As customer service leaders, we know all about the Jevons Paradox
...and why AI very likely doesn't spell the end of contact centres for now
A week ago, very few people would have been able to quote the Jevons Paradox. But since Monday, when Microsoft CEO Satya Nadella mentioned it in a post on LinkedIn, it's become a hot topic of discussion for those working in AI.
Nadella was responding to the launch of DeepSeek R1, a Chinese engineered foundational model that beats the established models at many tasks, all at a fraction of the compute costs.
The Jevons Paradox is apt here. It states that as technological progress makes a resource more efficient to use, in turn the use of that resource increases. In short, actions intended to reduce demand for a resource may inadvertently end up increasing it. Nadella’s point is that low cost AI will lead to an explosion in AI usage.
As with all economic models, it is a simplification of what actually happens in the real world, and debate persists as to whether the Jevons Paradox can be applied in different situations. But, in the world of customer service, leaders will understand this phenomenon implicitly.
The below chart illustrates this nicely.
The green line shows the number of Customer Service Representatives working in the United States (source: US Bureau of Labor Statistics), and the orange line shows global spend on customer service software applications (source: Gartner), both indexed to 2018.
What this suggests is that whilst there has been a 60% increase in dollars spent on customer service tech, much of which intended to improve efficiency, the actual number of people working in frontline customer service has barely changed.
What I think has happened here is the combination of three phenomena. The first two are direct analogies of the Jevons Paradox, and the last follows a similar line of reasoning.
1. As customer service channels increase and become easier to access, contact volumes increase
This is the effect that digitization has on customer behaviour. People check their bank balances more than ever, because it is so easy to do so on an app. It’s so much easier to send a WhatsApp message than to pick up the phone and call customer service, so people do it more often. As we make customer service channels easier to access, customers make more use of them.
2. As contact centres become more efficient, companies find new things to do with them
There is also an internal version of the Jevons Paradox. As the cost to serve a customer goes down, CSRs are deployed on a wider range of tasks than just service: sales, retentions, debt collection, order fulfilment etc. Because CSRs form a flexible resource that is relatively easy to train, companies have an uncanny knack of finding new things for them to do, often supporting the growth of a business.
3. As customers' digital lives become more complex, they have more reason to contact
As AI proliferates, so too with the day to day problems that need to be resolved. Imagine a world that is full of autonomous AI agents, all interactive with each other to achieve independent goals. It doesn’t take much of a stretch of the imagination to think of all the things that could possibly go wrong in such a complex system. Customer service capabilities will be needed to deal with all of those. This is what we saw with the rise of digital self-service. Many banks still see 15-30% of customer service contacts relate to help with logging in and using the app.
None of this means that AI has no value in customer service. It's just that, based on historical experience, that value is unlikely to be realised as a pure cost saving. Instead, the tech can enable companies to achieve greater value through scalable growth and better quality outcomes from their customer service teams. And that is a really good thing.
Latest perspectives from BCG
Retail spotlight: personalization in action
The top retailers on the BCG Personalization Index can achieve an estimated $570 billion in incremental growth by harnessing first-party data to make customer interactions faster, easier, and more convenient
Several core actions can help retailers to realize this full potential—and go from good to great
I agree - if customer service is "worth it" - more people will use it. It still takes someone on the back end to check the work of the AI. In my company, we offshored our customer support for a while - but found that the combination of non native English speaking customers and support agents from different countries - led to a lot of misunderstandings. Even though things have improved on the AI front - you *still* need someone checking to make sure that there wasn't miscommunication.