Introduction
by Matt Baron in association with Roman Kitsela & Max Clayton
A data-driven transformation has swept across industries, and the contact centre sector is no exception. Powered by the wealth of data available, Artificial Intelligence (AI) is a promising opportunity in Contact Management businesses. As businesses strive to maintain a competitive edge, optimise operations and drive profitability, AI emerges as a pivotal tool in redefining the future of customer service.
But what does AI in the contact centre industry look like? How can businesses effectively wield this powerful technology, and what would deployment look like in practice? Is this the holy grail to increasing operational efficiency that many are promising, or is it simply a buzzword used to generate interest for otherwise uninspiring companies? To veterans of the contact centre industry, new technologies appearing on the horizon are not a recent trend, but the cautionary tale of Robotic Process Automation (RPA) highlights the inherent risks of rushing implementation due to hype.
This executive overview delves into the growing role of AI within customer service: the challenges, the risks and the opportunities. There is no doubt that the potential benefits are immense – from cost savings through automation to brand enhancements via improved customer experiences. As always, however, the devil is in the details and without well-planned pragmatic deployment, companies risk investing cash, time and effort into tools that ultimately will only provide marginal benefits.
What is AI: CS?
Defining the Modern Age of Customer Service.
At the heart of AI’s transformation of customer service (CS) lies the ability to understand, anticipate and address customer needs proactively – AI: CS represents the amalgamation of AI-driven tools within the contact centre industry.
The modern contact centre, powered by AI: CS, goes beyond merely handling calls, emails and maybe chats. It’s a digital hub where customer contacts are addressed and resolved efficiently, with a personal touch, ensuring higher customer satisfaction. For the business, this means a reduction in agent handling times, increased first-time resolution rates and agent productivity. For the customer, this means an easy, seamless experience – data relevant to their query is collected efficiently, problems are resolved the first time and calls last less time than ever before.
Given the digital nature of today’s customer interactions, integrating AI is not just a luxury; it’s a competitive necessity. A delay in embracing AI: CS could mean losing out to competitors who offer faster, smarter and more efficient customer solutions.
Why should companies take AI: CS seriously? The reason is twofold:
(1) Customer Expectations: In an era where information is at the fingertips, customers expect immediate, accurate and hassle-free solutions. AI equips contact centres to meet these expectations head-on.
(2) Operational Efficiency: Beyond the immediate customer interaction, AI can reduce operational costs, minimise human errors and expedite resolution times.
We highlight two prime technologies particularly well-suited to contact centre deployment – with minimal operational risk, straightforward deployment and a clear pathway to profitability.
- Semantic search (the ability to search text using the natural meaning of phrases and sentences) – allows agents to more effectively search through knowledge bases using conversational style questions instead of traditional keyword-based searches.
- Asynchronous messaging – employs AI to suggest content for replies through to drafting replies to “asynchronous” (written contacts not requiring a real-time response) messages, which an agent can edit or validate prior to sending.
How much can you benefit from AI: CS?
Boosting Profits and Enhancing ROI.
- Operational Efficiency: Introducing AI into customer service augments human capabilities. AI chatbots, adept at handling simple routine queries, free up human agents to resolve complex issues. The resulting reduction in manual intervention streamlines operations, decreasing overheads and minimising the need for extensive manpower. The end result is a significant cost saving and a more efficient operational model.
- Data-Driven Insights: AI’s inherent data analysis capabilities can turn every customer interaction into actionable insights. Predictive AI: CS tools can model customer behaviour based on historical data – anticipating customer needs ahead of time, predicting churn, recommending next-best actions, etc. With AI, businesses transition from merely reactive to proactive stances, with insights necessary to address problems before they happen.
- Enhanced Return on Investment: Investing in AI: CS is more than just a mere technological upgrade; it’s a strategic blueprint for future-proofing customer service. The initial investment yields long-term dividends, both tangible: operational savings through a reduction in Average Handle Time (AHT) and increased rate of first-time resolution, and intangible: improved brand reputation through more personalised customer service and more consistent branding standards
In a broader sense, AI adoption can be seen to not only offer monetary returns but also strategically position companies within an ever more competitive market.
- Elevated Customer Experience: Leveraging semantic search and chatbots, contact centres can rapidly delve into vast data repositories, delivering precise and consistent answers. This drastically reduces waiting times, ensuring customer queries are addressed with both speed and accuracy.
In what order should you deploy AI: CS?
Laying Down the Blueprint.
Each company may need a different approach to implementation, and much will depend on the specific capabilities that exist within the company before AI adoption. The starting point for any business should be to look inward – examine the way your business is currently set up.
Firstly, before any thought of implementation: Optimise Existing Processes, Cleanse Data, and Validate Knowledge
Before diving into AI, streamline existing workflows/processes with a view towards AI implementation. Perform comprehensive data audits to ensure your data repositories are in a usable condition. Whereas a human agent can filter useful data on a per-use basis, Machine Learning (ML) algorithms can become wildly misaligned if cleansing and validation of data is not completed.
Identifying underlying inefficiencies within your workflow is key to extracting the most from AI integration. There may be assumptions within operational designs that would weaken the effectiveness of AI: CS upon launch. At the end of the day, it is easier to integrate AI into a well-oiled machine rather than expect AI to paper over cracks.
With solid operational foundations in place, it is possible to begin the process of AI implementation. However, for many businesses, taking this next step represents a potential risk. Rushing client-facing AI services or exposing key data to machine learning models may be a step too far – starting small is a wise move.
Then, with solid data foundations in place: Deploy Proof of Concept
Successful internal pilot projects are the best way to build confidence within major company stakeholders about the effectiveness and viability of new AI: CS tools. These should be trialled internally within the business to guarantee minimal disruption to the overall operation and will serve as a clear demonstration of the potential benefits to productivity within the AI-augmented workforce.
Start by splitting a team of agents into a control group (consisting of regular agents) and an AI-augmented group – armed with AI: CS tools trained on a well-maintained knowledge base. The performance of each group will be compared directly – quantifying the true added value these AI: CS tools provide.
To mitigate risk, initial trials should be deployed on small-scale internal data. We recommend the following to start with:
– Agent knowledge base – for asynchronous messaging
– Internal HR databases – for semantic search
What does this look like for semantic search?
Training AI models on internal HR datasets empowers AI-augmented agents with advanced search capabilities. Within the company, employees with the semantic search are better able to find answers to their own questions – reducing training and onboarding times for reduced overheads, locating subject matter experts quickly to boost collaboration, and enabling knowledge discovery for continued upskilling.
What does this look like for asynchronous messaging?
Large language models trained on relevant datasets (such as the agent knowledge base) are particularly well-suited to drafting responses to customer enquiries when an immediate response is not required. For such asynchronous channels, the AI-augmented agents are served suggested responses written by AI: CS tools on which they use their personal judgement to make edits before sending out to customers. This allows the agent to respond to contacts sooner – reducing waiting times, ensuring optimal resolution pathways – reducing repeat contact rates and adding consistency to responses – in line with company branding. The endpoint of a human agent making final edits also puts safeguards in place to insulate the business from the risk of an autonomous client-facing AI model.
Both above examples should also ensure that the data contained within the source repositories is cleansed, validated and ready to use.
As familiarity with AI-tools improves and the understanding of how best to utilise them within the company grows, it will become possible to start expanding your AI use cases. Yet rushing to embed the new technology within every facet of the business can lead to “integration hell” wherein separate tools may work well in isolation but coordinate together poorly. The key is to treat every new technology as another learning opportunity – embrace an iterative approach to deployment.
Finally: Scale Gradually
With each successful pilot run, implement lessons learnt and expand AI usage. For agents handling asynchronous messaging queries, you may find immediate productivity gains, so expanding these tools to the entire team is the next step. With lower handling times, agents will be free to focus on more complex tasks (which AI:CS tools are not yet able to handle effectively). If you find that agents are having to make fewer and fewer edits to AI-constructed responses, you should assign a success percentage. E.g. 9 times out of 10, the agent is making no edits to the AI-written response. You may even want to try a fully automated response. This scaling process will be ongoing across every AI: CS tool and organisations will need to be agile in their decision-making, ready to respond to changes in customer and agent behaviour stimulated by their interaction with AI: CS.
Some trials may prove to be unsuccessful. For these, taking a step backwards may be necessary to examine what went wrong. In the long run, the failed attempts could yield some of the biggest benefits to your organisation, either through focusing your attention on use-cases inherently better suited to your business or by further improving your underlying processes.
To make the most of AI: CS, you will eventually need to scale up deployments. At each step of this process, continued evaluation and refinement are essential to ensure that employees are continually trained, transitions are smooth, and you do not overextend.
How do you deploy AI: CS for maximum benefit?
From Vision to Reality.
With the hype surrounding AI, you could be forgiven for assuming that AI technology is a silver bullet for each and every problem your company may have. Businesses are scrambling to appear cutting edge by labelling every new tool “AI-powered” or “ML-driven”.
The reality is that for many use cases, AI may not be effective or even appropriate. Even in cases when AI is appropriate, the fact remains that the biggest bottleneck to the benefit you can extract from AI as a tool is caused by the strengths or weaknesses of your underlying processes. A poor process is still a poor process, even if enabled by AI.
Our view for deployment can be summarised in 3 pragmatic steps:
- Sort your processes out
- Cleanse knowledge repositories
- Organise your data infrastructure
In the detailed study by Avocado55, every single one of the 14 service businesses analysed showed measurable benefits upon the introduction of AI: CS – both in terms of increased Customer Satisfaction and reduced Unit Costs.
The takeaway is that while AI is not a one-size-fits-all solution, there is undeniable value in its adoption within customer service. Ultimately, the success of this deployment is dependent on a structured, pragmatic approach and the alignment of AI: CS with your business needs.
AI-Augmented, Not AI-Replaced
AI: CS tools should not be viewed as replacing the human touch – but rather enhancing it. While AI is clearly better at processing vast data and identifying patterns rapidly, humans bring depth, nuance, and an emotional insight that machines cannot replicate. Merging the two creates a synergy that elevates the service experience. In this partnership, it’s clear: the whole is greater than the sum of its parts.
Continuous Improvement Protocols
Successful AI:CS deployment is not a one-off task; it is an ongoing commitment to excellence. Your business must establish protocols for continuous assessment and improvement of AI:CS tools. Regular evaluation ensures that the system remains updated, relevant, and effective in meeting your business needs.
Feedback loops, both from customers and internal teams, play a vital role in identifying potential areas of improvement and adapting the AI system accordingly. Moreover, just as the AI: CS tools require structured updates, so does your team. Regular training sessions ensure that your employees are maximising the full potential of AI – after all, an AI solution is only as effective as the people and processes that surround it.
The AI Blueprint: Theory into Practice
- Assess Your Needs: Every contact centre is unique. Begin by assessing the types of interactions that occur frequently and understanding the specific requirements of your business.
Evaluate the following:
- How might semantic search enhance your customer experience?
- Are written asynchronous contacts significant in your operation?
- Is your knowledge repository up to date?
- Do you possess in-house technical expertise, or will you outsource AI: CS solution implementation and maintenance?
- Choose the Right Tool(s): Invest in AI tools that align with your needs. Ensure they are scalable, easy to integrate and come with robust support.
Evaluate the following:
- Which tool(s) best serve your needs?
- How adaptable is the AI solution to changes in other parts of your tech stack?
- Are there any legal or regulatory considerations in your industry regarding AI/ML use?
- Do your chosen tool(s) scale effectively within your budget?
- Optimise and Cleanse: Review all key processes (we suggest rapid re-engineering) and the ways your customers and agents interact with customer service tools. Evaluate the issues you may have with your data and put together a plan for regular data management.
Evaluate the following:
- What is your existing process for handling asynchronous communication?
- How often do you update your customer communication protocols?
- How will you audit your knowledge/data repositories?
- Are your processes accurately mapped, reflecting the real-world operations of your business?
- Collaborate Across your Business: Encourage inter-departmental cooperation to get the most out of AI: CS. From IT to operational managers to customer service agents, everyone plays a role in successful AI deployment.
Evaluate the following:
- How are you gathering feedback from different departments on the AI’s performance?
- What platforms or channels are in place for interdepartmental communications about the AI?
- How open are you to collaborating with AI vendors for tailored solutions?
- Educate & Train: Equip your team with the necessary skills to work alongside AI. Continuous learning is key to adapting to this ever-evolving landscape.
Evaluate the following:
- How do you train your agents to handle asynchronous communication?
- How are you ensuring the human touch is not lost in customer interactions as AI’s role expands?
Conclusion
In the rapidly evolving landscape of the contact centre industry, the future is undeniably intertwined with artificial intelligence, though the introduction of AI presents both opportunities and challenges.
Foremost among these challenges for companies will be issues of data privacy. As AI: CS tools gain increasing access to private company data, effective safeguards must be put in place to ensure data is insulated within the company. With the automation of data handling, there’s an inherent risk of breaches or misuse, especially if the AI systems are faulty or compromised – putting both customer trust and corporate reputation in jeopardy.
Additionally, exposing AI:CS tools to customers directly without human intervention introduces new multifaceted hurdles to overcome. Not only is there a risk of AI tools delivering incorrect answers and experiencing hallucinations, but there is also the possibility of increasing impersonality as human-to-human interactions become rarer. The brand risks are, therefore, about more than just accuracy; they also include the customer’s perceived value when interacting with an algorithm vs. a human.
Despite the risks and challenges, with correct pragmatic deployment, the benefits of the AI tools will outweigh the drawbacks, and risk factors can be managed. We believe AI: CS is not just a buzzword; it’s a strategic imperative for businesses to thrive in an increasingly competitive and digital world.
With a methodical approach outlined above, the rewards of integrating AI: CS into your business can be monumental, and risks understood, managed and minimised. Any business decision carries with it an opportunity cost of actions not taken. At Avocado55, we see the opportunity cost of ignoring AI: CS as now being too great to ignore. In the past, the question may have been, “Can I benefit from AI: CS?”, now we see it as “How much can I benefit from AI: CS?”.
