# Metrics, baselines, and readings

> Metrics, baselines, readings, assumptions and value formulas: how Roiva calculates value, and what to record next.
>
> Source: https://roiva.ai/docs/metrics-baselines-and-observations
> Checked against the product on October 06, 2026

Roiva calculates value by comparing what things were like before an initiative with what they're like now. Four pieces make that work. Metrics, assumptions and formulas each have a tab under **Method** in the sidebar, beside **Chain** and **Charters**, and metric entries are under **Entries → Metric Entries**.

#### Metrics

A **metric** is something you measure: tickets resolved, agent minutes per ticket, won deals, tokens used. Each has a type (count, sum, average, ratio, percentage, currency, duration or score) and a unit. A currency reading names its currency: a CRM's won deal value for a month is in the deals' currency when they share one, and in dollars at the European Central Bank's rate when they mix. The **Metrics** tab opens on **In Use**, the metrics your initiatives hold figures for and your formulas read or produce, once there are any. **Library** lists every metric you can use, Roiva's and your own, and **New Metric** adds your own.

#### Metric entries: baselines and readings

A **metric entry** is one value of a metric for a period. Two kinds matter:

- A **baseline** is the before-value: what the metric was before the initiative started.
- A **reading** is a value after it: this month's agent minutes per ticket, say.

Metric entries come from connected integrations, from a query against your own data warehouse, from value formulas, from CSV import (**Entries → Metric Entries → Import**, which takes baselines too), by hand, or through the Roiva Assistant, which marks the ones it records as its own.

A metric entry is called an **observation** in the API, in the MCP tools and in the columns of the CSV export and import. It is the same record under the engine's own name.

Baselines mostly fill themselves in. When your connected systems have figures for the three complete months before an initiative started, or for any of them, Roiva calculates the baseline from them. Otherwise Roiva may hold an estimate, and replaces it with a baseline from your history if those months sync in or are imported later. A baseline you enter or import yourself is never overwritten.

Every baseline says how it was measured: from history, by a time study before go-live, by a sample timed without the AI after it, from a benchmark, or as an estimate. An estimate says what it rests on. The method decides whether the value calculated from it rests on a measured, reconstructed or estimated before-figure, and every report shows that split. For an initiative that is already live, see [Baselines for an initiative that's already live](https://roiva.ai/docs/baselines-for-an-initiative-already-live). Where no system records how long the work takes, [run a time study](https://roiva.ai/docs/running-a-time-study): the arm timed without the AI becomes the baseline and the arm timed with it a reading, each with an interval and a date it's due again.

#### Assumptions

An **assumption** is a figure a formula takes as given: a loaded labor rate, minutes saved per automation run, a price per million tokens. Roiva ships a default for most, usually with its source, and the **Assumptions** tab is where you confirm or change them. While a formula runs on a Roiva default, the value it calculates is held for review.

#### Value formulas

A **value formula** is the arithmetic that turns readings, baselines and assumptions into value entries: for example, (tickets agents resolved) × (agent minutes per ticket before − minutes now) ÷ 60 × (labor rate). That one prices people's time, so it books freed capacity rather than cash. The **Formulas** tab (**Method → Formulas**) lists them, and each shows its inputs and what it calculated. A calculated value entry's **Formula Inputs** section shows the arithmetic behind that one entry. See [How Roiva calculates value](https://roiva.ai/docs/how-value-is-calculated).

#### What to record next

- The **Missing Inputs** page lists the figures formulas are waiting on, and you can enter them there. Some say where to find the figure.
- The **Inbox** lists every figure your active initiatives need from a person — a missing baseline, an after-value to compare with one, a figure due to be measured again, or a baseline still an estimate six months after it was set — each with the form that records it.
- Each initiative's **Setup** tab lists what that one initiative still needs.

When a measurement falls due, your **Inbox** carries a row for it with the form that records it. See [What an initiative needs before it has a number](https://roiva.ai/docs/before-an-initiative-has-a-number).