Skip to main content

Pricing Experiments

Test pricing strategies with A/B experiments before rolling them out to all customers.

What Are Pricing Experiments?​

Experiments let you test different prices with a portion of your traffic to find what works best:

flowchart LR
    subgraph Traffic[Incoming Bookings]
        A[All Visitors]
    end

    subgraph Split[Random Assignment]
        A --> B[50% Control]
        A --> C[50% Variant]
    end

    subgraph Pricing[Different Prices]
        B --> D[Current Price: R400]
        C --> E[Test Price: R450]
    end

    subgraph Results[Measure]
        D --> F[Compare Conversions & Revenue]
        E --> F
    end

Creating an Experiment​

Step 1: Define Your Hypothesis​

Before starting, be clear about what you're testing:

  • "A 10% price increase will not significantly reduce bookings"
  • "A twilight discount will increase afternoon utilization"
  • "Weather-based pricing will improve wet-day revenue"

Step 2: Create the Experiment​

  1. Go to Pricing > Experiments
  2. Click New Experiment
  3. Configure:
FieldDescription
NameDescriptive name (e.g., "Weekend Premium Test")
Control RuleYour current pricing rule
Variant RuleThe new pricing to test
Traffic SplitPercentage for each (default 50/50)
DurationHow long to run (recommended: 2-4 weeks)

Step 2A: Example Experiment (plain-language)​

Name: Weekend Premium Increase
Status: Running
Scope: Course
Traffic %: 50
Control rule: Weekend Premium +15%
Variant rule: Weekend Premium +25%
Start date: This Friday
End date: 4 weeks from start

Step 3: Set Success Metrics​

Choose what to measure:

MetricWhat It Tells You
Conversion Rate% of viewers who book
Revenue per SlotTotal revenue / available slots
Booking VolumeNumber of bookings
Average Order ValueRevenue per booking

Experiment Lifecycle​

stateDiagram-v2
    [*] --> Draft: Create experiment
    Draft --> Running: Start experiment
    Running --> Paused: Pause if needed
    Paused --> Running: Resume
    Running --> Completed: End date reached
    Running --> Cancelled: Stop early
    Completed --> [*]: Apply winner or discard

Status Descriptions​

StatusMeaning
DraftConfigured but not active
RunningLive and collecting data
PausedTemporarily stopped
CompletedFinished, review results
CancelledStopped without conclusion

Reading Results​

Key Metrics​

After your experiment runs, review:

Experiment: Weekend Premium Test
Duration: 14 days
---------------------------------
              Control    Variant
Visitors:     1,240      1,235
Bookings:       186        172
Conv. Rate:   15.0%      13.9%
Revenue:     R74,400    R77,400
Rev/Visitor:  R60.00     R62.67

Statistical Significance​

Results include a confidence indicator:

ConfidenceMeaning
< 80%Not enough data, keep running
80-95%Trending, but not conclusive
> 95%Statistically significant
Wait for Significance

Don't end experiments early. Let them reach 95% confidence for reliable results.

Quick Start Checklist​

  1. Pick one change to test (price only).
  2. Set Traffic % to 10-20% for the first run.
  3. Run for at least 2 weeks to avoid short-term noise.
  4. Review results and only apply when confidence is high.

Best Practices​

Experiment Design​

  1. Test one thing at a time - Don't change price AND conditions
  2. Use meaningful differences - Test 10-20% changes, not 1-2%
  3. Run long enough - At least 2 weeks, ideally 4
  4. Pick representative periods - Avoid holidays or unusual weeks

What to Test​

Good experiment ideas:

TestControlVariant
Price sensitivityR400R450 (+12.5%)
Twilight timing-20% after 4pm-20% after 3pm
Weekend premium+15%+25%
Lead time discount-5% at 14 days-10% at 14 days

What Not to Test​

Avoid experiments that:

  • Could upset regular members (test with visitors first)
  • Run during peak tournament season
  • Have too small a price difference to measure

Managing Experiments​

Pausing an Experiment​

If something goes wrong:

  1. Go to Pricing > Experiments
  2. Find the active experiment
  3. Click Pause
  4. Investigate the issue
  5. Resume or Cancel as appropriate

Ending Early​

Sometimes you need to stop early:

  • Clear winner: One variant significantly outperforms (>95% confidence)
  • Clear loser: Variant is hurting revenue badly
  • External factors: Event or situation makes data invalid

Applying Results​

After a successful experiment:

  1. Review the Results tab
  2. If variant won, click Apply Winner
  3. This promotes the variant rule to your standard pricing
  4. The experiment is archived

Conflict Prevention​

The system prevents conflicting experiments:

  • Only one experiment can run per price rule
  • Overlapping date ranges are blocked
  • You'll see warnings if experiments might interfere

Example: Testing a Weekend Premium​

Setup​

Name: Weekend Premium Increase
Control: Current weekend rate (+15%)
Variant: Higher weekend rate (+25%)
Split: 50/50
Duration: 4 weeks
Success: Revenue per available slot

Results After 4 Weeks​

              Control (+15%)  Variant (+25%)
Bookings:     324             298
Revenue:      R149,040        R156,450
Rev/Slot:     R298            R313
Confidence:   97.2%

Winner: Variant (+25%)
Insight: 8% fewer bookings but 5% more revenue

Decision​

With 97% confidence that the variant generates more revenue, apply the +25% weekend premium as the new standard.

Troubleshooting​

Results are not changing​

  1. Confirm the experiment status is Running.
  2. Make sure the Traffic % is greater than 0.
  3. Ensure the selected Control and Variant rules are active.

Results show 0 exposure or 0 bookings​

  1. Verify the experiment date range includes current dates.
  2. Confirm the experiment scope matches the course you are testing.
  3. Increase Traffic % temporarily to confirm data is flowing.

A variant looks worse but confidence is low​

  1. Let the test run longer.
  2. Avoid changing rules mid‑experiment.
  3. If the variant is clearly harming revenue, pause and cancel.