Measurement Regularity
Baseline measurements that arrive at uneven intervals can distort the “normal range” VEKTIS uses to score impact. When this happens, you’ll see a regularity warning on your dev item — a heads-up that the baseline may be less reliable than it could be.
How VEKTIS spots irregular measurements
Section titled “How VEKTIS spots irregular measurements”A regularity warning appears when both:
- You have at least three baseline measurements
- The intervals between them aren’t reasonably consistent
| Pattern | Why it triggers a warning |
|---|---|
| Two or more measurements recorded on the same day | The intervals collapse to zero — VEKTIS can’t tell what your “normal cadence” is |
| Largest gap is more than 3× the smallest gap | One stretch of measurements is much sparser than another, suggesting inconsistent tracking |
| Otherwise even spacing | No warning — your cadence looks regular |
Why cadence matters
Section titled “Why cadence matters”Statistical impact scoring works by comparing post-release measurements against the spread of your baseline values. If those baseline values were collected at wildly varying intervals — daily for a week, then nothing for a month, then once more — the spread reflects measurement timing as much as it reflects the metric’s natural behavior. The result: weaker confidence in whether a post-release change is real.
How to space measurements
Section titled “How to space measurements”You don’t need to be perfect. The goal is consistency, not precision.
Practical guidance
Section titled “Practical guidance”| You see | What to do |
|---|---|
| Regularity warning on a baseline you’re still building | Stick to your cadence going forward — once you have enough evenly-spaced data, the warning will disappear on the next measurement |
| Warning appears after you missed a few weeks | Resume regular measurements; the older irregular data still counts toward the baseline, but new even-spaced data will dilute its effect |
| Warning persists after consistent recent measurements | Earlier irregular measurements are dragging the assessment — consider whether to remove obvious outliers |
| No warning, but you suspect noise | Add more measurements — more data points beat fewer, even at the same cadence |