Pillar Drift
Trend
The directional signal PCR cannot provide — how fast a pillar is improving or deteriorating.
This page describes the intended design. A null placeholder for it already exists in the metrics data model, but nothing computes a real value yet. It doesn't appear on your dashboard yet, and no org currently has a live value for it.
− PCR_30_days_ago
What it signals
PCR is a point-in-time snapshot. It tells you where a pillar stands today. PDT is the directional signal — it tells you which way a pillar is moving and how fast. Two pillars at 75% PCR are in the same place. One with PDT of +8 is recovering. The other with PDT of −12 is deteriorating and will reach crisis before PCR alone signals it.
A pillar at 75% PCR with PDT of +8 is not a problem that needs your attention today. A pillar at 88% PCR with PDT of −9 is a problem that is not yet visible in PCR — but will be within two sprints. PDT surfaces that risk before PCR crosses a threshold. That is the gap it fills.
PDT is the early warning PCR cannot provide. Without it, you are looking at a number that tells you where you are. With it, you are looking at a vector that tells you where you are going. Architectural governance that waits for PCR to fall below 70 before acting is reactive. PDT makes it predictive.
How rkito produces it
Who this metric is for
PDT tells the VP Eng which architectural domains are in decline before they become crises. A pillar with a healthy PCR but three consecutive negative PDT readings is a domain to watch — not to ignore. PDT converts architectural health from a status report into a risk register.
PDT tells the architect where to direct their attention in the next review cycle. A pillar with stable PCR and flat PDT does not need an architecture review. A pillar with declining PDT does — regardless of where PCR currently sits. PDT prioritizes architect time before it is urgently needed.
Single-sprint declines are noise
PDT of −3 for one sprint is noise. A large batch of PRs, a temporary team context switch, a focused refactoring push — any of these can produce a negative PDT reading that reverses the following sprint without any structural change.
PDT of −3 for three consecutive sprints is a structural signal. That is not variance — that is a pillar losing ground consistently, and the cumulative delta is −9 PCR points over the period. That warrants investigation.
Do not escalate single-sprint declines to leadership. They create alarm without diagnostic value and erode trust in the metric. Wait for the trend to confirm across two to three review periods before acting.
What to do at each threshold
Step-by-step response playbook for each signal state.
Pillar conformance is improving or holding. No remediation required.
Note stability in next architecture review. A positive PDT sustained over two or more sprints is worth surfacing — it signals the pillar is governed and improving.
No action required. Continue monitoring at the normal review cadence.
Pillar is declining slowly. Observe the trend before escalating.
Watch for two more review periods. A single yellow PDT reading does not warrant action — it warrants attention.
If the negative trend persists across two or more consecutive periods, open a Lens review with the pillar owner before PCR crosses into yellow territory.
Pillar is declining sharply. Immediate pillar owner engagement required — but root-cause before prescribing a response.
Four root causes account for most red PDT readings, and each requires a different response: new team onboarding (awareness gap), new tech stack (Lenses not updated for new patterns), outdated Lenses (now misfiring on legitimate patterns), or genuine architectural drift.
Do not conflate root causes. An awareness gap needs targeted enablement. Outdated Lenses need a recalibration sprint. Genuine drift needs a structured remediation plan. Treating any one of these as another wastes architect time and erodes trust in the governance system.
The point-in-time state PDT tracks over time. PCR tells you where a pillar is. PDT tells you which direction it is moving.
Once a pillar enters decline, PDRP measures how quickly it recovers. PDT identifies the decline; PDRP tracks the remediation.
The org-level aggregate. PDT per pillar is the early signal that, if unaddressed, will eventually move DCS. PDT is the upstream warning.
Which of your pillars is drifting?
rkito computes PDT per pillar from your first design audit. See which domains are recovering and which are deteriorating — before PCR alone can tell you.