Definition
A/B testing is an effective way to compare two versions of a webpage or mobile app and see which one gets better results. This technique helps you determine what works best so you can optimize your content and engage more users.
In the room
How the word shows up in a real conversation.
A software startup is developing a new productivity app for mobile devices. They have invested a lot of time and resources into the design and development of the app, and it will be ready to launch soon.
At the startup, they make an A/B testing experiment. They create two versions of the app and have slight differences in their user interface, features and functionality.
They then randomly divide their existing customer base into two groups, with one group using Version A and the other using Version B. The team closely monitors the usage and feedback from each group.
After the experiment was wrapped up, it was determined that Version B had higher engagement & adoption rate and better feedback.
Based on these results, the team decides to launch Version B of the app to the public. The A/B testing experiment allowed them to validate their design choices & make data-driven decisions and improve the overall quality of their app before launching it to the market.
Common mistake
Calling a before/after change an A/B test. If traffic was not split at the same time, you measured a season, not a variant.
India note
On India traffic, split by city and device. A Delhi desktop win can hide a tier-2 mobile loss.
See also
Product-Market fit
Degree to which a product satisfies market demand and customer needs.
Read definition →
Customer acquisition
Process of gaining new customers through marketing and sales efforts.
Read definition →
Churn rate
Percentage of customers who stop using a service over a given period.
Read definition →
Minimum Viable Product (MVP)
The smallest product that can test the riskiest assumption with real users.
Read definition →
From the store
Research more
Open the word in an AI tool with a ready prompt. We do not send your data.
Get one word every week
A 3-minute read: 1 word, 1 founder lesson, 1 tool.