This is a made-up story, though a familiar one. A small platform team at an imaginary online retailer is told to move the checkout service to a new deployment setup before the busy season. Everyone is capable, and everyone is confident. What saves the project is not a clever tool but a plain page titled “Things we are assuming,” started in the first week. If you have ever wondered what writing down assumptions looks like inside a real team, read on.
Week one: the kickoff and the awkward question
The team has four engineers, a product owner and a manager who hears about the work mostly in summaries. In the kickoff, the most experienced engineer asks a question that gets a few uncomfortable looks: “What are we taking for granted?” Nobody has a ready answer, so the team does what the engineer suggests and spends twenty minutes at a whiteboard. By the end there are eleven sentences. A few examples:
- The checkout service has no hidden dependency on the old server’s local disk.
- The payment provider’s sandbox behaves like its live service for our purposes.
- Traffic during the sale will resemble last year’s pattern.
- The database can be copied without a long outage.
Each sentence receives a confidence word and, where possible, an owner. The product owner takes the traffic belief, since that role holds the sales calendar. One engineer takes the local-disk question, being the person with the code open all day.
Week two: the first surprise
The local-disk assumption had been marked “probably fine.” The engineer decides to spend an hour checking it anyway, since the cost is low, and finds that the service writes temporary invoice files to a path on the old server and a nightly job reads them. Nobody had mentioned it because it had always worked, and a thing that always works stops being talked about. Had the team discovered this at cutover, the nightly report would have broken on the first night in the new setup, in the same week as the sale.
Instead the discovery is logged on the page: the line is struck through, replaced by what was found, with a note of the fix chosen. The team does not make a fuss. They simply move on to the next line.
Week four: a disagreement that the list settles
Two engineers disagree about whether the new setup needs a separate warm standby for the database. One says traffic will not justify it; the other has a bad feeling. The old habit would be a long debate. Because the page already contains a line about traffic, both can point to it. The line says “resembles last year’s pattern, confidence medium, check pending.” The check is overdue, so they run it that afternoon: they compare recent weeks with the same weeks of the previous year using the figures they have. The pattern roughly holds, but one campaign launched later than usual, so the peak will be steeper. The standby goes ahead, and the argument ends with a fact instead of a winner.
Week six: the line nobody wants to test
Every list has one. Here it reads “The support team can handle a rise in tickets during cutover.” It is marked low, and for two weeks nobody touches it because it feels like someone else’s job. The manager, reading the page, sees the lonely low-confidence line and has a short conversation with the support lead. The result is a plan for extra cover on the night of the move. It is a human arrangement, not a technical one, and it would not have appeared in any architecture diagram. The team can see that the page paid for itself on the people side as much as the code side.
For a wider view of the people dimension of this kind of change, the discussion of managing the cultural side of DevOps adoption pairs naturally with the story. Likewise, anyone curious about the ways delivery work can be made safer can browse the guide to handling risk when shipping software reliably.
A short aside on the confidence words
The team had briefly argued about whether to use percentages. They settled on three plain words instead: solid, probable and unsure. Percentages looked scientific but invited false precision; nobody could honestly say whether a belief was seventy or eighty percent likely. Words were quicker to assign and easier to discuss. The only rule was that “solid” had to be backed by something someone had actually seen, such as a test run, a log excerpt or a written answer from the owner of a connected system.
That rule quietly changed behaviour. People began saying “I have seen it” or “I have only heard it” in conversation, which sharpened ordinary stand-ups in ways nobody had planned.
Cutover night
The move goes quietly. Because the assumptions page had gradually changed from eleven hopeful lines to a mix of confirmed lines, two struck-through ones and three added late, the team arrives at cutover with a realistic picture. One surprise does appear: a certificate renewal date that nobody had listed. It takes minutes to fix, and afterwards the team adds a fresh line and makes renewals a standing item for the next move.
What the team took from it
- The page was short enough to be read in a stand-up.
- Every line had an owner and a confidence word.
- Checks were small: an hour, a conversation, a comparison of figures.
- Wrong assumptions were treated as good news, since they were found early.
- The list survived the project and became the starting point for the next one.
Borrowing the idea for your own team
You need no special software. Start with a shared document and the title “Things we are assuming.” Ask the group to fill in five lines in ten minutes, without debate, and then rate them together. Anything below high gets a check and an owner. Review the page at the same cadence as your regular meeting, and keep the history of changes.
If you want to see how training-oriented sites in the field present related material, one example is NoOpsSchool, an India-based website that, by its own description, lists certifications, courses, consulting and service pages for IT professionals. It is only one reference point; the habit is yours to shape.
Closing thought
The retailer in this story is fictional, but the pattern is not. Big moves seldom fail on the dramatic risks everyone discusses; they fail on small beliefs that nobody wrote down. A page, a few minutes and a willingness to say “we are not sure” can change the odds in a project’s favour.









