Let’s understand how a conversational AI can help connect the problems to the solutions effectively in real-time. Take an instance where there is a conversation between a customer care executive of a fashion store and a customer that is effectively solved with the help of a conversational AI.
The conversational AI uses sentiment analysis to understand the customers’ state of mind, makes recommendations to the executive for better solutions, and quick recommendations related to the offers of the brands the customer liked or has in their wishlist increases the overall satisfaction of the customer.
This is how conversational AI helps in solving customers’ issues in real-time. Many contact centers are already infusing AI into their operations and help desk tasks. Although AI is not going to eliminate the staff, it is going to aid a huge scale-up of the help desk services.
Are you a contact center organization looking for real-time & customized analytics solutions for your business?
You would have learned how Advanced Analytics aids the contact centers to enhance the customer experience thereby converting contact centers to revenue centers. Our Conversational AI product FLASH comes in handy providing meaningful insights in just a matter of seconds.
Conversational AI-led analytics is indeed the game-changer for contact centers was a 3 part blog. The part 1 concentrated on the problems faced by the contact centers which was followed by the solutions conversational AI could offer in part 2 and how the conversational AI solved the problems in real-time in part 3 of the blog.
BPO – Analytics: A prerequisite for operational efficiency:
The contact center has extensive information about customers, their behavioral and non-behavioral data, product experience, and the interactions with the frontline staff. With appropriate analytics, the information could be modeled to predict very useful characteristics to promote the product, up-sell/cross-sell, predict the intent of the customer call, etc.
BPO sector sets forth the following expectations which can be achieved through Advanced Analytics:
The historical and current data needs to be analyzed to uncover the masked opportunities for the strengthening of the organization.
The potential customers have to be identified, segmented, and targeted for the marketing campaign.
The unstructured data from the customer complaints have to be extracted and analyzed further to enhance customer service strategies.
The data from social media requires to be mined to determine emotions and opinions about brands, products, and services.
The customers who are likely to cease the service from the business need to be predicted.
The intent of the customer has to be predicted to give what exactly the customer wants before they ask for it.
There is a need for prediction of the future using extensive evaluation of the past performance
Recurrent issues of a product should be categorized for future enhancements.
The best products to be recommended to customers are to be predicted so that they can easily find their desired product.
The total value a customer can bring to a company throughout their lifetime is to be measured.
Analytical Solutions for Contact Centers:
geniSIGHTS understands the pressing needs of Business Process Outsourcing companies and has come up with a handful of analytical solutions. These intelligence solutions yield valuable insights to organizations ensuring the highest level of customer satisfaction.
Some of the analytical solutions offered are:
The Uplift Modelling ensures that the customers who have higher propensity scores are rightly identified and targeted for the marketing campaigns. The customers are segmented into four categories based on their behavior after being encountered with a marketing action.
Sure Things – Customers who would have responded whether they were targeted or not.
Lost Causes – Customers who will not respond irrespective of whether or not they are Targeted.
Sleeping dogs – Customers who are less likely to respond because they were targeted.
Persuadables – Customers who only respond to the marketing action because they were Targeted.
The cross-sell recommendation suggests the customers for the complementary product when they purchase a new product. The upsell recommendation shows up the customers with a better version of the product that they plan to buy. This system uses affinity analysis/association rules mining to analyze the co-occurrence of relationships among activities performed by customers and discover the cross-sell/up-sell opportunities accordingly.
The recommendation system exceptionally helps the business to predict the most likely product that the customer will buy. This system can therefore provide more personalized offers for the items of the customer’s interest.
Predicting customer churn rate
Customer Churn analysis predicts the customers who are likely to cancel their subscription from the company. The analysis also helps to understand why the customers are leaving and the possible ways to reduce the churn.
Sentiment mining extracts the necessary information from social media to understand what people think about a product or how they look up to a brand. The emotional tone behind each mention in social media can be derived from this analysis.
Customer Effort Analysis
The customer effort analysis brings out the amount of effort a customer puts in getting his issues resolved through various touchpoints.
These are some of the advanced analytical solutions that the conversational AI could offer for the problems faced by contact centers. Wait for edition 3 to find out how Conversational AI resolves all the problems of contact centers in real-time.
Are you a contact center organization looking for real-time & customized analytics solutions for your business?
Learn how Advanced Analytics aids the contact centers to enhance the customer experience thereby converting contact centers to revenue centers. Our Conversational AI product FLASH comes handy providing meaningful insights in just a matter of seconds.
Conversational AI-led analytics is indeed the game-changer for contact centers is a 3 part blog. Part 1 concentrated on the problems faced by the contact centers which is now followed by the solutions conversational AI could offer in part 2 and how the conversational AI solves the problems in real-time will be in part 3 of the blog.
We feature Mr. Dorai Thodla in our monthly Feature corner, He’s Founder of Technology Strategies LLC and iMorph Innovation Center Pvt. Ltd. this month. He has co-founded 4 software companies – two in India and two in the USA. Mr. J. S. Sridharan, our Co-Founder, and Senior Vice President ask a few questions to Mr. Dorai Thodla. You may follow Dorai on Twitter (@dorait) and LinkedIn.
He helps organizations leverage emerging technologies for building skills and creating innovative products. His companies build Information Assistants, a collection of tools for gathering and analyzing information.
He considers the ability to do informal research and tracking technology trends as important skills for the future of work.
Sridharan – With reference to the informal research process, could you tell us more about which areas this would apply to?
Dorai – There are a few areas where quick informal research can help.
1. A product team trying to validate a set of ideas. They may need to know what already exists and whether they should build a product.
2. A startup with reasons similar to product groups, but may need additional information about the market size, how the market is growing, current players, and whether there is a viable business model.
3. Finding gaps and opportunities in emerging technology trends. Inferring latent needs from discussions and analyzing competitive market spaces.
Sridharan – Why do you think organizations would undertake informal research and at what level?
Dorai – It depends on the size of the organization. Small to medium organizations may use it at a tactical level. For example, a small product group may use it to do competitive analysis, opportunity analysis for product extensions, etc.
Sridharan – We know that we can leverage Twitter as a tool for any informal research. Can you share insights on how twitter hashtags will help organizations in this research?How can we leverage LinkedIn for this? What are some other tools we could use for doing this informal research?
Dorai – Hashtag’s usage is not uniform. Some of the most influential tweeters (like Paul Graham) do not use hashtags at all. An analysis of hashtags shows that they are used when you want to join conversations or when you are in a conference or other similar events.
Twitter is a broad platform serving a wide variety of people. You can see authors, journalists, news organizations, technology people from the small, medium, and large organizations. LinkedIn is different. It started serving the needs of HR and recruiters. For a while, now LinkedIn is broadening the base but the type and quality of conversations are very different. It is easier to find influencers and people from large organizations. The profiles on LinkedIn are much richer.
The major problem with LinkedIn is a highly restricted API. So gathering information using automated tools is a bigger challenge.
With respect to other tools, If you cannot afford to subscribe to expensive research resources from experts in the field (students and startups have this problem), there are many tools you can cobble together to do your own research. You can call this DIY Research or Informal research. These tools include:
– Search engines
– Blogs and syndicated feeds
– News sources (both raw and aggregated)
– Websites and Portals
– Social media tools like Twitter but also LinkedIn and to some extent Facebook
– Wikis including Wikipedia and its properties
– Trade and Research publications
– Data and Scholarly search engines
Sridharan – How can startups leverage social media for tracking technology trends?
Dorai – If you want to speak the language of your customers, you need to use the vocabulary of your market place. These include idioms, phrases, and memes in those places. For example, tl;dr (too long did not read) is a common term among software developers. Other terms include “code smells”, “tech debt” etc. You can gather tweets of developers, do a frequency analysis of the terms to understand the terms of the group.
The importance of understanding the vocabulary of users may be much fold. If you want to use content marketing, you may want to use that language in your content.
There are a few ways. The more open the social media, the better. That is why Search and Twitter are our primary tools of choice.
1. Google, Facebook, LinkedIn, Twitter will provide you data about the size of a specific niche audience.
2. The search parts of Google, Twitter, and others will provide discovery tools. I prefer tools that provide good APIs. These tools make data-driven decisions easier. For example, on Twitter, I can type a search, find popular hashtags, lists, and mentions, and can mine them for more information. This is not so easy to do for example in LinkedIn (despite their having an API) and Quora. So people resort to using scrapers and RPA (robotic process automation) tools.
3. Blog searches or Feed Searches
4. Job searches – for certain types of research jobs are a leading indicator
5. Conferences, meetups, and discussion forums are useful tools for identifying watch signals.
6. Funding patterns, especially Angel-funding are leading indicators. You can also track other rounds of funding and private equity to understand the market space.
7. Books and articles are mostly lagging indicators but they are good sources to track the popularity of certain technologies.
Sridharan – How can start-ups and organizations leverage this informal research for new product markets and finding new opportunities?
Dorai – There are several stages for a product startup. Let us look at the needs at each stage.
1. You come up with an idea for a product. You need to find out whether products similar to yours are there already and what are the pros and cons of each product (an analysis of alternative solutions). Having some competition validates that there are a problem and the need for a solution. But a lot depends on how established the competition is. At the end of this research, you identify a niche and some potential early adopters.
2. You build a prototype or a minimum viable product. You do this in stages. Once you have a proof-of-concept prototype or a functional, usable prototype, you need to validate that it solves the problem. You need research to identify these users, ways to reach them, and interact with them.
3. Once you have a few users using your product, you need to expand the number and also identify paying customers. You need research to go from 100s of users to thousands or tens of thousands of users.
4. Once you have sufficient users to feel comfortable with your validation, you need to price the product. If similar products exist, you need to research their pricing models. If there are no similar products but the market is being served by services, you need to research those pricing models.
As the number of users and types of users grows, you need more research to get to the next stage. Most of this research is not available in the market. Even if available, it may not be current.
Sridharan – Could you elaborate on some new research methodologies that can help startups?
Dorai – Startups have several free and paid resources for product discovery, content discovery, keyword discovery, and competitive analysis. Informal research strings together available tools to create your own Research Assistants. I see this as a multi-stage process.
– Info tools – tools to gather data from a variety of sources – Web, RSS feeds, Tweets, Posts, Discussion boards, Forums, and Blogs.
– Analyzers – Analyze information gathered via info tools – segment/cluster, discover topics, mine entities, derive the vocabulary of conversations, etc. The analyzers use Machine Learning, Deep Learning, Natural Language Processing, semantic tagging, and other emerging techniques.
Sridharan – With reference to businesses tracking technology trends, how do you think digital analytics help track the trends?
Dorai – Data gatherers will bring in a lot of data in various digital forms – text, images, speech (podcasts), videos. You need analytics for different levels of filtering. The first level simple analytics are early filters to separate signal from noise. Next level analytics provide good inputs for inference and prediction.
Sridharan – To understand customers and market size, which tools or platforms to look for?
Dorai – We can start with some of the available free tools. These include search engines, Twitter, LinkedIn, Facebook. They help you understand the market sizes since they need to know it to guide you in advertising. For example, using Facebook or Google ad products you can understand the size of a reachable market.
Using product hunt, beta list, and other similar services you can identify similar products. But nothing beats search. You can come up with a list of key terms that describe your product and try searches and analyze the results. Google provides a search API that you can use to automate this process.
You can use services like angel list to locate startups. You can use search to locate directories, professional associations in your product or technology space.
Twitter is becoming a great resource for discovery. Twitter APIs allow you to automate searches. Using Twitter you can find companies, products, trends, research reports, influencers, and discussions. You can create lists to manage the vast amount of information, retrieve tweets and links, and analyze the results.
Sridharan – How does text analytics help organizations in the changing business landscape?
Dorai – The outcome of searches (web search, Twitter search, Blog Search, Product Search, etc.) is short documents.
You need to text analysis to mine useful information from documents. Let us take an example of a blog comparing several products. You can use topic and keyword extraction techniques to derive useful information. You can use entity extraction to identify companies, products, and events mentioned in the article.
The lack of access to physical stores due to the pandemic has increased the traffic for contact centres worldwide. Contact centres are bridging the gap to communicate with customers, emergency helplines, on-call services, and inquiries. Even before contact centre agents were required to work from remote locations,Cisco’s 2020 survey revealed that 62% of decision-makers plan to implement a cloud contact centre in the coming year. 93% agree that technology is very important in creating a better customer experience. Contact centres are a treasure of data waiting to be discovered. They have data acquired through daily processes like telesales, telemarketing, customer surveys, collections by means of outbound services and customer services, emergency response past inbound services. Analyzing contact centre data can help not only the customers but the contact centre decision-makers as well, to improve service and increase efficiency. Artificial Intelligence-based software improves the customer experience by monitoring key phrases and then prompting the agent in giving proactive responses. They can identify challenging calls for agents to handle and route the calls depending on complexity. AI is also gaining popularity through chatbots and solving issues without the need for agent interference. Chatbots need to understand the tone of the customer and respond accordingly. A healthy Contact Centre must conserve excellent hardware and software solutions providing a direct impact on quality assurance/success metrics, call statistics, customer implementation, and proficient Contact Centre agents. Contact centres today can answer more than just calls. Let us look at what other questions can be unraveled through analyzing contact centre data.
1. Is the Customer Satisfied?
According to a 2019 survey byDeloitte Digital, Customer experience or CX is a high priority (57%) among the surveyed centers.Harvard Business Review found in a study that a focus on increasing (Customer Satisfaction) CSAT will help businesses retain 74% of customers for another year leading to a substantial increase in revenue. We can increase customer lifetime value by looking after the key determinants like customer acquisition, customer retention, and customer margin. IVR (Interactive Voice Recording) Systems have been in use recently to reduce the number of calls reaching an actual agent. Valuable insights on the reason for the call, the caller, Opt-out Rate, self-Services rates, and incorrect routes can be evaluated through IVR data.
Representative customer satisfaction metrics from FLASH, the analytical dashboard
2. Is your Product facing a regular or common problem?
Collating and categorizing issues is a great way to understand the product. A repeated complaint of the product needs to be fixed for future versions so that customers do not engage in expressing distress and reducing brand value. Categorizing these issues also helps in creating a common answer pool for contact centre employees. AI can be used to direct the calls to Subject Matter Experts to handle issues. This will reduce escalations and help employees with better ways to tackle problems. We can also perform sentiment mining to analyze the sentiment of the customers on service level parameters such as resolution effectiveness, feature variety, etc. by leveraging social media sentiments of the public on various products or services of clients for improved business services.
3. Is your Contact Centre Efficient?
First call resolution (FCR) is an important metric that measures the efficiency of the contact centre to solve issues in the first call. A long waiting time is another factor for low customer satisfaction. Measuring the average delay of calls or the call abandon rate or the response time will give us an idea about the need for more staffing to handle the surge in calls. Response time metrics and call resolution metrics are indicators of efficiency(image below depicts metrics from an AI-driven dashboard, FLASH). Customer Effort Analysis can be performed to measure the efforts spent by customers to get various services rendered through various touchpoints. Customer journey analysis can be used to identify the pain points that customers face while traversing channels for a specific purpose. Path analysis can help identify the patterns in the menu/caller path in a time slot. All metric dimensions can then be evaluated to arrive at actionable insights about the customer segment in qualifying dominant paths, dependent paths, and average time spent on the path, etc.
4. Is your Employee productive?
Average Sales per Agent is the measure of sales that an agent closes within a time period. We can use this metric to award agents performing well and provide targets for employees to work towards. An agent’s Utilization and Average Handle Time are two primary contact centre metrics to focus on when trying to measure an agent’s productivity. We must also be careful of the escalations that an agent faces and train employees who need to improve performance.
5. How many Customers at risk of attrition?
According to contactcenterpipeline.com, Attrition is considered the No. 1 challenge 27% (up from 19.2%) for contact centres in 2020. Measuring the Net Promoter Score or NPS of customers helps us to plan a better approach for managing customers who are at risk of attrition. Predicting churn rate qualifies the steady-state level of customers at any point in the network.
6. Are there any opportunities for Upsell and cross-sell?
Affinity analysis/association rules of mining are used to analyze the co-occurrence of relationships among activities performed by customers and discover the cross-sell/up-sell opportunities accordingly. Recommendation Systems help identify products customers need and help contact centres create personalized offers based on purchase history and products commonly bought together. Use of Uplift Modelling can be used to identify the right set of customers(the persuadable) to be targeted for marketing campaigns.
7. Can your competitor be a reason for low sales?
As contact centres are a hub of engagement, customers often give honest feedback on products. You also find that many potential clients are already using your competitor products. Employee engagement plays an important role in analyzing these potential risks to the business. Predicting the customer churn rate helps to identify the customers who are likely to leave the network.
Contact Centres are now moving into a space of self-service where customers are adapting to emerging artificial intelligence software and chatbots. Customers today do not like waiting for calls to be answered, they would rather finish queries through chatbots that do not consume time. IVR systems are slowly being replaced by intelligent assistants that provide human-like interactions. Having an advanced chatbot and analyzing its response can help you understand your customers. With these forms of service, direct calling and email, data is being collected from multiple sources. Decision-makers need regular reports to keep track of their performance. While automating this data is now a necessity, the use of AI to analyze this data is a competitive advantage. AI-based dashboards give you the insights needed to understand your data. They highlight possible improvements and forecast staffing depending on query traffic. They group dissatisfaction and anger sentiments and analyze agents who deal better with these emotions. Customer sentiments change with time, people are more impatient and now more sensitive due to long hours indoors. Many factors can be associated with good or bad performance and identifying these factors can be tricky for decision-makers. AI learning, analyses your data in real-time and helps you get a holistic view of your business. Analyzing and processing this data turns your contact center into an answering hub. Companies that ignore customer service face an inevitable fate of resentment and destroyed brand image. FLASH, a product by geniSIGHTS ensures your contact centre does more than just calls. FLASH is a one of a kind voice-powered AI tool that gives your contact centre the power to stay ahead of the competition and give highly efficient results. Our solution uses advanced analytical techniques through various statistical models and machine learning algorithms to qualify underlying patterns and mine hidden insights from data. It is lightweight and cloud-based, facilitating analysis at your fingertips even as you sit at home during this lockdown. Contact us at https://flash.genisights.com/ for a demo today.
The BI Survey research found that 58% of the surveyed companies relied more than half of their decisions on gut feel or experience over data. But two-thirds believe that information will be valued for decision making in the future. Highly data-driven organizations are 3X more likely to report significant improvement in decision-making, according to a survey by PWC.
The HiPPO effect, coined by Avinash Kaushik and reiterated by Bernard Marr in Forbes, is usually the most experienced, most powerful, or highest paid in the room. Once they voice their opinion, others find it disrespectful to voice their own. This may not lead to desired results. Let’s think if the decision would go on a similar path if the time and resources would be different. Given more time, you would probably rely on the opinion of a board of directors or your experienced colleagues. Will you still be completely confident?
Data however changes this. You have an unbiased opinion based on facts that leads to better results. Data analytics uses information from all sources and simplifies it to visual representations that highlight what you need. You are now capable of viewing all your information in just one screen. But data is useless until interpreted. Data-driven decision making or DDMM lies in interpreting this data for actionable insights. A term fondly called as data storytelling is considered an art. Artificial intelligence-based complex algorithms make sure to see more than what the human eye and mind are capable of. These insights help drive your decisions in directions that lead to success. So you have got your data, got your insights from this data, why then have you not used it to make the decision given to you 10 minutes ago?
The problem with many AI solutions is the inability to get exactly what you want when you want. You have a ton of data and a ton of insights but how does your BI tool know what you need to reach the answer for the question given to you 10 minutes ago. You have knocked on the door of data analytics and are seeking the solution. But can’t you just ask?
FLASH, by geniSIGHTS, gives you the power to simply talk to your BI tool and seek the information you need to come to the decisions that will open the right doors. In turn, FLASH also talks back to you with insights. FLASH is a one of a kind voice-powered dashboard that provides quick insights into your organization’s metrics. It goes beyond the functionality of Business Intelligence and uses conversational AI in the dashboard to respond to queries and generates real-time insights for actionable decision-making.
FLASH is AI-driven, lightweight, and easy to use. It can easily be integrated with your existing data systems without having any extra software. It is also accessible from any location and insights can be exported and shared with the members of your organization. The visualization can be customizable and easily understood without any expert knowledge of data. FLASH ensures that your entire team is on the same page, making your user experience smooth and without any delay. Unlike other solutions, FLASH learns continuously to understand what you need for your business and becomes your virtual data assistant over a period of time.
FLASH is your AI agent who has all the answers you need to make decisions in a FLASH. We have seen how Amazon’s Alexa has made life easier. Shouldn’t your dashboard be able to do that? Now you can make your decisions within 10 minutes and be sure of it.