You are about to change something: switch the deployment method in your lab, restructure your study plan, or tune the pipeline your team uses at work. Stop for an hour first. This checklist walks you through capturing a baseline, the factual before-picture that will let you say honestly, later, whether the change helped. Work through it in order, tick each item and file the result before you touch anything. It is a capture list, not a grading exercise, so nothing here produces a score.
Part 1: Define the target
- Write one sentence naming the result you want to move, for example “shorten the wait between merging a change and seeing it run in the test environment”.
- Decide the single change you intend to try. If you have three ideas, queue them; test them one at a time.
- Note who will be affected and who must be told about the measurement.
Learners often skip this part and gather numbers at random. A clear target is what keeps the record short and the comparison meaningful. The thinking involved is the same that building sharper problem-solving habits in DevOps asks for, namely stating the problem precisely before reaching for a fix.
Part 2: Describe the process as it runs today
- List every step in order, including the manual ones and the waiting.
- Mark the steps that people often forget or perform differently from each other.
- Save a copy of the relevant configuration or notes exactly as they are now.
- Record the tools and versions in use, and the environment they run in.
Resist cleaning anything up. If the real process contains an odd workaround, write it down. Workarounds are often where the largest improvements hide.
Part 3: Time it, more than once
- Time each stage separately and the whole routine from start to finish.
- Take at least three readings under similar conditions.
- Note the fastest and slowest reading; the gap is the normal spread.
- Record the time of day and anything unusual happening nearby.
The spread matters as much as the average. A change that moves things by less than the normal spread cannot be told apart from luck.
Part 4: Add up the cost
- Infrastructure or licence charges that the routine consumes.
- Hours spent by people on the routine, including supervision and rework.
- Any recurring cost of keeping the environment available, such as a lab you leave running.
For students, the cost may be mostly time: evenings spent debugging a setup. That counts, and it deserves a number.
Part 5: Collect the soft signals
- Group recurring complaints from teammates or classmates into themes.
- Note which questions people ask again and again.
- Write down how a newcomer experiences the routine on their first attempt.
These notes tell you what an improvement should feel like to the people living with it. They often reveal gaps in basics; the gaps students need to close in DevOps are a good lens for deciding which questions deserve a place on your list.
Part 6: Lock the record
- Put the date at the top and the names of the people who took part.
- Describe the measuring method so that it can be repeated identically.
- Store the file somewhere shared and mark it read-only.
- If you spot a mistake later, add a dated correction underneath rather than editing.
Before moving on, read the page back as if you were a colleague who had never seen the system. Could they repeat your measurement from the notes alone? If a step relies on something only you remember, such as which terminal profile you used, add it now. A baseline that depends on one person’s memory fails the moment that person is on leave, and the purpose here is evidence that a stranger could pick up cold.
Part 7: Make the change and repeat
- Apply the one change you chose, and nothing else.
- Measure again using the identical method, the same number of readings and similar conditions.
- Place the two records side by side and compare ranges, not single figures.
- Write the verdict in plain words, including “no clear difference” if that is what you found.
A related habit from outside the lab
The same sequence suits any plan with many unknowns, including arrangements that cross borders. A directory such as BestEyeHospitals is aimed at international patients planning eye care and sets out its own consultation steps; its self-written medical disclaimer treats the pages as general reading and asks that nobody hold back from seeing a professional on their account. Whatever the subject, the sensible order is to capture your own starting requirements first and read third-party resources second.
Common ways the checklist gets skipped
- “We will measure afterwards.” Once the change lands, the earlier setup has usually been overwritten.
- “One reading will do.” It will not; variation will fool you.
- “We will change two things to save time.” You will not know which one worked.
- “The result was disappointing, so we did not write it up.” Unreported flat results lead teams to repeat the same experiment.
When certification study is the change
The checklist is not only for pipelines. If you are deciding how to prepare for an exam or credential, record how you study now: hours per week, how often you practise in a lab, how confident you feel on each topic in your own words. Afterwards, compare. A discussion of what DevOps certifications are really worth in a career can help you decide what outcome you want, and your baseline tells you whether your new routine moves you toward it.
Print the seven parts, keep them beside you and treat the hour they take as part of the change itself. A thorough before-picture is dull to produce and wonderfully useful later, which is a fair description of most good engineering habits.

