AI-Powered Insights: Who’s Asking for Help (or Not) and What to Do Next

In many of my previous articles, I have encouraged calculating and reducing the rate of customer-initiated contacts, rather than focusing solely on volumes. A good example is CPO, which stands for contacts per order shipped, rather than the number of contacts. Examining rates adjusts for changes in order, revenue, or usage, and, in general, this metric should decline. Over the years, we have found that about 45% of customer contacts represent mistakes or confusion, and another 35% should be automated.

Further, in my February 2025 CustomerThink article, “Making it Right”¹, I shared five practices to ensure your customers’ needs are met when they contact you.

However, there are several additional dimensions to improve customer experience (CX) and employee experience (EX), including how to handle:

1. Customers who do not bother to contact you.
2. Cultural, regional, and demographic differences.
3. First-time vs. experienced customers.

Taken together, I’m calling these three dimensions the propensity to complain, or PTC. These dimensions:

• Underpin my five recommended practices for “Making it Right” and my five practices for “Getting it Right” (in my most recent CustomerThink article²).
• Produce the much sought-after Win-Win-Win = lower total costs + greater revenues + more loyal customers.
• Help explain Hirschman’s three ground-breaking responses all customers have – exit, voice, or loyalty³.
• Improve agent or customer-facing experiences (EX) that are tied closely to CX.

Let’s review each dimension and see how AI and predictive analytics can support them.

1) How to handle customers who do not bother to contact you.

It is estimated that only 5 to 20% of customers contact organizations when they need help⁴. However, in our client projects, we have discovered you need to isolate complaint rates by intent, not an average. For especially irritating intents such as “My Internet keeps dropping” or “Where’s my order [I need today]?”, you will see a much higher contact rate⁵. Still, on average, most customers don’t even bother. Why is that? It could be they:

• don’t know whom to contact or where to lodge their complaint; or,
• haven’t had success in the past, so why bother?; or,
• maybe they find other ways to fix their problem, such as posting on social media.

In The Frictionless Organization, my co-author David Jaffe and I argued that some of these customers are “happy campers” who don’t need to contact the company, while the others are “silent sufferers”⁶. AI and predictive analytics can now identify the split between happy campers and silent sufferers, leading to specific action plans for each group and each individual, especially the silent sufferers.

You can trace this practice well before the age of AI. Telecommunications companies such as MCI in the 1990s routinely analyzed customer usage of add-on paid services. When customers contacted MCI, saying, “Why is my bill so high?” or “Why am I being charged for this service?”, MCI would either confirm the fees’ validity or acknowledge non-usage and refund the customer. Since most customers didn’t bother to contact MCI for these intents, MCI reached out to them by phone, email, or snail mail to tell non-complaining customers that these additional charges had been removed. This was an early form of downsizing to meet the customer needs, rather than waiting for customers to contact MCI to complain about toxic revenues⁷.

Now, let’s say that in your company, on average, 15% of your customers who have a problem contact you for help, with 85% divided into happy campers at 25% and silent sufferers at 60%. The silent sufferers can be identified using AI, LLMs (large language models), and predictive analytics. For example, if a coat shipped by an e-commerce company was mis-sized and the company heard about it from a small percentage of customers, they could easily reach out to all other similarly situated customers and apologize for the mis-sizing, offering a free replacement or a refund.

2) How to handle cultural, regional, and demographic differences.

Over the last 40 years, academic researchers have compared customer complaint behavior patterns from country to country in the same settings, such as hotel customers in China versus hotel customers in Brazil. Their findings typically illustrate variations explained by the five Hofstede cultural dimensions⁸, Davidow’s A-CRAFT model⁹, Schwartz’s dimensions of national culture¹⁰, and other theories.

For example, customers in Asia are less willing to contact companies to complain or to seek compensation or retribution, versus customers in the USA and certain European markets who contact the same company two to five times more often for the same issue. This doesn’t mean that the Asian customers are any less dissatisfied, so it is important to reach out to them to confirm their disquiet and then offer the appropriate resolution or compensation¹¹.

By the same token, there are regional differences within countries, such as customers in the Philadelphia to Boston corridor complaining significantly more than customers in the US Southwest or Northwest. Again, it’s important to recognize which regional customers are not complaining, but whose needs should be addressed and serviced as much as those who contacted the company¹².

In addition, there are significant demographic differences in customer contact rates. For example, older customers and Millennials tend to reach out more frequently than Gen Z and younger customers, who instead complain directly to their friends or on social media rather than the organization. There are similar differences by sex, income, and other factors. Once again, AI and predictive analytics can now determine those differences and provide the right response to noncomplaining customers.

3) How to handle first-time versus experienced customers.

Studies have also shown that first-time customers tend to contact companies three to five times more frequently than longer-term customers who have grown accustomed to processes and policies, have reset their expectations, and/or have found other paths for resolution¹³. Therefore, it is essential to identify first-time customers and provide them additional time and guidance, whether they contact the company or not, so they can quickly gain experience.

This is especially important when companies have marketing campaigns to bring in new customers. Without using this algorithm, customer service and support teams are swamped with new customers who are contacting them for help, and all customers suffer as a result. This is also why a lot of electronics providers work hard to provide a clear and easy out-of-the-box and onboarding experience so that the customers do not need to contact them for help to install or start using the products.

Four Questions to Get Started

AI, LLM, and other advanced analytics inform companies to determine which customers are similarly situated as those who contact them (the silent sufferers or happy campers) and respond accordingly before they leave. These tools also help explain the cultural, regional, and demographic differences in contact rates and their propensity to complain. And finally, these tools will help identify first-time customers and provide the appropriate handling advice and mechanisms for them.

Therefore, the next time you report contact rates or examine contact rates in your organization, ask yourself or your analysts the following four questions:

1. What is the rate of customer-initiated, agent-handled contacts, not just the volume?
2. What percentage of our customers who have encountered the same problem bother themselves to contact us, and, for those who do not, what percentage of them are happy campers versus silent sufferers?
3. What are the differences in response rates by culture, region, and other demographic characteristics, and how can we normalize that by reaching out to those who are not contacting us?
4. How can we identify first-time customers before they reach out to us and offer them special help so they can become experienced users of our services?

By following these four questions and using AI, LM, and predictive analytics, you will be able to strive for the elusive Win-Win-Win.

Please contact me to get the research papers and frameworks.

Notes

¹ “Making it Right in Customer Service: Five Practices to “Flip the Turtle”, February 13, 2025:
https://customerthink.com/making-it-right-in-customer-service-five-practices-to-flip-the-turtle/

² “Getting it Right: Five Practices to Become Frictionless, Save Money, and Delight Customers”, April 23, 2025: https://customerthink.com/getting-it-right-five-practices-to-become-frictionless-save-money-and-delight-customers/

³ Albert Hirschman’s short 1970 book, Exit, Voice, and Loyalty: Responses to Declines in Firms, Organizations and States, is well worth (re)reading with today’s CX challenges in mind:

https://pages.ucsd.edu/~bslantchev/courses/ps240/05%20Cooperation%20with%20States%20as%20Unitary%20Actors/Hirschman%20-%20Exit,%20voice,%20and%20loyalty%20%5BCh%201-5%5D.pdf

⁴ See John Goodman’s excellent TARP studies.

⁵ From Driva Solutions’ client projects for Contact Optimization using the Skyline program. As I have encouraged in earlier CustomerThink articles and my books, companies should create a 25-50 intent taxonomy, not thousands of reasons nor useless, high-level categories such as “Billing” or “Admin”. There should not be any “Other” or “General”, and each one needs to show customer churn; NPS, Customer Effort, or another loyalty metric; FCR; and NBA.

⁶ Bill Price & David Jaffe, The Frictionless Organization: Deliver Great Customer Experiences with Less Effort, Barrett-Koehler, 2022. 

⁷ From the author’s personal experience, when I led MCI Call Center Services from 1991 to early 1999 before I joined Amazon to lead its global customer service.

⁸ Geert Hofstede’s seminal studies was his analysis of 72,000 IBM employees around the world, from which he wrote “Cultures and Organizations: Software of the Mind”: https://tinyurl.com/wmy6jtec.

⁹ Moshe Davidow’s 2014 paper, “The A-CRAFT Model of Organizational Responses to Customer Complaints and Their Impact on Post-Complaint Customer Behavior”, is featured in the book he and Gautam Mahajan and I co-authored, Zero Complaints: The Path to Continuous Value Creation, Productivity Press, 2025: https://tinyurl.com/4xpj739b

¹⁰ Shalom Schwartz started advancing his three polar dimensions of national culture, updating Hofstede’s earlier employee studies: https://changingminds.org/explanations/culture/schwartz_culture.htm

¹¹ From the author’s experience establishing customer service operations at Amazon.co.uk and more recently with a global consumer products company, whose Asian countries repeated this same cultural reluctance to complain.

¹² From the author’s Driva Solutions clients, including telcos and B2B companies.

¹³ From the author’s Amazon years, when we eliminated handle time for new customers and two other types of customers. Later, Driva Solutions clients, including a leading TV set-top provider and other companies. See also Taghizadeh & Panahi’s 2013 study, “A comparative study of complaint behavior of loyal customers versus first customers”.