Tools & Operations 2026

A/B Testing
CTAs & Titles
Lifting conversions with small improvements

*How well it works varies by individual and site. Results and amounts are not guaranteed
One at a time
Don't change several at once
Look at rate
Ratio, not click count
Don't rush
Don't decide on a few
9 slides
2

What is A/B testing? (gently)

2 / 9
🅰️
An experiment comparing two versions: prepare A and B and let reader behavior show which responds better
🧠
Drop the assumptions: a confident title may flop; a casual phrase may win. The reader tells you the answer
🧱
Stack small improvements: no single leap. Small gains add up to a difference over six months
💡 Whether you improve varies by individual and site. Continuing to test does not guarantee results
3

Three elements to test first

3 / 9
📰
Title: the first thing seen in search results and lists. A difference here affects inflow itself
🗣️
CTA wording: "Sign up now" vs "First, see how it works" carry different psychological pressure
🔘
Button color & placement: testable without touching text. Mid-article or end, pop or blend
⚠️ Start where impact is large and change is easy. Change only one thing per test
4

How to test on a small site

4 / 9
📅
Time-period comparison: A this week, B next week. No tool needed; align conditions
📄
Page-duplication comparison: A/B on separate URLs, same period. Mind duplicate-content settings
🐣
Start with time-period: "two versions, two weeks each" is plenty. Making it a habit is what counts
💡 A holiday or sudden market move changes reader behavior. Keep external factors in mind
5

How to set up measurement (GA4)

5 / 9
🎯
Decide one goal: e.g. "sign-up button clicks" — the action you want to increase
📌
Place an event on it: record the button click as a GA4 event and count how often it's pressed
📊
Distinguish A/B by name: compare by "rate," not raw clicks (impressions differ between pages)
💡 Not perfect is fine. Confirming just "sign-up clicks are counted" is enough as a foundation
6

How to read results

6 / 9
🎲
Small counts swing on chance: 2 vs 1 clicks out of 10 isn't "twice as good." Count matters
🤝
A narrow gap is a draw: if the difference is tiny, don't force a winner — "no difference emerged"
🔍
Suspect external factors: market moves, holidays, social shares can move numbers, not the test
⚠️ Don't decide on a few clicks. Adopt only once a clear difference holds over a certain count
7

Common mistakes

7 / 9
①
Changing several at once: you can't tell what worked, so it can't be reproduced. Change one thing
②
Too-short a period: cutting it off in a day or two turns day/time skew into a conclusion
③
Vague goal: not defining success up front means interpreting results to suit yourself
⚠️ ④ Don't bait clicks with hype. "Definitely earn / no risk" breaks the law and drives early churn
8

The one thing to try today

8 / 9
1
Pick one article with the most traffic. Effects of improvement show up more easily there
2
Make two versions of "just one" — the title or CTA (e.g. "Sign up now" ⇄ "First, see how it works")
3
Test two weeks each and compare by rate. Don't decide on a few; judge after it accumulates
💡 What you need: "verify with numbers, not assumptions." One element at a time, no rush, honestly

Lifting conversions with small improvements recap

1
A/B testing compares two versions by reader behavior. Verify with numbers, not assumptions
2
Start with title, CTA wording, and button color/placement. Change only one element at a time
3
Measure "rate" in GA4, don't decide on a few. Exaggeration is a no (law & early churn)
4
Start by making two CTA versions for one article. Results vary by individual and aren't guaranteed
Read the three key metrics →