A/B Testing overview

Use A/B Testing to compare destination URLs on an existing Pointerly link. Traffic is split by variant weight; results accumulate until you pause, complete, or declare a statistical winner—and when a winner is found, its URL can be applied back to the link automatically.

What you'll do

• Open Dashboard → A/B Testing (requires the ab_testing feature)
• Create a test bound to a link that does not already have an A/B test
• Configure variants (2–5), traffic weights (must sum to 100%), and a goal metric
• Start, pause, resume, or complete the test from the list or detail dialog
• Check for a winner when enough sample and confidence are met

Before you start

• Plan includes A/B Testing (otherwise you see an upgrade block)
• You are in the correct team workspace
• At least one tracked link exists and is not already attached to another A/B test
• You know which URLs you want to compare (control keeps the link’s original URL)

Steps

1. Open A/B Testing

1. Go to Dashboard → A/B Testing.
2. Confirm the page header shows A/B Testing with description Test different link variations to optimize conversions.
3. Use search and the status filter (All / Draft / Running / Paused / Completed) to find existing tests.

2. Create a test

1. Click Create Test (keyboard create shortcut is shown on the button when available).
2. In Create A/B Test, select the Link to test. Only links without an existing abTestId appear.
3. Review Current URL for the selected link (this is the control destination).
4. Set a Test name (auto-suggested from the product title or slug when you pick a link).
5. Choose a Goal metric:
   • Clicks (CTR) — optimize click-through
   • Conversions — optimize conversion rate
   • Revenue — optimize revenue per visitor
6. Configure Variants (minimum 2, maximum 5):
   • Control always uses the selected link’s original URL.
   • Each additional variant needs a name, weight (1–99%), and alternative URL.
   • Use Balance to redistribute weights evenly so they sum to 100%.
7. Optional — open Advanced Settings:
   • Confidence threshold (80–99%, default 95%)
   • Minimum sample size (impressions per variant before analysis; default 100)
8. Click Create Test. Pointerly creates the draft test and attaches it to the selected link.

3. Start, pause, and monitor

1. From the test card menu (⋯) or the detail dialog:
   • Draft → Start Test (status becomes Running)
   • Running → Pause Test or Complete Test
   • Paused → Resume Test or Complete Test
2. Click a card to open the detail dialog for impressions, clicks, and per-variant results.
3. While a test is Running, variants cannot be edited (weights and URLs are locked).

4. Declare a winner and apply the URL

1. When enough traffic has accrued, use Check winner / complete with statistical evaluation from the detail flow.
2. If confidence and sample size are met, Pointerly may complete the test and apply the winner URL to the linked short URL automatically.
3. If there is not enough data yet, you will see a message to keep collecting traffic.
4. You can also Complete Test manually without waiting for auto-winner selection.

Tips

• Test one meaningful change per experiment (landing page, marketplace, or offer URL)—not many unrelated changes at once.
• Keep early traffic balanced (50/50 or even splits) until you have a clear sample; skew weights later only if you must protect revenue.
• Prefer conversions or revenue as the goal when clicks alone do not reflect monetization.
• Raise minimum sample size for high-traffic links so a few bursts do not crown a false winner.
• A link can only participate in one A/B test at a time—finish or detach before starting another.

Troubleshooting

Related

• Links overview
• Create a link
• Analytics overview