Key takeaways
Because they are the easiest calls to avoid and the most expensive to let through. A repeat call is either a simple request that comes back every day in the same form, or a customer who calls back because their issue wasn't resolved the first time.
The two add up. The first group covers recurring contact reasons: where is my case, when is the technician coming, why is this amount on my bill. Each call is legitimate, but the answer is always the same and an advisor adds nothing to it. The second group, callbacks, comes from incomplete handling: a transfer without context, a partial answer, a promised callback that never happens.
The cost is well documented. According to SQM Group, repeat calls represent 23% of the average contact center's operating budget, and every point of first call resolution gained cuts operating costs by about 1%. These figures date from 2020, but the order of magnitude still speaks for itself: reducing repeat calls has a direct effect on support costs.
It listens to the request in the caller's own words, with no keypad menu, and matches it to a known contact reason. Understanding customer intent is about what the caller wants to achieve, not the words they use.
Take a customer who says "I got a weird bill." They may want to dispute an amount, understand an adjustment or ask for a payment plan. A traditional IVR made them choose between two keys, often the wrong one. The voicebot asks a clarifying question if needed, then routes the call.
This understanding relies on a library of contact reasons. DialOnce's AI Voice Agent draws on a bank of intents built over ten years of customer relations projects with more than 150 large organizations. In social housing, for example, the agent covers more than 550 intents, from moving in to ending the lease. The same library powers the digital agent and the mailbot: a customer who starts on the website and then calls is understood in the same way. To see how these tools compare, read our article on the difference between a mailbot, a voicebot and a chatbot.
All those whose answer already exists in your knowledge bases or business tools: regulatory information, case tracking, appointment booking, opening a request. The customer gets an answer during the call, or hangs up with an action already under way.
Phone automation is about more than reading out an FAQ. Connected to the CRM or business tools, the voicebot checks the status of a case, logs a request, schedules a callback or sends an SMS or RCS message during the call, for example with a link to the document the customer was expecting.
We showed this live at the Congrès Hlm 2026, the annual gathering of the French social housing sector. A tenant calls because they don't understand the service charge adjustment shown in their tenant portal. The agent explains that the adjustment reconciles advance payments with actual costs each year, answers the follow-up question about those costs, then offers to guide the tenant through the portal to find the details. No advisor steps in, and the tenant has no reason to call back the next day.
These requests are handled 24/7. A customer who calls on a Sunday evening gets an answer right away, instead of calling back on Monday morning when the queues are already full.
The voicebot passes the advisor the identified reason for the call, the information gathered, the actions already taken and the history of the conversation. The advisor picks up a qualified request, and the customer doesn't have to explain everything again.
This is often where callbacks start. A transferred customer who has to give their contract number again, explain the problem again and then wait a second time loses patience. If they hang up before the end, they will call back. A contextual handover removes this break.
In the Congrès Hlm demo, the same call about a water leak takes a different route. The agent recognizes the emergency and hands over to an advisor without keeping the tenant waiting, along with the context already gathered. The advisor knows what is going on before even saying hello.
The handover can also switch channels. An administrative request moves to a digital journey and is handled later, while an emergency goes straight to a person. In a previous article, we explained how to reduce contact volumes without compromising the customer experience by routing each request to the right channel.
No. The voicebot sits upstream of the contact center and integrates with your existing CCaaS and CRM. It doesn't replace the advisors' tool: it sends it requests that are already qualified.
Integration with contact center platforms determines the quality of the handover. Without a connector, context is lost the moment the call reaches the advisor. DialOnce integrates with the main platforms on the market, including Kiamo, an official partner, as well as Genesys, NICE inContact, Odigo, Diabolocom and Salesforce. The solution is developed and hosted in France.
To scope a project, start from your contact reasons and the tools you already use.
Lower support costs come from two effects: fewer calls reach advisors, and the ones that do are handled faster. Across all DialOnce customers, the agent placed upstream of the contact center avoids up to 25% of contacts and improves reachability by up to 35 points.
The gain isn't only about volume. An advisor who receives a qualified request doesn't have to piece the situation together, which reduces average handling time. They spend their time on cases that call for judgment: a dispute, a customer in difficulty, a sensitive file.
Vattenfall offers an example on digital channels. The energy supplier's omnichannel AI agent handles requests autonomously, 24/7, from signing up to managing the contract. The result: 35% savings on cost to serve, an NPS that rose from 5 in 2021 to 40, and 90% customer satisfaction. The AI Voice Agent applies the same logic to the phone, with the same intent library.
Read the Vattenfall customer story to see how the energy supplier reduced its cost to serve.
A traditional IVR makes callers choose from a keypad menu. A voicebot understands the request in the caller's own words, then resolves it or hands it over with its context. A visual IVR can complement this setup, as we explained in our article on visual IVR and voicebots as complementary solutions.
No. It handles the repetitive requests it can resolve on its own. Advisors keep their time for situations that call for listening or expertise.
Track a few indicators before and after deployment, on the same contact reasons: the self-service resolution rate, the transfer rate to advisors, the average handling time of transferred calls and the callback rate, meaning customers who call back about the same request. Then compare the calls avoided and the time saved with the cost of a call handled by an advisor.
A voicebot answers several calls at once. Its capacity depends on the number of phone lines routed to it, not on staffing. It can absorb peaks in activity, such as the start of the month or an incident, without lengthening the wait for customers who need to speak to an advisor.
Yes. DialOnce's AI Voice Agent handles every language, which helps call centers whose customers don't all speak French.
In France. DialOnce is a French software company, and its solution is developed and hosted in France.