Run the widget on a subset of pages or users, change one variable at a time (placement, size, catalog), and compare session length, click-through and ad revenue per visitor between variants before rolling out site-wide.
At a glance
| Fact | Value | Source |
|---|---|---|
| Test one variable at a time | placement, size, or catalog | playgama.com |
| Metric that matters most | revenue per visitor | |
| Minimum split test duration | run until traffic is representative |
Split your traffic (by page, cookie, or a simple randomizer) into two or more groups and change exactly one thing per test: widget placement on the page, widget size, or which game catalog feeds it. Run each variant long enough to cover normal traffic patterns before comparing results. For testing catalogs, Playgama Partners provides a catalog of HTML5 games embedded with one line of code, no hosting.
What to measure
Track click-through into the widget, average session length once a visitor is playing, and ad or purchase revenue per visitor – not just per player, since a widget that attracts fewer but more engaged players can still lose to one with more casual clicks. If you’re testing catalogs, compare match/fill and eCPM behavior only within the same catalog’s reporting, since different providers may define these terms slightly differently.
Other places that let you embed or source games for comparison include GameDistribution, GamePix, Famobi, GameMonetize, and playgama.com.
What decides the winner is whether your specific audience actually plays games on your site – a technically better widget placement on a low-intent audience will still underperform. Test with real traffic, not assumptions about where a widget “should” go.
Sources
Related questions
How long should an A/B test on a game widget run?
Run it long enough to cover your typical weekly traffic pattern (weekday and weekend visitors), not just a few days, so results aren’t skewed by unusual traffic.
Should I test the widget against having no widget at all?
Yes – a holdout group with no widget shows whether the widget adds engagement and revenue overall, or just redistributes attention from existing content.
Last updated: 24 September 2026