traffic, and drive more conversions during the experiment. Works with both Frequentist and Bayesian stats and supports strategies like Thompson Sampling, Epsilon-Greedy, and UCB. Once enabled, traffic allocation is fully automatic, with clear reporting to show how it evolves over time. Learn more: https://support.convert.com/hc/en-us/articles/multi-armed-bandit-mab-auto-allocation-in-convert

You can now combine audience conditions using AND logic, not just OR. This means you can target users who meet multiple conditions at once (e.g. from Google and Yahoo), giving you tighter, more accurate audience definitions. To use this, make sure your project is running on the new tracking script. Learn more: https://support.convert.com/hc/en-us/articles/115000009891-define-audiences-for-your-experience

You can now remove specific data directly from experiment reports, no support ticket needed. Owners and Admins on Pro and Enterprise plans can delete selected event types and date ranges (including specific conversion goals) straight from the report view. This gives you more control when cleaning up test data caused by setup issues, QA traffic, or unexpected events. The flow is simple: choose the date range, select the event types, review the impact, and confirm. Changes may take a few seconds to reflect, and once data is removed, it can’t be restored—so you’re always in control. Learn more in our Help Center: https://support.convert.com/hc/en-us/articles/remove-report-data-feature

We’ve added Average Order Value (AOV) and Average Products Per Order (APPO) to experiment reports to give you deeper insight into order-level performance. AOV (Average Order Value) shows the average revenue per order for the selected date range or segment. It appears as a currency column alongside RPV in the Revenue table, while all RPV statistics (lift, confidence, etc.) continue to work exactly as before. APPO (Average Products Per Order) shows the average number of items per order. It uses the same product count data as APPV, but is calculated per order instead of per visitor. These metrics make it easier to understand whether changes are driving bigger orders, more items per order, or both—not just more visitors converting. Learn more about experiment reports and these metrics: https://support.convert.com/hc/en-us/articles/understanding-report-metrics-in-convert

All new experiments now use t-tests for Frequentist statistics (experiments created before Oct 13, 2023 continue using z-tests). You can configure key settings based on how critical your test is, including confidence level (95% or 99%), test type (two-tailed by default, one-tailed optional), and multiple comparison correction (Sidak, Bonferroni, or none). To make setup easier, Sensible Defaults let you quickly apply recommended settings for standard or mission-critical experiments. Learn more: https://support.convert.com/hc/en-us/articles/10157554451085-statistical-methods-used

We’re adding Risk and Expected Uplift to Bayesian reports to help you make faster, more confident decisions. See the realistic impact a variant could deliver and the downside to watch out for; alongside Chance to Win. Great for high-traffic tests, revenue-critical flows, early stopping, and comparing close winners. Metrics are optional and can be toggled on or off. Learn more: https://support.convert.com/hc/en-us/articles/understanding-risk-and-expected-uplift-in-bayesian-reports