Deciding Data
  • Home
  • Blog

Recent posts

  • How does Deciding Data work?
  • Why not build your own?
  • Why not just let the platform learn?
  • Is this difference significant?
  • The World Makes the Choice
  • When Creative Is Cheap, Learning Is the Bottleneck
  • Why the Fastest Teams Are the Ones Who Test
  • Your Creative Process Can Validate Patterns
  • How Our Eyes Taught Me About Smarter Ad Tests
  • Why You Want to Watch Out for Related Features
  • Mastering The Science of Marketing: Experimentation Processes
  • The Problem of Cumulative Advantage
  • How to Improve Your Site with Coordinated Tests
  • How much should I spend testing new ads or audiences?

How does Deciding Data work?

October 29th, 2026
Diagram: one loop. Theories claim hypotheses. Hypotheses suggest the next test. You run it, and your tests add to the evidence. The evidence revises the theories. Hypotheses are also checked against the evidence.

We turn your past tests into ideas about your customers, then help you test the most promising one. Each test adds to what your team knows.

Continue reading...

Why not build your own?

October 29th, 2026
Illustration: a pantry closet drawn front-on. An ice cream maker sits on the middle shelf among folded towels, a cooking pot, and jars, its cord wound around its base.
I own an ice cream maker. I still buy my ice cream at the store.
Continue reading...

Why not just let the platform learn?

October 29th, 2026
Diagram: two loops. Left, platform learning: your team's new ads are tried on different people, spend moves to the winners, and the only thing that returns to your team is a request to send more ads. Right, Deciding Data: your tests become findings about which traits drive results, and those return to your team as what to make next.
Meta's tests teach Meta's algorithm. Ours teach you.
Continue reading...

Is this difference significant?

October 8th, 2026
Chart: three 95% ranges for urgency over today's subject line, in clicks per 1,000 recipients. Test send: +0.4 to +9.6. Small send: −4.1 to +14.1, crossing zero. Whole-month send: +0.2 to +2.8.
What a significance test can and cannot tell you about an email test, and the numbers that answer the questions a decision needs.
Continue reading...

The World Makes the Choice

July 17th, 2026
Tools expand the options, experiments narrow them
Cheaper, faster tools give us more creative options than ever across email, paid ads, and our sites. Experiments do the reverse: they narrow the field to what actually works.
Continue reading...

When Creative Is Cheap, Learning Is the Bottleneck

April 15th, 2026
Two rows of stages, Ideas to Tests to Understanding. Before, Ideas is the smallest stage. Now, Ideas and Tests have grown while Understanding stays the same size, so Understanding is the bottleneck.
AI has solved the creative production bottleneck for ad teams. The new constraint is learning velocity, how fast you can turn test results into genuine understanding.
Continue reading...

Why the Fastest Teams Are the Ones Who Test

December 5th, 2025
Timeline comparing creative cycles with and without systematic testing
Speed and testing aren't opposing forces. Learn why systematic experimentation actually accelerates your creative cycle, especially during peak seasons.
Continue reading...

Your Creative Process Can Validate Patterns

December 5th, 2025
Diagram showing the validation cycle: estimate patterns, create new ads, validate or falsify, repeat
You don't need to build a separate system to validate insights from experimentation. Your content creation process does it.
Continue reading...

How Our Eyes Taught Me About Smarter Ad Tests

May 13th, 2025
Spending distribution over time
Discover how the biological concept of eye saccades inspired a great approach to ad testing, helping us find clear signals in a noisy marketplace and deliver more reliable, and faster results.
Continue reading...

Why You Want to Watch Out for Related Features

April 30th, 2025
Diagram showing related ad features and their associations
Learn how related ad features can weaken your test results and discover strategies to isolate true drivers of performance.
Continue reading...

Mastering The Science of Marketing: Experimentation Processes

April 4th, 2023
Diagram of three experimentation processes
How to build experimentation processes to continuously update your most important decisions.
Continue reading...

The Problem of Cumulative Advantage

March 9th, 2023
An abstract representation of growth
Cumulative advantage can increase the randomnes in ad performance. We have some ways to see through it, and use it.
Continue reading...

How to Improve Your Site with Coordinated Tests

September 29th, 2022
Diagram: eight ideas, Idea 1 to Idea 8, are tested in ads, where traffic is plentiful. Two of them, Idea 3 and Idea 6, are refined on the site, where there are fewer visitors and each is worth more. The ads send lessons and traffic to the site; a better site lowers what each ad costs. Deciding Data tests the ads; the site's own testing tool tests the site.
Experiments that focus on the ideas you are testing, rather than the platform, can help you learn even faster.
Continue reading...

How much should I spend testing new ads or audiences?

March 31st, 2022
A person exploring the mountains
Happily, there is a technique that does a great job finding the best balance between testing new ads and using the best ones we've found.
Continue reading...
Blog
Docs
Contact Us
Login
Experiment Results
Privacy Policy
Terms and Conditions
© 2026 Deciding Data Inc.