Collecting Baseline Data

Before you can tell whether your feature made a difference, you need a picture of what “normal” looks like. Baseline data captures how the metric behaved before your change — giving you the “before” in a before-and-after comparison.

Where does the data come from?

By default, VEKTIS collects it for you. When a dev item is set to “VEKTIS measures it for me” (the default), VEKTIS reads baseline measurements from your product’s analytics automatically — matched to the dev item by its Feature ID and measurement type. Brand-new features with no history yet establish their baseline through a short burn-in period. Nothing to type in.

To turn this on, add the VEKTIS tracker to your product — the Quickstart walks you through it, and its Use AI option can point your AI assistant at these docs to wire it up for you. You pick automatic vs. manual when you create the dev item.

Entering baselines by hand is optional. If you chose “I’ll enter results myself” — for a metric VEKTIS can’t reach automatically — you pull the number from wherever you already track it (analytics dashboard, observability tool, internal reports, a spreadsheet) and add each data point yourself. This works with any metric from any source, no integration required.

Example: For a metric you’re entering by hand — say “Page load time” — you’d check your performance monitoring tool for the past few weeks’ average load times and enter each week’s value as a separate baseline data point.

When to collect baselines

You can add baseline metrics while your dev item is in Planning, Development, or Measuring status. The key rule: each baseline data point must be dated before your release date (if one is set).

Tip

Start collecting baseline data as early as possible — even while the feature is still in planning. The more history VEKTIS has, the better it can distinguish real changes from noise.

Adding a baseline by hand

These steps are for dev items you measure yourself. If VEKTIS measures the dev item automatically, baseline data fills in on its own and you can skip this.

1. Click on your dev item from the dashboard to open its detail page — you’ll find the Baseline Metrics section below the dev item details
2. Click “Add Baseline Metric”
3. Fill in the fields:
   • Date — When this measurement was recorded. Can’t be in the future or after your release date.
   • Value type — How the number should be displayed (e.g., percentage, count, currency). See Metric Value Types for all options.
   • Value — The actual measurement
   • Notes — Optional context (up to 500 characters)
4. Save — VEKTIS recalculates your baseline statistics automatically

Note

Each date can only have one baseline metric per dev item. If you need to correct a value, edit the existing entry rather than creating a duplicate.

Value type consistency

The first metric you add — whether baseline or post-release — locks the value type for that dev item. All subsequent baseline and post-release metrics must use a compatible type.

• Duration types (seconds, minutes, hours) can be mixed with each other — VEKTIS converts them automatically
• All other types must match exactly — you can’t mix percentages with counts, for example

This ensures your before-and-after comparison is always apples-to-apples.

Baseline strength

As you add data points, VEKTIS shows how reliable your baseline is:

Strength	Data points	Recommendation
Insufficient	0–1	Add more data — can’t calculate statistics yet
Weak	2	Bare minimum. Results will work but may not be accurate
Moderate	3	Good start. One more data point gets you to “Strong”
Strong	4+	Reliable baseline — you’re ready to ship with confidence

Caution

You can mark a feature as released with a weak baseline, but your impact results will be less reliable. If the metric normally fluctuates a lot, you may need even more than 4 data points for accurate results.

Editing and deleting baselines

You can edit or delete baseline metrics while the dev item isn’t in Complete or Archived status.

• Editing — Change the date, value, value type, or notes. Date and type compatibility rules still apply.
• Deleting — Removes the data point and recalculates statistics. If this drops you below 2 data points, statistics can’t be calculated.

Best practices

1. Aim for 4+ data points — This gives VEKTIS a strong picture of normal behavior
2. Spread measurements over time — Weekly data points over a month capture natural variation better than 4 measurements in 2 days
3. Be consistent — Measure the same thing the same way each time
4. Include typical variation — Don’t only measure on “good” days. Include normal ups and downs to give VEKTIS an accurate sense of the range.
5. Record before you ship — You can add baselines after releasing, but dates must still be before the release date

Note

VEKTIS monitors the regularity of your baseline measurements. If your data points are spaced at very inconsistent intervals — or if multiple measurements fall on the same day — you’ll see a warning encouraging you to establish a more regular cadence. Consistent spacing leads to more reliable baselines.