Data Lab

Why Data Lab exists

The standard results page answers the most common question: “Did the variation beat the control?” But experimentation programs need deeper analysis. Data Lab lets you explore the raw data behind your experiments — segment by any dimension, compare time periods, analyze visitor journeys, and build reports for stakeholders who need more than a summary card.

What Data Lab offers

Capability	Standard results page	Data Lab
Primary and secondary metric results	Yes	Yes
Custom time range analysis	No	Yes
Custom segmentation dimensions	Limited	Unlimited
Visitor-level data exploration	No	Yes
Cross-experiment comparisons	No	Yes
Custom calculated metrics	No	Yes
Exportable reports and dashboards	CSV only	Interactive dashboards and CSV

Access Data Lab

1. In the Optimizely application, navigate to Data Lab in the left sidebar
2. Select an experiment from the experiment picker
3. The Data Lab workspace opens with your experiment data loaded

Build a custom analysis

Step 1: Select the experiment and time range

1. Choose the experiment from the dropdown at the top of the workspace
2. Set the date range — you can analyze the full experiment duration or zoom into a specific window
3. Data Lab loads all visitor events within the selected range

Step 2: Choose your metric

Data Lab provides access to all metrics configured on the experiment, plus the ability to create ad-hoc metrics.

Use an existing metric:

1. Click Add Metric in the analysis panel
2. Select from the experiment’s configured metrics
3. The metric appears as a column in the results table

Create a custom metric:

1. Click Add Metric > Custom
2. Define the metric using event data:
   • Event key — The event to measure (e.g., purchase, add_to_cart)
   • Aggregation — Count, unique count, sum, average, or median
   • Filter — Optional conditions on event tags (e.g., revenue > 5000)
3. Name the metric and add it to the analysis

Step 3: Segment your data

Segmentation reveals how different visitor groups respond to the variation.

1. Click Add Segment in the toolbar
2. Choose a dimension to segment by:
   • Device type — Desktop, mobile, tablet
   • Browser — Chrome, Safari, Firefox, Edge
   • Country or region — Geographic breakdown
   • Traffic source — Direct, organic, paid, referral
   • Custom attributes — Any attribute passed to Optimizely
3. The results table breaks down each metric by the selected segment

You can apply multiple segments simultaneously to create cross-tabulations (e.g., device type by country).

Step 4: Analyze the output

Data Lab displays results in an interactive table with the following columns for each variation:

• Visitors — Count of unique visitors in the segment
• Conversions — Count or sum depending on metric type
• Conversion rate — Conversions divided by visitors
• Improvement — Percentage change vs. control
• Significance — Statistical confidence for the segment

Tip: Segment-level significance requires higher visitor counts than the overall result. A segment showing “not significant” often means insufficient data, not that the effect is absent.

Common analysis patterns

Analyze time-windowed performance

Check whether the variation effect changed over time — useful for detecting novelty effects.

1. Set the time range to the first week of the experiment
2. Record the improvement percentage
3. Change the time range to the second week
4. Compare — if the improvement dropped significantly, novelty bias may be inflating the overall result

Compare revenue per visitor across segments

For e-commerce experiments, revenue per visitor matters more than conversion rate.

1. Add the Revenue metric
2. Segment by Traffic source
3. Check whether the variation lifts revenue consistently or only for certain channels

Identify underperforming segments

Find segments where the variation hurts performance.

1. Segment by Device type
2. Look for negative improvement on any device
3. If the variation wins on desktop but loses on mobile, consider a device-specific implementation

Build a stakeholder report

1. Configure the analysis with the metrics and segments your stakeholders care about
2. Click Save Report to preserve the configuration
3. Click Share to generate a link that stakeholders can access without navigating through the experiment
4. Export to CSV for inclusion in presentations or data warehouse pipelines

Export data

Data Lab supports two export formats:

Format	Best for
CSV	Importing into spreadsheets, BI tools, or data warehouses
Dashboard link	Sharing interactive reports with stakeholders

To export:

1. Configure your analysis
2. Click Export in the toolbar
3. Select CSV or copy the dashboard link

Data freshness

Data Lab uses the same event pipeline as the standard results page. Event data typically appears within 15 to 30 minutes of collection. If you need real-time data, use the standard results page, which refreshes more frequently for the primary metric summary.

Limitations

• Data Lab queries run against the experiment’s event data. It does not access external analytics data (use Integrate with Analytics for cross-platform reporting).
• Custom segments require the attribute to have been sent to Optimizely during the experiment. You cannot retroactively add segmentation dimensions.
• Very large experiments (millions of visitors) may experience slower query times. Apply date range and segment filters to reduce the dataset size.

Troubleshooting

Issue	Cause	Fix
No data appearing	Experiment has not collected enough events yet	Wait for at least 24 hours of data collection
Custom metric returns zero	Event key mismatch	Verify the event key matches what your code or snippet sends
Segment shows “insufficient data”	Too few visitors in that segment	Broaden the segment or extend the date range
Export file is empty	Filters too restrictive	Remove filters and try again

Related content

Continue your journey with these related pages.

📖Understand the concepts

• How Experimentation Works

🔧Follow a guide

• Interpret Results
• Set Up Metrics
• Custom Metrics and Events