Blog: Adopting AI for new-found efficiency and enhanced customer experience

This blog examines key considerations when introducing bots into the contact centre. We also show how is using AI to transform the customer experience and drive business growth.

Not all bots are created equal

We tend to see three categories of bots. First on the market, basic rule-based bots pick up on keywords and provide a predefined IVR-like response. They’re good since they are transparent and it’s easy to see the underlying logic and thought-process. However, they’re rigid because they can’t actually understand language. So, you need to anticipate every word and scenario otherwise the bot will fail. This makes them harder to maintain and also easier to break. In summary, they reveal what’s going on, but can rack up coding costs and are limited in terms of natural language processing and AI capability.

At the other end of the spectrum are fully-fledged AI and machine-learning bots that attempt to understand, learn and generate language. Like a baby, they can take a while to train and be equally unpredictable and difficult to control. More importantly, it’s not possible to examine and understand the underlying reasoning for a response. A famous example is Microsoft’s Tay, which was released to twitter and quickly became corrupted, offensive and racist. It was a sharp lesson learnt. This type of bot is generally not suitable for business.

In the middle is the hybrid bot, which provides a more flexible approach when it comes to understanding customer intents and requests. Although they aren’t able to generate new responses, these bots can handle the same question asked thousands of different ways. For example, by explaining the quickest route from A to B with a consistent response that also includes travel options and associated costs.

This is an over simplified description of artificial intelligence, and there are many nuances regarding how bots learn and extract meaning and reason to interact with humans. This research has been going for the last 70 years and is still far from complete.

Automatic incentives for bot introductions

There are several commercially-sound reasons to adopt bot technology. The most popular is lowering cost. Using bots to deal with FAQs and simple, repetitive tasks is a great place to start. Not least because it reduces call handling overheads (still one of the biggest costs for contact centres) and releases agents for more complex or revenue-creating work.

The next is maximising agent efficiency. Essentially creating a bot that works side-by-side with your teams to feed them information, so when the bot needs to handover the agent knows why the customer is calling and can serve them faster. That improves average handling time (AHT) and first contact resolution (FCR). Automating post-call functions, such as data entry into multiple systems can be another huge time saver.

AI-powered bots can help increase revenue. For example, spotting trends and patterns of high-value customers prime for potential upsell opportunities. Or intercepting website shoppers and offering reassurance before they abandon shopping carts and disappear.

Finally, bots offer new ways to grow customer loyalty. For instance, cost-effectively extending customer service beyond normal hours. Or, using natural language processing allied with AI to track agent performance and flag training needs. Two sure-fire ways to improve customer experience and net promoter score.

Top tips

When planning your bot journey, there are many ways to stack the odds in your favour:

  1. Categorise demand to better understand reasons for contact.
  2. Identify where AI self-service offers the biggest value for the lowest effort (often simple, frequent interactions).
  3. Map use cases and customer journeys (areas like complaint handling are probably best left well alone).
  4. Define metrics (such as AHT, FCR or sales conversions) balanced with other success criteria (like security and compliance).
  5. Conduct rapid iterative testing to see which bot services do and don’t work.
  6. Stop when seeing diminishing returns (for example, due to excessive complexity).

Best practice in AI execution

Headquartered in Brno, is one of the world’s fastest-expanding online travel platforms. Its proprietary algorithm, Virtual Interlining, allows users to combine flights from more than 750 carriers, including many that don’t normally co-operate.

Consistently growing bookings and doubling headcount every year, the company embraced AI to create a more scalable and efficient business. Voice bots with machine-learning capabilities were added to its Genesys Cloud platform in August 2018. They started processing common requests, reading out travel information pulled from internal systems and dealing with basic tasks like adding passport details or extra luggage allowances.

Intent capture, caller verification, flight updates and refunds were easily handled by voice bots. Tellingly, call abandonment rates with AI dropped to 10% – 15% compared to a dismal 40% achieved by customers using rigid tree-like IVR structures.

Within six months agent-assisted contacts halved, while bookings revenue steadily increased. However, the real value came from releasing the most skilled agents to deal with stressful situations like customers caught up in delays or missing their flights. Factors really shaping the experience and the principal reason customers keep returning to

Today, AI allows customers to self-serve in up to 25% of cases, saving around eight FTE. The company continues to develop new use cases. Like deploying voice bots to assist in handling emergency scenarios or adverse weather conditions. And it’s integrating chatbots with chat and messaging, two channels preferred by younger customers. Asynchronous chatbot messaging is also on the horizon, so busy customers can get on with their day and communicate when they have a moment.

If you would like to arrange a demo of AI or simply get a conversation going, please call 0330 403 0000 or email [email protected]

More about the author
Axel Ericsson

Axel Ericsson is a senior business development manager at Foehn Ltd.– specialists in cloud contact centre technologies. Axel has close to 10 years’ experience working in the tech industry at both large technology companies including Intel and start-ups such as Rebtel - a VoIP Platform as a Service company. Axel is passionate about AI and deep learning and have MSc degrees in both business and engineering together with hands-on experience implementing machine learning algorithms at Broad Institute of MIT and Harvard, the largest genomic institute in the world. Always keen to help companies making the most out of their technology investments by implement high impact digital transformation projects.