When AI Doesn’t Solve the Problem
Why Quality Customer Support Still Needs People
Artificial intelligence helps companies respond faster and handle more customer inquiries. But a fast response doesn’t necessarily mean the customer’s issue has been resolved. We explore where automation truly adds value, how to measure its effectiveness, and why human involvement remains essential.
Artificial intelligence is increasingly seen as a solution to a wide range of business challenges. Customer communication was one of the first areas to implement it. Chatbots can quickly handle routine inquiries, help customers find information, and reduce the workload of support agents. However, even the fastest system provides little value if the customer never gets the solution they need.

In an era when more and more processes can be automated, a company’s goal should not simply be to increase the percentage of conversations handled without human involvement. What matters is ensuring that customers receive accurate answers, ideally during their first interaction and with minimal effort.
A Fast Response Doesn’t Always Mean a Resolved Issue
Chatbots are effective when a request is straightforward and follows a clearly defined process. Problems arise when the system provides irrelevant answers, repeatedly asks customers to rephrase their questions, or makes it difficult to reach a human agent. A minor issue can quickly escalate into a complaint, and frustration can turn into a negative review.

The Genesys State of Customer Experience 2026 report highlights the importance customers place on service quality, fast issue resolution, and continuity across interactions. These expectations should be taken into account when implementing automation.
  • 84%
    of consumers give a virtual agent no more than three attempts to resolve their issue
  • 95%
    expect companies to preserve the context of their interactions when switching between channels
  • 85%
    have reduced their spending with a brand or stopped doing business with it because of poor customer service
Source: Genesys, State of Customer Experience, 2026. These figures reflect survey findings and do not represent the performance of any specific Viatel project.

In non-standard situations, such as complaints, disputes, or cases requiring an individual solution, it is particularly important to involve a human agent at the right moment.

However, according to Genesys, 48% of organizations do not automatically transfer previously collected information between virtual and human agents. As a result, customers have to explain their issues again, increasing handling time and reducing satisfaction.

Effective automation must therefore provide more than fast responses. It should also enable a seamless handoff to a human agent, preserving the necessary context while ensuring compliance with personal data protection requirements.
The goal is not to choose between technology and people, but to use automation where it genuinely improves the customer experience.
What Companies Lose When Automation Fails
An unresolved issue rarely disappears on its own. Customers contact support again, the team’s workload increases, and the expected cost savings from automation may never become real. Inappropriate or irrelevant responses undermine customer satisfaction and trust, ultimately affecting repeat purchases and long-term loyalty.

According to the Genesys study, 85% of consumers reported that poor service had caused them to spend less with a brand or stop doing business with it altogether. Another 66% would consider switching to a competitor after three or fewer negative interactions.

This is why evaluating automation solely by the number of conversations closed by a bot is not enough. Companies also need to assess whether the interaction resulted in an actual resolution and how much additional effort it required from both the customer and the support team.
How to Properly Measure the Impact of Automation
In an article published in April 2026, McKinsey & Company noted that 60% of respondents had not yet observed an impact from their AI programs on organization-wide EBIT (earnings before interest and taxes). This highlights the importance of distinguishing between the widespread use of technology and its proven business impact.

McKinsey suggests evaluating AI value across five dimensions: financial performance, strategic transformation, operational metrics, employee adoption, and technical effectiveness.

For Customer Support operations, it is useful to evaluate several performance indicators together. These can be aligned with industry quality management practices, including the COPC CX Standard, and adapted to the objectives of each individual project.
  • FCR · First Contact Resolution
    The percentage of customer issues resolved during the first interaction
  • CSAT · Customer Satisfaction
    A measure of how satisfied customers are with the support they receive
  • Repeat Contact Rate
    How frequently customers need to contact support again about the same issue
  • Transfer Success Rate
    The percentage of inquiries transferred to a human agent that result in a successful resolution
A high automation rate is not a success in itself if customers continue returning with the same unresolved issues. Performance metrics should be analyzed together rather than in isolation.
Our Approach to Automation at Viatel
At Viatel BPO, AI implementation starts with understanding business processes and operational needs, rather than selecting a technology.

Today, we use AI tools in selected Back Office processes, quality assurance, and employee training and assessment, specifically in scenarios where these tools do not process personal data.

When considering broader automation opportunities, we work with our clients through four structured stages:
  • Analyze the Process
    We identify operational challenges, team requirements, and activities where automation could deliver meaningful benefits.
  • Assess the Risks
    We evaluate the potential impact on service quality, information security, and the overall customer experience.
  • Run a Pilot
    We conduct a limited trial with clearly defined objectives, performance indicators, and monitoring requirements.
  • Make Data-Driven Decisions
    We determine what works, what needs improvement, and whether further scaling is justified.
This phased approach aligns with McKinsey’s recommendations: AI solutions should be scaled only when measurable results have been demonstrated against predefined performance indicators.

Not every process can or should be automated. Even a technology that performs successfully in one project may not be suitable for another.
Effective Automation Starts With the Right Questions
The future of customer support is not a competition between AI and human agents. It is about finding the right balance between technology and human expertise, taking into account business processes, customer expectations, service quality, and measurable outcomes.

Together with our partners, we identify which tasks are best handled by automated systems, where human involvement remains essential, and how to validate the benefits of a solution before implementing it on a larger scale.
Contact Viatel
Let's discuss your business needs
Tell us which business processes you'd like to improve or outsource. Our specialists will evaluate your requirements and recommend the most suitable operating model.
We treat every customer interaction as if it were our own, because delivering exceptional customer service and sales support is our business.
Viatel Romania
Home
Business Solutions
How We Work
About Us
Insights
Careers
For any questions regarding the processing of personal data, as well as for exercising your rights related to personal data protection, you may contact the Data Protection Officer of Viatel S.R.L.

You can contact the Data Protection Officer using any of the following methods:
by e-mail: date.personale@viatelbpo.com
at the Viatel S.R.L. office address: MD-2071, Republic of Moldova, Chișinău Municipality, 75 Alba Iulia Street, 7th floor