In 2018, DBS Bank came together with AI start-up impress.ai to implement Jim – Job Intelligence Maestro – a chatbot that helps the bank shortlist candidates for positions in their wealth planning team. This is primarily for screening for entry-level positions. Apart from process efficiency, the introduction of AI in the recruitment process is also aimed at eliminating bias and objectively finding the right candidate for the right job. The DBS chatbot uses cognitive and personality tests to assess candidates, as well as providing them with answers to the candidates’ frequently asked questions. The scores are then passed on to actual recruiters who continue with the rest of the recruitment process. DBS claims that they have curtailed the initial assessment time of each applicant by an average of 22 minutes.
While some organisations have started evaluating the use of AI in their HR function, it has not reached a mass-market yet. In the global Ecosystm AI study, we find that nearly 88% of global organisations do not involve HR in their AI projects. However, the use cases of AI in HR are many and the function should be an active stakeholder in AI investments in customer-focused industries.
Telstra employs AI to vet Applicants
Last month, Australia’s biggest telecommunications provider Telstra announced its plans to hire 1,000 temporary contact centre staff in Australia to meet the surge in demand amidst the global pandemic. In response to the openings, Telstra received overwhelming 19,000 applications to go through and filter, with limited workforce. To make the recruitment process more efficient, the company has been using AI to filter the applications – and has been able to make initial offers two weeks from the screening. The AI software takes the candidates’ inputs and processes them to find the right match for the required skills. The candidates are also presented with cognitive games to measure their assessment scores.
Ecosystm Principal Advisor, Audrey William speaks about the pressure on companies such as Telstra to hire faster for their contact centres. “Several organisations are needing to replace agents in their offshore locations and hire agents onshore. Since this is crucial to the customer experience they deliver, speed is of essence.” However, William warns that the job does not stop with recruiting the right number of agents. “HR teams will need to follow through with a number of processes including setting up home-based employees, training them adequately for the high volume of voice and non-voice interactions and compliance and so on.”
The Future of AI in HR
William sees more companies adopting AI in their HR practices in the Workplace of the Future – and the role of AI will not be restricted to recruitment alone. “A satisfied employee will go the extra mile to deliver better customer experience and it is important to keep evaluating how satisfied your employees are. AI-driven sentiment analysis will replace employee surveys which can be subjective in nature. This will include assessing the spoken words and the emotions of an individual which cannot be captured in a survey.”
In the future, William sees an intelligent conversational AI platform as an HR feedback and engagement platform for staff to engage on what they would like to see, what they are unhappy about, their workplace issues, what they consider their successes and so on. This will be actionable intelligence for HR teams. “But for a conversational AI platform to work well and to encourage users within the organisation to use it, it must be designed well. While it has to be engaging to ensure employee uptake, the design does not stop at user experience. It must include a careful evaluation of the various data sets that should be assessed and how the AI can get easy access to that data.”
AI and Ethics
With the increased use of AI, the elephant in the room is always ethical considerations. While the future may see HR practices using conversational AI platforms, how ethical is it to evaluate your employees constantly and what will be the impact on them? How will the organisation use that data? Will it end up giving employers the right reasons to reduce manpower at will? These and allied issues are areas where stricter government mandates are required.
Going back to AI-assisted recruitment, William warns, “Bias must be assessed from all angles – race, education, gender, voice, accents. Whilst many platforms claim that their solution removes bias, the most important part of getting this right is to make sure that the input data is right from the start. The outcomes desired from the process must be tested – and tested in many different ways – before the organisation can start using AI to eliminate bias. There is also the added angle of the ethical use of the data.”
Agents are the most valuable assets in a contact centre. In the current environment, the biggest threat is agents getting infected, causing the closure of contact centres for weeks or possibly even longer. We are already seeing the impact of that with offices being shut, students not going to school and industry gatherings and events being put on hold or cancelled. So having a business continuity plan (BCP) is critical. The BCP should include ways to continue to engage with customers.
The contact centre manages live interactions. Every second there are voice calls coming in, emails received and self-service tools being accessed. It is important to have multiple backup plans – both from a people and a technology perspective – to keep operations running effectively, without calls being put on hold too long or with other channels going unanswered. Contact centres battle with these challenges every day and the situation will get far more serious with the ongoing changes we are witnessing.
Some important considerations include:
Having a backup plan allowing agents working from home
More contact centres today are gearing up to agents working from home, but the process is not an easy one. To begin with, the initial set up includes having the right connectivity and a reliable network. Ensuring that the agent has the right working environment with minimal distraction is crucial. A good quality headset can help. A poor-quality headset will only create unwanted problems with understanding customer issues and handling them. Other concerns include security, tracking how data is being handled, agent under-performance and safety of the agents from an operational and health perspective. Measures such as listening to call recordings and storing them centrally are growing in importance. Multi-factor authentication and analytics using agent logs are some measures that can be put in place.
While there are lots of tools and technologies to monitor and check on agents, the key for home-based agents will be trust. Some outsourced contact centre providers that have been using home-based agents for years have stated that having trust and not micromanaging the agents, is essential for the model to work. Some contact centres have also deployed a BYOD policy for home-based agents assuming the right security, device management, application management and authentication measures are in place.
Organisations should also consider actively recruiting additional home-based agents. These agents could be retirees, currently unemployed or people with mobility issues who prefer to work from home.
Given the difference in the working environment, the metrics used to measure agent performance needs to be modified to be more realistic and fair to both agents and organisations.
Employing home-based agents will drive employment amidst challenges in the economy. Ecosystm research finds that more than a third of organisations do not have provisions for agents working remotely (Figure 1).
For a long time, the industry has talked about the rise of home-based agents and while it has received positive momentum, it has never really taken off in a big way. This time it will.
Managing spikes in voice and non-voice calls
In industries such as healthcare and airlines, call volumes are exceeding normal volumes. Having the ability to deflect the calls to other non-voice channels will be important. It might need the Interactive Voice Response (IVR) scripts to be changed from time to time to manage the flow of the calls. This is when cloud architecture becomes important. The cloud model can be used to make changes to call workflows easily. The sudden peaks will also require changing the channels easily and without intervention from IT. This is where the agility of cloud comes in as it allows changes and additions – for example when 500 agents need to be added or moved to work on other areas – to be made more easily. Ecosystm research finds that currently, only a third of organisations have their contact centre solutions fully on a cloud, with another 66% with partial cloud solutions. This is set to change with the rise in the number of home-based agents.
There should be thought and planning on how to deflect voice calls to other self-service channels. In the current environment, some organisations deploy a call back option when there is an overflow calls. Similarly looking at deflecting voice calls to self-service channels to ease the load on agents should be evaluated.
Managing back up locations (onshore and offshore)
Contact centre operators are looking at ways to isolate agents and keep them safe. Apart from very strict hygiene measures, organisations are also restricting agents to their specific floor. Some are looking at having agents split into different centres, to contain the risk of mass infection.
Several contact centre operators are building contingency plans to route calls to outside the onshore location in case the situation in a site or a cluster worsens.
For back-end contact centre activities and non-voice calls, taking the load off from the current onshore setup and pushing them offshore, can be an option. The best place to start would be by evaluating each client contract and SLAs especially on security, regulation and privacy issues regarding customer data-handling.
There will be a lot to be considered too should the country go into the full lock-down mode as we are starting to see with a few countries. This makes the case for employing home-based agents stronger.
Using messaging apps, the website and FAQs for daily notifications
Many contact centres are informing citizens and customers about the changes in business operations, services offered, refunds, where to go for help, what do to in an emergency and other essential information through the website, app or the updated FAQ. This will help reduce unnecessary voice and non-voice enquiries to the contact centre. During an emergency, it is normal that phone queries will rise and developing a detailed FAQ is critical to counter that. The more detailed the FAQ giving essential information, the more agents will be able to focus on the more essential day to day activities. Several companies are now sending pop-ups within apps about daily changes to avoid an overflow of inbound enquiries.
Virtual Assistants and Conversational AI can help to ease the load
The more intelligent the virtual assistant and conversational AI platform, the more a customer will be able to get the right response. The challenge has been that many platforms are poorly designed and customers get frustrated because they are unable to get the basic information they need. In times of high inbound activity, if answers to simpler queries can be provided through a chatbot, it can help ease the load on agents. It is good to start planning for this as it will take some time to get the virtual assistant platform up and running and even longer for the algorithms to learn from historical patterns to work well. While it may not be the perfect solution now, planning for a Conversational AI can bring some sort of balance back to the contact centre. Having a solid knowledge management system at the back-end cannot be compromised. Without a good knowledge management system, the virtual assistant solution will force customers to leave the self-service platform and place a call to the contact centre, defeating its very purpose.
The challenging situation we are in is undoubtedly putting pressure on contact centres. It is not uncommon now for customers to be put on hold – for more than two hours and in some extreme cases more than 7 hours! In times like this, understanding data and the patterns around data from each customer touchpoint will help plan the next steps on how best to navigate the situation. Testing and pre-testing the channels and the changes made before they go live must be done rigorously.
Whilst these are very challenging times for the economy, the good news is that contact centres are successfully piloting or have already implemented some or all of the above discussed here. Outsourced contact centre providers are running pilots across various locations and applying technology to deal with the challenges they are witnessing daily. Technology has also come a long way in the contact centre space, and by the application of the right technologies, scale, and business continuity measures, resilience can be achieved.
This blog was created with input from CX leaders across the entire Asia Pacific region. The author wishes to thank everyone for their valuable input.
