Analytics Deep Dive
The problem this solves
Section titled “The problem this solves”Out-of-the-box dashboards answer common questions — how much traffic did the site get, which experiment won, what is the conversion rate. But every organization has questions that standard dashboards cannot answer. You might need to compare content performance across regions, track a custom engagement score, correlate experiment results with downstream revenue over time, or feed analytics data into a BI tool your leadership team already uses.
This deep dive covers the advanced analytics capabilities in Optimizely: building custom reports, understanding the relationship between experiment analytics and web analytics, and getting data out of Optimizely into external systems.
Dashboards: the starting point
Section titled “Dashboards: the starting point”Optimizely Analytics ships with pre-built dashboards that cover the most common reporting needs. Before building custom reports, understand what is already available.
Overview dashboard
Section titled “Overview dashboard”The overview dashboard provides a high-level snapshot of site performance. It answers the question: “How is the site doing right now?”
Key widgets:
- Active visitors — Real-time count of visitors currently on the site
- Traffic trend — Sessions over time with day-over-day and week-over-week comparisons
- Top pages — Highest-traffic pages ranked by views
- Conversion summary — Overall conversion rate and total conversions for your primary goal
- Traffic sources — Breakdown by channel (organic search, paid, social, direct, referral)
Experiment dashboard
Section titled “Experiment dashboard”The experiment dashboard is tightly integrated with both Web Experimentation and Feature Experimentation. It answers the question: “Which experiments are running, and what are the results?”
Key widgets:
- Active experiments — List of currently running experiments with status indicators
- Results summary — For each experiment: conversion rate by variation, statistical significance, and confidence interval
- Stats Engine indicators — Whether the experiment has reached statistical significance, the false discovery rate, and the recommended action (keep running, declare winner, stop)
- Metric trends — How experiment metrics change over time, useful for detecting novelty effects or seasonal patterns
Pages dashboard
Section titled “Pages dashboard”The pages dashboard helps content teams understand how individual pages perform. It answers: “Which content is engaging visitors, and which is not?”
Key widgets:
- Page views and unique visitors — Traffic volume for each page
- Engagement metrics — Average time on page, scroll depth, bounce rate
- Entry and exit rates — Which pages visitors land on first and where they leave
- Content groups — Performance aggregated by content category (blog, product pages, support articles)
Commerce dashboard
Section titled “Commerce dashboard”For sites with e-commerce, the commerce dashboard tracks revenue and purchase behavior. It answers: “What is driving revenue?”
Key widgets:
- Revenue and orders — Total revenue, order count, and average order value over time
- Product performance — Views, add-to-cart rate, and purchase rate by product
- Revenue attribution — Which experiments, campaigns, and content pieces contributed to revenue
- Cart analysis — Abandonment rate, items per cart, and checkout completion rate
Experiment analytics vs web analytics
Section titled “Experiment analytics vs web analytics”Optimizely Analytics contains two measurement systems that serve different purposes. Understanding the distinction prevents misinterpretation of data.
| Dimension | Experiment analytics | Web analytics |
|---|---|---|
| Purpose | Determine whether a change caused an improvement | Describe what visitors are doing on the site |
| Statistical model | Sequential testing with Stats Engine (always-valid p-values) | Descriptive statistics (counts, rates, averages) |
| Attribution | Attributes conversions to experiment variations | Attributes traffic to sources and campaigns |
| Time scope | Duration of the experiment | Any date range you select |
| Identity | Visitor bucketed into a variation | Anonymous session (cookie-based) |
| Output | Winner declaration, confidence interval, lift estimate | Reports, trends, comparisons |
How they connect
Section titled “How they connect”Experiment analytics and web analytics share the same underlying event data. When a visitor views a page, clicks a button, or makes a purchase, that event feeds both systems. The difference is in how each system interprets the event:
- Web analytics counts the event and attributes it to a traffic source
- Experiment analytics checks whether the visitor is in an active experiment and attributes the event to their variation
This shared data model means you can:
- Drill from experiment results into web analytics — See how a winning variation affected downstream page views, bounce rates, or time on site
- Segment web analytics by experiment — Filter web analytics reports to show only visitors who were in a specific experiment
- Correlate content performance with experiment outcomes — Understand whether a content change that won an experiment also improved overall engagement metrics
Custom reports
Section titled “Custom reports”When standard dashboards do not answer your question, custom reports let you build your own analysis.
Custom metrics
Section titled “Custom metrics”Define calculated metrics that combine standard metrics into measurements specific to your business:
- Engagement score — A weighted combination of page views, scroll depth, and time on page
- Revenue per session — Total revenue divided by sessions, useful for comparing experiment variations
- Content efficiency — Conversions divided by content production cost (requires custom event data)
Custom dimensions
Section titled “Custom dimensions”Segment data by properties that are meaningful to your organization:
- Customer tier — Segment by loyalty program level (Gold, Silver, Bronze)
- Content category — Group pages by editorial category beyond the URL structure
- User role — Distinguish between logged-in customers, prospects, and internal users
Custom dimensions require developers to pass the dimension values with events. Once configured, they appear as filter and breakdown options in all reports.
Report builder
Section titled “Report builder”The report builder provides a drag-and-drop interface for constructing custom views:
- Select metrics — Choose which numbers to display (page views, conversions, revenue, custom metrics)
- Add dimensions — Choose how to break down the data (by page, traffic source, device, custom dimension)
- Apply filters — Narrow the data to specific segments, date ranges, or conditions
- Choose visualization — Table, line chart, bar chart, funnel, or pie chart
- Save and share — Save reports to your dashboard or share them with team members
Data export
Section titled “Data export”Analytics data can be extracted from Optimizely for use in external tools. Three export methods serve different use cases.
CSV export
Section titled “CSV export”The simplest option. Any report or dashboard widget can be exported as a CSV file for use in spreadsheets or ad-hoc analysis.
Best for:
- One-time analysis in Excel or Google Sheets
- Sharing data with stakeholders who do not have Optimizely access
- Quick data pulls that do not need automation
API access
Section titled “API access”The Analytics API provides programmatic access to metrics and dimensions. Use it to build custom dashboards, automate reporting, or integrate analytics data into internal tools.
Best for:
- Custom dashboards built with tools like Grafana, Retool, or internal platforms
- Automated daily or weekly report generation
- Integrating analytics data into CRM or marketing automation systems
Data warehouse sync
Section titled “Data warehouse sync”For enterprise analytics, Optimizely can sync analytics data to your data warehouse (Snowflake, BigQuery, Redshift). This enables cross-system analysis where you join Optimizely data with data from other platforms.
Best for:
- Enterprise BI reporting in Tableau, Looker, or Power BI
- Cross-system attribution models that combine Optimizely data with ad platform data
- Machine learning models that use behavioral data as features
Decision guide: choosing the right analysis approach
Section titled “Decision guide: choosing the right analysis approach”| Question | Approach |
|---|---|
| ”Did this experiment win?” | Experiment dashboard — check Stats Engine results |
| ”How much traffic did we get last month?” | Overview dashboard — traffic trend widget |
| ”Which content type drives the most conversions?” | Custom report — conversions by content category dimension |
| ”How do our experiment results correlate with revenue over quarters?” | Data warehouse sync — join experiment data with revenue data in BI tool |
| ”What is our team’s custom engagement score by audience segment?” | Custom report — engagement score metric by customer tier dimension |
| ”How does our analytics data feed into our broader data stack?” | API export or warehouse sync — depends on destination system |