Saw a question on a Q&A site about shift-left testing not delivering results; here I share Qiucao’s answer

Summary
The questioner’s problem is actually not simply a matter of how to shift testing left. The so-called “testing becoming a bottleneck” is more the result of various problems piling up.
And applying agile development does not mean it is agile just because you run an iteration every two weeks. From the question, we can see that this iteration actually still ran as a mini-waterfall, still with requirements, design, test submission, testing; the team did not actually become agile.
Challenge Analysis
Let’s first analyze the many challenges the questioner mentioned:
- Incomplete requirements documents
Agile indeed does not emphasize documentation, but that does not mean requirements within an iteration can be unclear and incomplete.
In agile practice, before an iteration begins, the PO should have completed the PBI list and worked with the team to complete requirements grooming. And an important condition for starting an iteration is that at the sprint planning meeting, the team jointly confirms the SBI and the workload. Uncertain requirements should not enter the SBI.
Of course, agile embraces change; changes and additions to requirements during an iteration are possible, but it is not advocated to have them frequently within an iteration, because with fixed team productivity, changes cause WIP artifacts to change, and burndown charts and velocity charts will deviate from the ideal curve. Mature agile teams avoid such situations as much as possible.
- Low developer involvement in testing
An agile team actually does not define two different roles, development and testing. That is, in agile, there is no absolute division of labor between development and testing.
“Eating your own dog food” is a phrase often mentioned in agile practice, which essentially makes clear the necessity of developer self-testing.
Agile in which developers do not test is necessarily pseudo-agile.
- The team lacks an effective communication mechanism
This is even harder to explain.
All agile principles and practices are fundamentally aimed at breaking down barriers and improving communication and collaboration efficiency. If it is still hard for the team, especially development and testing, to communicate, then this is simply not agile at all.
So, first make agile actually agile, then look at the results of shift-left testing.
Should We Implement Shift-Left Testing?
What does the questioner’s shift-left look like?
Have testers get involved in requirements review earlier, and also try to write more detailed test plans, but the effect is not obvious
First, we need to stress again that in an agile team there is no separate role of tester; apart from
POandSM, all members are theDev Team. So there is no such thing as “letting testers participate in requirements review earlier” - in requirements grooming and SBI determination, testing should necessarily be involved.As for more detailed test plans, this is actually weakened in agile; agile advocates test strategies like one-page plans, where clarifying 6W2H before testing is enough.
So the so-called shift-left here is originally the normal operation of testing.
Even in waterfall mode, requirements review and test planning are things testers should participate in.
This is not real shift-left!
How to Shift Left
So in agile, when we talk about shift-left testing, what should actually be done?
The normal understanding is still early participation in requirements review and design review, as well as pairing with developers while coding, assisting with unit testing, and so on.
But I want to say that real shift-left testing in agile is not simply moving testing activities left. Testing activities should run through the entire development activity, and every team member should carry them out and be responsible for them.
Just as BDD and ATDD practices follow the adoption of agile, the starting point of shift-left should be concrete business and user acceptance scenarios. Or more directly, the requirements themselves.
Shift-left testing should be more about testers, as a quality perspective within the team, injecting quality and risk control genes from the very beginning (from when the PO creates requirements), and helping the whole team build and popularize quality awareness and conduct risk control.

Improving quality and efficiency by a single testing role alone is empty talk. Only when the whole team values quality and embeds risk avoidance into all activities from the very beginning is it fundamental.
