Learn — Smart Experimentation
Smart Experimentation
The smarter and easier way to run A/B tests.
A/B test everything — recommendation algorithms, layouts, push audiences, etc.
Collecting data
CTR · recommended_for_youProbability to be Best
Variant B
32.8%
Variant C
26.2%
Variant A
39.6%
Conversion Rate (CTR %)
0.760%
B
0.755%
C
0.755%
A
Tabular Results
| Variant | P(Best) | CTR | Clicks | Δ Mean |
|---|---|---|---|---|
Variant B★ winner | 84.2% | 0.833% | 181 | +18.6% |
Variant C | 10.7% | 0.729% | 154 | +3.9% |
Variant A(baseline) | 5.1% | 0.706% | 150 | — |
Auto-promotes winners
Traffic automatically shifts toward the best-performing variant while testing continues. Losers are phased out without any manual intervention.
Full-stack testing
Test recommendation models, UI layouts, headline variations, and paywall logic. If it affects reader behaviour, you can optimise it.
What you can test
If it touches your readers, you can optimise it.
Recommendation algorithms
Which model drives more recirculation?
Headline variations
Auto-promote the winning title in real-time
Paywall triggers
Which propensity threshold converts best?
UI layouts
Article page, homepage, widget placement
Push audiences
Broad vs. targeted — measure churn impact
Newsletter order
Does article position 1 drive opens?