Outcome, with confidence
A traffic-light outcome status (green, yellow, red) defined per agent type, plus a confidence score for the label itself. A clear "not interested" is red with high confidence; a half-finished call is yellow.
此页面目前仅提供英文版本。 查看英文页面
Call analytics
AI call analytics that turn every conversation into an actionable record
Not a transcript dump. Each conversation becomes a structured record: what the caller wanted, what happened, how certain the outcome is and what should happen next. The same record feeds the CRM, dashboards and reports, so support, sales and operations teams work from one outcome standard.
After the form, Dring AI calls in about two minutes, identifies itself and qualifies the workflow. See the callback flow or review the operations benchmark.
The end-of-call record
One schema for every agent, every channel.
Voice, WhatsApp or chat, inbound or outbound, the record has the same shape. That is what makes a support line and a sales campaign comparable on one dashboard.
Reason tags
One to three atomic tags explaining the outcome: pricing question, wrong number, callback requested, decision next week. Fixed vocabulary where it matters, free tags where it does not, each with its own confidence.
Conversion likelihood
For sales, qualification and appointment agents: a 0 to 1 likelihood in tenths, banded by outcome, so sales gets a ranked list and not a pile of transcripts. Support and survey agents report engagement quality instead.
Sentiment and CSAT
The customer's tone at the end of the call, positive, neutral or negative, scored on every conversation. A CSAT score is recorded only when the customer was actually asked and gave a number, never inferred.
Next action
What should happen now: the agent calls or messages again, your team calls or messages, nothing, or do-not-call. With channel and, when the customer named a time, a scheduled slot.
Summary, warnings, transfer
A short summary in the call's language, a fixed set of warning flags for things a human should look at, and whether the call was transferred and why.
outcome_statusgreen · confidence high
conversion_signalqualified for next step
reason_tagspricing_question · decision_next_week
sentimentpositive
summaryReached the operations manager. Needs weekly pallet shipments to Germany. Wants a rate sheet. Decision after the board meeting next week.
next_actioncompany call · phone · scheduled
warningsnone
transferredfalse
micro_insightCustomer opened up after the agent named the shipping lane first; keep the lane question early.
Outcome standard
Green means the same thing for the same kind of agent.
Outcome status is defined by the agent's purpose, not by mood. That is why a support line at one customer and a support line at another can be compared, and why a sales agent is never judged like a survey agent.
| Agent type | Purpose | Green | Yellow | Red |
|---|---|---|---|---|
| Qualification | Collect information and interest | Planned information collected, weak or strong lead alike | Half the information, "later", undecided | Clear "not interested" or "do not call" |
| Appointment | Get a booking or an informed acceptance | Booked, or invitation accepted with understanding | Undecided, wrong person, mechanical yes | "I do not want an appointment" |
| Support | Resolve the issue by AI | Resolved, customer closed the call | Not resolved, ticket or transfer | Rare: explicit "never call about this again" |
| Survey | Collect answers | Core question block completed | Availability problem, block incomplete | Refusal to take part |
| HR interview or reference | Collect candidate or referee information | Required items completed, regardless of verdict | Required field missing, consent not obtained | Declined to take part |
| Notification | Deliver a message to the right person | Delivered and acknowledged | Not reached, wrong person, voicemail | "Stop these notifications" |
| Booking or confirmation | Complete or confirm a transaction | Completed and confirmed | Information missing, technical failure | Customer declined |
| Collection or cancellation | Complete the requested action | Completed | Hesitation, terms unclear | Refused |
Dashboards
Per agent, per line, live.
Volumes, answer rates, outcome distribution, sentiment, transfer rate, top reason tags and delivery signals, per agent and per line, updated as calls end. Filters for campaign, period, language and channel. Drift alerts when an agent's scores leave their expected range.
- Outcome funnel: reached, conversation, green share
- Reason-tag ranking: why calls end the way they do
- Transfer rate and transfer reasons
- Median and p95 response latency
- Unique customers and contact frequency, so nobody is over-called
Conversations in scopeCurrent view
Resolution across current production workloadsApproximately 72%
Agent conversation minutesMore than 10,000 per day
Outcome signalsTracked per workflow
Record deliveryPer call
Top reason tagsorder_status · return_request · payment_question
Monthly report
A report that says what to do, not only what happened.
Every customer receives a monthly performance report in their own brand colours. Two sections, the same order every month, so trends are visible at a glance.
The month in four numbers
Total interactions, reached or answered, green share, sentiment. Plus the one sentence that summarises the month.
Where the calls came from
Volumes by day and hour, answer or reach rates, unique customers and contact frequency, channel split, after-hours share.
What happened and what next
Outcome distribution against the agent's standard, reason-tag ranking, transfer reasons, sentiment trend, and insights written as applicable recommendations with the expected effect.
Insights follow a rule, not a hunch
Each recommendation names the observation, the metric it is based on, the action and the expected effect. Insights that cannot point to a number in the report do not make it into the report.
Same metric dictionary everywhere
The report, the dashboard and the CRM fields use one metric dictionary and one outcome standard. A number in the report can be traced to the calls behind it.
Where the data goes
Into your systems, in seconds.
The end-of-call record is delivered by webhook or email, written to your CRM through Dring Connect, and kept on the Dring dashboards. Your fields can be added per agent as extra analysis items. Across current production workloads, Dring handles approximately 300,000 customer conversations per month.
- Hang-up
Analysis runs
The conversation is analysed against the agent's outcome standard and your extra fields.
- Seconds later
Record delivered
Webhook, email or CRM write-back, with outcome conditions so green, yellow and red can go to different places.
- Continuously
Dashboards update
Per agent and per line, with drift alerts to Dring first and to you when action is needed.
- Monthly
Report published
In your brand colours, same structure every month, with recommendations and the actions we took.
FAQ
Analytics questions.
How is automatic analysis checked?+
Outcome labels are reviewed against human QA criteria. Disagreements feed the calibration set, and the test pipeline checks for drift before a controlled release.
Can we add our own fields?+
Yes. Extra analysis items are defined per agent with a name and a short instruction, and appear in the record, the webhook and the dashboards.
Do you infer CSAT from tone?+
No. A CSAT value is recorded only when the customer was explicitly asked and gave a number. Sentiment is reported separately and never converted into a score.
Can we export everything?+
Yes: webhooks per call, CSV exports from the dashboards, and the monthly report as PDF. Recordings and transcripts follow the retention set in your contract.
See a real monthly report.
Submit context and consent for a callback. Dring AI calls in about two minutes, identifies itself and qualifies the reporting workflow before we walk through an anonymised report.