A/B Testing · Front Studio

Test everything. Win faster.

Every headline, image, and label on your front page can be A/B tested against live traffic. Kilkaya calls a winner at 95% confidence — and does it faster than anything else on the market.

No black boxes. No tricky probabilities. Just a clear winner, backed by real clicks.

A/B Test: concluded
Variant C is the winner with 95% certainty.
CTR increase: +74%

PM: "We must act now" — economic crisis looms

CTR: 5.4%Clicks: 216

Any element. Any slot.

Headlines and images are only the start. Anything inside an article slot can become a variable in your test.

Headlines

The single most powerful element. Shorter, more emotional, more specific titles win again and again.

Images

Test the crop, the focal point, the image itself. Face-detected cropping keeps the subject in frame.

Lead text

The teaser copy under the headline. Different angles on the same story can produce very different CTRs.

Section markers

Category labels shape what readers expect. Try "Breaking" vs "Analysis" and see whether engagement shifts.

Background & colours

Small visual changes — a highlight colour, a card background — change which articles catch the eye.

AI-generated variants

Spin up several headline alternatives with AI and let real traffic pick the winner.

Why the Frequentist method

We tried Bayesian methods and multi-armed bandits. The classical approach wins on speed, simplicity, and clarity — which is what you need when content has a short shelf life.

Fastest to a winner

Classical statistics is built to reach a reliable answer quickly — crucial when a story's traffic window is hours, not days.

95% confidence threshold

We only call a winner when the data is statistically solid. No early conclusions, no noise mistaken for signal.

Transparent and auditable

No black-box probabilities. Every result is something you can explain to any editor — and trust on the spot.

We only count visible impressions — and clicks

An impression only counts when the slot is actually visible on screen. But visible impressions still vary wildly by position — a slot at the top of the page can hit 50% CTR, while the same slot further down might pull just 2%. That's a 25x gap. So we use clicks, not views, as the primary measure — the only metric that holds up regardless of where the slot sits.

Statistical precision

Clicks = accuracy

After running thousands of A/A tests — pitting identical variants against each other to measure natural variation — we found that with just 100 clicks per variant, we can detect a 23% CTR increase at 95% confidence.

The inverse-square law sets the rules here: catching smaller wins takes exponentially more clicks. For big gains, results come quickly. For subtle ones, patience pays.

Clicks needed per variant
+50% CTR
Large gain — detectable quickly
21
clicks
+23% CTR
Our default threshold
100
clicks
+10% CTR
Subtle gain — needs more data
529
clicks

Built to reward real engagement

A headline that gets clicks but drives readers straight back out the door isn't a winner. Kilkaya filters those out.

Quick-exit filtering

By default, Kilkaya throws out every click where the reader leaves within 5 seconds. Clickbait titles can inflate CTR while delivering nothing real — the filter makes sure the winner actually earned its traffic.

What makes a winning headline?

Across thousands of tests, the strongest titles tend to share a few traits:

  • Concise — but not cryptic
  • Transparent about the story
  • Emotionally resonant — curiosity, urgency, empathy
  • Specific enough to build trust

Ready to let your audience decide?

Every slot on your homepage is an experiment waiting to happen. Start testing — and start finding winners faster.