ProductUse casesPricingBlogContact
Dashboard Sign in Start free
Artificial Intelligence in Business

Voicebot KPIs: What to Measure in the First Month

Voicebot KPIs: What to Measure in the First Month

The voicebot KPIs worth tracking in the first month are resolution rate, handover rate, repeat contact rate and abandoned conversations, read together rather than one by one. Each number alone can look good while customers are unhappy. This guide explains what each metric means, how to count it from conversation logs and which combinations tell you something is wrong.

The first weeks after launch are when a bot changes fastest. You add missing answers, fix wording that sounds odd when spoken and adjust the scope. Without a few clear measures, those changes are guesses. With too many, nobody looks at them. Four voicebot KPIs, reviewed weekly, are enough to start.

Before you measure: set a baseline and a definition

Write down how you count each metric before the first conversation, and keep the definitions fixed for the whole month. If you change what “resolved” means in week three, the trend line means nothing.

Also note what you have now without the bot: how many contacts your team handles per week, which topics dominate and how often customers come back about the same issue. You do not need precise numbers; a rough picture from your inbox and phone log is enough to compare against.

Finally, check what your platform lets you export. Any vendor should give you conversation transcripts or logs with timestamps; ask what is included before you sign, because the metrics below are counted from them.

Voicebot KPIs: the four core metrics

1. Resolution rate

Resolution rate is the share of conversations in which the customer got the answer they needed without a person. It is close to what chatbot teams call containment, and the guide on how to measure containment rate covers the counting rules in detail.

For a voicebot, be strict about what counts as resolved. A conversation that ended because the caller gave up is not resolved, even though no human was involved. A practical rule: count a conversation as resolved when the bot gave an answer from the knowledge base, the customer did not ask for a person and did not come back about the same topic within a few days.

2. Handover rate

Handover rate is the share of conversations passed to a person, by request or by the bot’s own rules. A high handover rate in the first month is not a failure in itself. It often means the bot is correctly refusing questions outside its scope.

What matters is why the handovers happen. Tag each one with a reason: the customer asked for a human, the topic was out of scope, the bot did not understand, or the question needed account access. The quality of the handover matters too; the article on handing over without losing context explains what the agent should receive.

3. Repeat contact rate

Repeat contact rate is the share of customers who come back about the same issue within a short window, through the bot or another channel. It is the best check on the resolution rate, because it catches answers that sounded complete but did not solve anything.

Counting it takes some work. You need a way to connect contacts from the same person: an order number, an email address given in the conversation or a ticket reference. Where you cannot link them, sample: pick a set of conversations marked as resolved and check by hand whether those customers wrote or called again.

4. Abandoned conversations

An abandoned conversation is one the customer left before getting an answer or a handover. In voice, this often means silence, a long pause before the reply or repeated misunderstandings. Listen to a sample of abandoned conversations every week; the pattern is usually visible within a handful.

Common causes are slow responses (see where voicebot silence comes from), answers that are too long to follow by ear and speech recognition errors on product names or addresses.

How to read the metrics together

One metric on its own can mislead. The combinations below are the ones worth checking every week.

What you see Likely meaning What to do
High resolution, high repeat contact Answers sound complete but do not solve the problem Read the resolved conversations of returning customers and rewrite those answers
Low resolution, high handover Scope too narrow or sources missing Group handover reasons, add answers for the top topics
Low handover, high abandonment Customers give up instead of asking for a person Make the route to a human easier to find and say it earlier
Abandonment rising after a change New answers are too long or slower Compare transcripts before and after, shorten spoken answers

Two supporting signals

Besides the four core voicebot KPIs, keep a simple list of two things.

  • Unanswered questions: every time the bot says it does not know, log the question. This list is your content backlog. The piece on honest “I do not know” answers explains why these refusals are useful.
  • Misheard words: product names, street names and numbers the bot gets wrong. Fixing them often means adding spelling variants to the sources or rephrasing the question the bot asks.

A weekly review routine for the first month

  1. Week one: read as many transcripts as you can. Do not tune metrics yet; fix obvious wrong answers.
  2. Week two: start counting the four metrics with the fixed definitions. Tag handover reasons.
  3. Week three: act on the biggest gap: the most common handover reason or the most common cause of abandonment.
  4. Week four: compare with week two and with your pre-bot baseline. Decide whether to widen the scope or keep fixing the current topics.

Keep the review short: one person, one hour, the same day each week. The goal is a list of three changes for next week, not a report.

If you run a voice assistant on your website, such as Botino’s, it answers from your own knowledge base, so most improvements are edits to that content: a clearer FAQ entry, a missing policy page or a shorter spoken answer.

FAQ

What is a good resolution rate for a voicebot?

There is no single benchmark that fits every business, because it depends on the scope and the questions customers bring. Compare your own trend week by week and check it against repeat contacts.

Should customer satisfaction surveys be part of the KPIs?

A short question at the end of the conversation can help, but few customers answer it and those who do are often the very happy or very unhappy ones. Treat it as a supporting signal, not the main measure.

How long should I measure before changing the bot’s scope?

Give each change at least a week of real traffic before judging it, and change one thing at a time so you know what caused the effect.

Pick the four voicebot KPIs, write their definitions down today and review them weekly. After the first month you will know which answers work, which need rewriting and whether the bot is ready for a wider scope.