Parloa study finds big firms make customer contact hard
Parloa has published a global study that found many large companies are hard for customers to contact. The research reviewed more than 10,000 enterprise websites and nearly 4,000 support interactions.
Among the websites analysed, 43.3% showed neither a visible phone number nor a chat option for customer support. Of those that offered a support channel, phone was available on 54.1% and chat on 28.9%.
More than half of the websites required users to scroll to find support information. Almost 95% of chat widgets appeared only on the homepage rather than on support pages, where users with active issues would be more likely to look for help.
The research covered Forbes Global 2000 company websites and included more than 800 businesses across 27 industries. Parloa used AI discovery agents to examine online support access, then carried out secret shopper tests across web, chat and voice channels.
Chat Limits
In chat, the study found a low rate of successful problem-solving. Fewer than 1 in 11 chatbot sessions ended with the customer's stated goal resolved.
When an automated chat interaction failed, transfer to a human agent worked only 10% of the time. The report also suggested that many companies still rely on older chatbot systems rather than newer large language model-based tools.
Less than 8% of classifiable chatbots in the research showed large language model features, while 92.5% were mainly rule-based. It also found that 4 out of 5 chatbots did not proactively tell users they were AI-powered.
Latane Conant, chief marketing officer at Parloa, said the findings reflected a broader approach to customer service among large organisations.
"We developed this study with the hypothesis that companies think of customer experience, particularly the service aspect, as a cost center, and not as a chance to deepen customer relationships. The data show this to be the case, given a clear contact deflection strategy starting at the front door. Savvy companies that shift their CX execution to focus on and enrich the actual customer experience, including service journeys, will gain the competitive advantage and rightfully earn stronger customer lifetime value," Conant said.
Voice Bottlenecks
The study painted a similar picture for phone support. Researchers mapped more than 100 enterprise IVR systems and found that many were hard to navigate, especially for the AI voice agents used in the testing.
Reaching a person usually took three to four menu levels. In some cases, hold times exceeded 90 minutes, while many systems offered neither a callback option nor visibility into queue times.
The AI voice agents were often blocked by touch-tone menus, strict authentication requirements and call flows designed only for human callers. That led the report to examine whether large companies are prepared for a model in which a consumer's AI assistant deals directly with a company's service agent.
Agent Readiness
On that measure, readiness was limited. Only 1% of tested enterprises appeared ready for agent-to-agent interaction.
The report argued that this matters because consumers are increasingly using AI tools to handle routine tasks such as returns, delivery tracking, reservations and bill management. In many cases, existing business systems treat AI callers as a threat rather than as a new form of customer interaction.
Legacy technology was identified as a central obstacle. Natural language requests from software-based callers often could not be processed, while authentication systems built for human users failed when the caller was an AI system.
Conant summed up that constraint directly.
"The limited readiness is not a reflection of insufficient ambition. It is a reflection of architectural constraints," she said.
Parloa said the study is intended to serve as a benchmark for enterprise customer experience across web, chat and voice. The company employs 430 people across offices in New York, Berlin, London and Munich.