Decision Drift
Decisions rarely fail in one moment. They drift—quietly—until reality no longer matches the original intent.
Most decisions do not break.
They drift.
Not because someone reversed them. Not because the team changed its mind.
But because reality kept moving while the decision was never revisited.
Decision Drift is the silent divergence between a decision’s intent and the system’s lived reality.
Drift Is Not a Failure Event
Teams often look for a single point of failure.
A release. A commit. A specific change.
But drift rarely has a clean timestamp.
It happens through small, reasonable steps:
- a safeguard is added
- an exception is introduced
- a “temporary” workaround becomes permanent
- a new constraint appears and silently reshapes the behavior
Each step is defensible.
The result is not.
The Most Common Pattern
A decision is made.
The team has a clear intent:
- what problem is being solved
- why this trade-off is acceptable
- what signal should change if the decision was right
Then the decision ships.
And it collapses into code, configuration, and habit.
Weeks later, the system still exists. The decision does not.
Only outcomes remain.
Dashboards show movement. But not meaning.
A Simple Example
Decision: Reduce onboarding friction.
Expectation: Time-to-value should go down.
Then, over time:
- a new verification step is added “for safety”
- a modal appears “to clarify”
- a new field becomes required “for segmentation”
- an edge case adds a fallback flow
No one announces a reversal.
But onboarding is now heavier.
If the team looks at metrics, they see only the outcome:
- drop-off increased
- completion time went up
What they cannot see is the drift:
- the system no longer reflects the original decision
Why Drift Is So Hard to Notice
Decision Drift hides inside good intentions.
It is rarely caused by malice.
It is caused by:
- local optimizations
- different teams solving different constraints
- a growing set of exceptions
- the absence of a shared decision memory
Each change is small.
But drift compounds.
And once compounded, it becomes political:
- “We never decided that.”
- “That’s how it’s always been.”
- “We had to do it for X.”
All of these can be true.
And that is the problem.
Drift vs. Bugs, Drift vs. Debt
Drift is not a bug.
Bugs are deviations from specified behavior.
Drift is a deviation from intended reasoning.
It is also not technical debt.
Technical debt is code-level cost.
Drift is decision-level cost.
When drift accumulates, it eventually becomes Decision Debt:
- more exceptions
- more workarounds
- more “we’ll fix it later”
The system pays interest.
But the team cannot locate the principal.
The Silent Revert Is Drift’s Cousin
Sometimes drift ends with a hard outcome:
The decision is effectively reversed.
But no one says it.
No note. No timestamp. No explicit learning.
That is a Silent Revert.
Decision Drift often creates the conditions for Silent Reverts.
Because when intent has already dissolved, reversal feels “natural.”
The Real Cost: Learning Breaks
The biggest cost of drift is not the outcome.
It is the loss of learning.
When drift happens, teams are forced to guess:
- what we thought would happen
- why we chose this path
- which assumptions were true
Without that context:
- success is misattributed
- failure is moralized
- retrospectives become storytelling
And the same debates return.
Not because teams are irrational.
Because the decision no longer exists as an object.
What Stops Drift
Drift is not stopped by more dashboards.
Dashboards operate downstream.
Drift begins upstream.
What stops drift is treating decisions as durable entities:
- the decision is explicit
- expectations are explicit
- signals are defined
- reality is checked against those signals
- revisits are normal, not shameful
In other words:
Decision-Centric Development.
A Practical Test
Ask this about any important change:
“If we come back in 90 days, will we still know what we expected to happen?”
If the answer is “probably not,” drift is not a risk.
Drift is already scheduled.
The Point
Decision Drift is what happens when decisions are treated as moments.
But decisions are not moments.
They are living entities whose correctness is tested over time.
If teams want learning—not just motion—drift must become visible.
Not to blame.
To remember.
Afterchange Team
Helping teams track decisions and measure impact.