> ## Documentation Index
> Fetch the complete documentation index at: https://www.voiceflow.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Measure

> Transcripts, evaluations, tests, analytics and credits: the words for reading back what an agent actually did, and what it cost to do.

The words for reading back what an agent actually did, and what it cost to do, one section per term.

## Conversations

### Transcript

The record of one conversation the agent actually had, with every turn and trace. The transcripts
table shows each at a glance; filters narrow it by date, platform, user ID, environment and more.

**Also asked as:**
- read back a conversation the bot had
- find a chat by the customer's ID
- the history of calls and chats

**Related:** [Transcript property](#transcript-property) · [Evaluation](#evaluation) · [PII redaction](/docs/resources/glossary/knowledge-and-data#pii-redaction) · [Call recording](/docs/resources/glossary/deploy#call-recording)

**Documented in:** [Transcripts › Understanding the transcripts table](/docs/documentation/measure/transcripts#understanding-the-transcripts-table) · [Transcripts › Filtering transcripts](/docs/documentation/measure/transcripts#filtering-transcripts) · [Transcripts › Reviewing individual transcripts](/docs/documentation/measure/transcripts#reviewing-individual-transcripts) · [Observe (API)](/docs/api-reference/sections/observe)

### Transcript property

A label you define once and then set on individual conversations, so your own outcome data can be
filtered alongside Voiceflow’s. Unlike variables, properties aren't used during the conversation.

**Also asked as:**
- tag a conversation with our own outcome
- attach custom metadata from a workflow
- filter transcripts by a label we set

**Related:** [Transcript](#transcript) · [Variable](/docs/resources/glossary/knowledge-and-data#variable)

**Documented in:** [Transcripts › Attaching custom properties](/docs/documentation/measure/transcripts#attaching-custom-properties)

### Transcript settings

Preferences for the transcripts view: whether test conversations are saved to transcripts, and
which columns the table shows.

**Also asked as:**
- hide test chats from the transcript list
- add an evaluation column to the table
- exclude test conversations from transcripts
- which columns show in the transcripts table

**Related:** [Transcript](#transcript) · [Test](#test)

**Documented in:** [Transcripts › Configuring transcript settings](/docs/documentation/measure/transcripts#configuring-transcript-settings) · [Transcripts › Customizing the table view](/docs/documentation/measure/transcripts#customizing-the-table-view)

## Evaluations

### Evaluation

A criterion a model scores past conversations against, turning quality into a number you can track.
Each has a name, a model, a metric type and instructions; results feed the Analytics dashboard.

**Also asked as:**
- score conversations automatically
- rate how well the assistant handled each chat
- a rubric the AI applies to transcripts

**Related:** [Default evaluations](#default-evaluations) · [Resolution](#resolution) · [Batch evaluation](#batch-evaluation) · [Analytics](#analytics)

**Documented in:** [Evaluations › Understanding evaluations](/docs/documentation/measure/evaluations#understanding-evaluations) · [Evaluations › Creating an evaluation](/docs/documentation/measure/evaluations#creating-an-evaluation) · [Evaluations › Viewing evaluation results](/docs/documentation/measure/evaluations#viewing-evaluation-results) · [Insights (API)](/docs/api-reference/sections/insights)

### Default evaluations

The three evaluations included out of the box: Customer satisfaction (a 1–5 rating from tone and
content), Deflection rate (was the issue resolved without a human) and Resolution rate (the outcome
category). Edit them or create a custom evaluation for different criteria.

**Also asked as:**
- the scores Voiceflow gives every conversation
- built-in quality metrics
- customer satisfaction, deflection and resolution scores
- the metrics that come pre-configured

**Related:** [Evaluation](#evaluation) · [Resolution](#resolution)

**Documented in:** [Evaluations › Default evaluations](/docs/documentation/measure/evaluations#default-evaluations)

### Resolution

The default evaluation that sorts every conversation into one of five outcomes; these feed the
Resolution chart on the Analytics dashboard.

**Also asked as:**
- was the customer's problem actually solved
- resolved versus escalated conversations
- outcome categories on the dashboard

**Related:** [Evaluation](#evaluation) · [Analytics](#analytics)

**Documented in:** [Evaluations › Resolution categories](/docs/documentation/measure/evaluations#resolution-categories) · [Analytics › Drilling into evaluation results](/docs/documentation/measure/analytics#drilling-into-evaluation-results)

### Batch evaluation

Running an evaluation over transcripts that already exist. Evaluations run automatically only on new
transcripts, so scoring history means selecting transcripts and running a batch.

**Also asked as:**
- apply a new scoring rule to older transcripts
- apply an evaluation retroactively
- backfill scores on old chats
- grade older conversations with a new evaluation
- run an evaluation on last month's transcripts

**Related:** [Evaluation](#evaluation) · [Transcript](#transcript)

**Documented in:** [Evaluations › Running evaluations on past transcripts](/docs/documentation/measure/evaluations#running-evaluations-on-past-transcripts) · [Transcripts › Running batch evaluations](/docs/documentation/measure/transcripts#running-batch-evaluations)

### Evaluation costs

Each evaluation consumes a small amount of credits because a model analyses the transcript; the
exact cost depends on the model chosen.

**Also asked as:**
- do evaluations use up credits
- how much does scoring cost
- how many credits an evaluation uses
- does automatic scoring cost anything

**Related:** [Evaluation](#evaluation) · [Credits](#credits)

**Documented in:** [Evaluations › Evaluation costs](/docs/documentation/measure/evaluations#evaluation-costs)

## Tests

### Test

A scripted conversation replayed against the agent, with checks asserting what it does. Tests never
run automatically: you run one, or a batch, yourself.

**Also asked as:**
- a simulated conversation to check the bot's behaviour
- regression tests before shipping a change
- make sure the assistant still escalates when it should

**Related:** [Test turn](#test-turn) · [Test check](#test-check) · [Test run](#test-run) · [Persona](/docs/resources/glossary/knowledge-and-data#persona)

**Documented in:** [Tests › Key concepts](/docs/documentation/measure/tests#key-concepts) · [Tests › Creating a test](/docs/documentation/measure/tests#creating-a-test) · [Tests › Running tests](/docs/documentation/measure/tests#running-tests) · [QA (API)](/docs/api-reference/sections/qa)

### Test turn

One side of a scripted conversation: either what the user says, or the agent reply the checks assert
against.

**Also asked as:**
- one message in a test script
- a user message or agent reply in a test
- the steps of a scripted test conversation

**Related:** [Test](#test) · [Test check](#test-check)

**Documented in:** [Tests › Creating a test](/docs/documentation/measure/tests#creating-a-test)

### Test check

One assertion about an agent turn. Compares the reply itself, or the behaviour behind it such as which
tool was called.

**Also asked as:**
- verify the bot called the right tool
- assert what the reply must contain
- assert that the agent escalated
- verify the reply contains a phrase

**Related:** [Test](#test) · [Test turn](#test-turn)

**Documented in:** [Tests › Adding checks to an agent turn](/docs/documentation/measure/tests#adding-checks-to-an-agent-turn)

### Test run

One execution of a test, carrying how many of its checks passed. Runs are asynchronous, so poll until
the status settles. Past runs are listed per environment.

**Also asked as:**
- see which tests failed last time
- run every test at once
- results of the last test batch
- which checks failed in my test

**Related:** [Test](#test) · [Credits](#credits)

**Documented in:** [Tests › Running a single test](/docs/documentation/measure/tests#running-a-single-test) · [Tests › Running multiple tests](/docs/documentation/measure/tests#running-multiple-tests) · [Tests › Reviewing past runs](/docs/documentation/measure/tests#reviewing-past-runs)

## Analytics and cost

### Analytics

Aggregate usage and cost: tokens, calls, conversations, unique users. The dashboard is filtered by
environment and date range, compares each metric with the preceding period, and breaks conversations
down by evaluation outcome.

**Also asked as:**
- how many people talked to the bot this week
- usage numbers per day
- a dashboard of conversation volume
- daily count of visitors who chatted with the assistant
- unique users per day
- conversation counts over time

**Related:** [Evaluation](#evaluation) · [Resolution](#resolution) · [Business impact](#business-impact) · [Credits](#credits)

**Documented in:** [Analytics › Filtering your view](/docs/documentation/measure/analytics#filtering-your-view) · [Analytics › Customizing the dashboard](/docs/documentation/measure/analytics#customizing-the-dashboard) · [Analytics › Trends](/docs/documentation/measure/analytics#trends) · [Analytics (API)](/docs/api-reference/sections/analytics)

### Business impact

The Time saved and Dollars saved widgets, which estimate the value the agent provides from figures you
enter.

**Also asked as:**
- put a dollar figure on what the assistant saves us
- report ROI from the dashboard
- estimate hours and dollars saved by the agent
- show ROI on the analytics dashboard

**Related:** [Analytics](#analytics)

**Documented in:** [Analytics › Estimating business impact](/docs/documentation/measure/analytics#estimating-business-impact)

### Credits

The unit Voiceflow bills usage in. Agents consume credits for AI responses, phone calls, evaluations
and tests; paid plans include a monthly allotment, and a bigger bundle or auto top-ups add more.

**Also asked as:**
- what does each message cost us
- does a bigger model burn more credits
- we ran out of credits mid-month

**Related:** [Plan](/docs/resources/glossary/core-objects#plan) · [Auto top-up](/docs/resources/glossary/core-objects#auto-top-up) · [Faster processing](/docs/resources/glossary/build#faster-processing)

**Documented in:** [Credits pricing table](/docs/documentation/account-management/billing/credits-pricing-table) · [Billing overview › Credits](/docs/documentation/account-management/billing#credits) · [Tests › Test costs](/docs/documentation/measure/tests#test-costs)