Reflections on a Four‑Week Research Project About Google’s Agile Core

From 2017 onward, the cloud community has been buzzing about agile and DevOps, always trying to break through the testing barrier and do more front‑line work.

About a month ago, I saw the "DevOps Case Deep‑Research Program" and immediately signed up. After applying and being evaluated, the topic "Google Agile Core" landed on my desk, and I was appointed as the research team leader.

Excited yet nervous, I had to deliver an in‑depth study within four weeks.

First, I watched a video of last Sunday’s case‑sharing session to feel the lively atmosphere.

Following my usual approach, I discussed with the team members and started gathering materials. The difficulty quickly appeared – what exactly is Google’s agile core? How does Google practice agile? No one on the team knew, there were no references, no concrete practices, and no way to probe the problem.

After the first week’s sprint review, we realized we had found no Google‑related agile content.

Sprint 1 over!

Sprint 2 began with little progress, and the SOS meeting highlighted the difficulty.

Other groups faced the same issue, so we received some suggestions, slightly adjusted our goal, and made "Sprint Design" and "Redefining the Team" the core research objects.

How should we plan the time and implement the work?

After discussion, the plan was as follows:

Sprint 2 – Create a mind map to define the scope

Sprint 3 – Prepare PPT and simple trial presentation ideas

Sprint 4 – Refine the PPT

Through SOS meetings and subsequent iterations, the process seemed to follow the plan, but many problems arose.

Because the team was fully autonomous and remote, progress control was difficult. Everyone had their own primary work, family responsibilities, reading, and summary writing. Network‑based communication also had many issues. In this case, my laissez‑faire autonomous management failed for me, but the team as a whole succeeded.

At the start, the biggest confusion was that members didn’t know what to do because only the result was defined, not the process or weekly plans. This led to a feeling of “not knowing what to complete” similar to writing a weekly report without clear tasks.

Until the third iteration, after the mind map was organized and PPT presentations were given, team members began to sense the problems, realizing the lack of clear logic and the need to speak in a more understandable way.

After four iterations, on July 7th at JD’s Shanghai base in Beijing, we delivered and showcased our research findings on "Google Agile Core", completing the final delivery.

The deliverable scored about 60 out of 100; 20 points were lost because I felt we didn’t fully organize and describe how Google conducts testing in a systematic way, and another 20 points were deducted for being somewhat superficial and offering many objective excuses.

Finally, thanks to Dong Ge, SRE Xu Lei, Teacher Li Jie, and the organizers for the meeting, as well as JD’s Dan Bing and Guang Yin for the venue and coordination; sharing with many teams was immensely beneficial.

For those interested in similar activities, stay tuned for next week’s community line‑up events and scan the QR code in the poster below to register.