Automation Before Automation (ABA): The Missing Phase in Modern Software Testing
We've all been there. You just finished (or changed) an API endpoint. You open Postman or cURL, send a few requests, play around with the data, and then what?
Most teams jump straight from this quick manual check into writing "proper" automated tests. But in recent years, something important has appeared in between — something that is already happening every day but didn't have a proper name. I call it Automation Before Automation (ABA).
What is ABA?
Automation Before Automation (ABA) is the practice of automatically generating and running hundreds of exploratory, validation, robustness, and protocol-level tests right after software is built — before you start creating long-term, maintainable regression test suites. It’s automated exploration before automated verification.
Think of it as the bridge between the chaotic "let me just try this" phase and serious test automation. Instead of manually poking around, you let a tool generate dozens or hundreds of variations — invalid inputs, boundary cases, protocol quirks, security headers, oversized payloads, and more.
Why This Matters Now
Traditional testing literature usually talks about two big things:
Exploratory testing (manual, creative investigation)
Automated regression testing (stable, repeatable checks)
But reality has changed. Thanks to fuzzers, mutation tools, schema-driven generators, and AI-assisted testing, teams are now doing high-volume automated exploration very early. This activity doesn’t really fit into either classic category. That’s why I defined ABA as a distinct intermediate phase.
Key Characteristics of ABA
Tests are mostly generated, not hand-written
High volume, fast feedback
Disposable (you often throw them away after the session)
Focused on discovery, not regression prevention
Low maintenance
In short: ABA answers the question "What should we automate?"
Traditional automation answers "How do we continuously verify it?"
They complement each other — they don’t replace one another.
The industry has been doing Automation Before Automation for a while without naming it. By giving it a clear name and framework, we can start treating it as a proper testing layer — one that sits between quick manual checks and heavy regression suites.
If you work in software development, I believe you’ll recognize this phase in your own workflow.
Read the full White Paper: https://qaontime.com/research/automation-before-automation.html