Afiniti is one of the companies that apply artificial intelligence (AI) and machine learning (ML) technology to contact centers. The company has developed an artificial intelligence-based technology that matches contact center agents and customers to help contact centers significantly improve their business performance. Afiniti is one of Avaya's AIConnect initiative partners, but the company's solution works with any contact center or telephone exchange.
Afiniti is a passionate believer who believes that the contact center should be the guarantee for customers to get business services. This core function of matching customers to agents should satisfy this premise by routing customers to the agents most likely to achieve the desired results, whether it is new sales, resales, customer retention or problem resolution. The company's artificial intelligence algorithms enable intelligent routing to optimize these expectations and find the one that works best.
The idea of ​​as much as possible to agree on the outcome of the agent with the customer's expectations is not a new concept; tracking and monitoring agent performance is a standard contact center process. The Afiniti approach is unique in that it improves the overall performance of the entire contact center, not just the individuals who perform well. In addition, its effectiveness can be accurately measured. Afiniti is very confident in its technology. It is willing to install its solution, not asking for any fees from customers, and only paying for actual performance improvement results.
How does the Afiniti solution work?
Afiniti's technology uses several monitored machine learning techniques, including regression analysis and classification analysis, as well as Bayesian analysis and heuristic analysis. Using regression analysis, Afiniti's system complements customer characteristics and behavior with agent characteristics and performance to estimate the likelihood that a particular agent will achieve the desired business outcome when dealing with a particular customer.
Although this sounds simple and straightforward, it is actually quite complicated. The following figure shows the logical flow when matching an agent to a customer.
Afiniti's "client agent" matching algorithm uses artificial intelligence machine learning to optimize contact center performance
It begins with the customer's automatic numbering or caller ID display of the IVR system. The Afiniti System Application Data Linking System accesses customer-related information from more than 100 public and private data sources by leveraging the customer's phone number as the basis for initial data queries. For example, using a phone number, Afiniti can find a person's name, street address, city, and state. Using this location information, the company can access census data to determine the characteristics of the person's community, city, or town. It can then extract the person's data from credit reporting agencies, companies that track online user behavior, social networking sites, and even companies that track purchase information.
All of these data with customer attributes are entered into a neural network to calculate the customer type or customer characteristics, which is a process that is completed in microseconds. Companies using the Afiniti system typically have a small set of customer types, each of which can be further subdivided into subtypes. It is not uncommon for a company to have more than 100 different customer subtypes in a queue.
We should note that many end-user companies already have their own customer segmentation plans. Afiniti uses a company's own segmentation type as input to the neural network, as well as all other personal data collected by Afiniti about a particular customer, to produce a unique Afiniti customer segment.
The process of customer matching agents is done by a machine learning technique called regression analysis. The Afiniti system uses data from customers and duty agents as raw data for regression algorithms. Customer data includes customer types and any other existing data from the CRM database. IVR data and queue data are also involved in the analysis because they provide information about the customer's intent - the reason they call. For example, do they want to buy something or want to talk to technical support, or do they have a request to interrupt the service? Agent attributes, including performance, are also fed back into the regression model. Other attributes used by the model are based on an optional 20-minute survey of new contact center agents or an initial setup based on the Afiniti solution. These other attributes include various factors such as like/dislike, married/single, male/female, child/no child, hobbies/interests, sports or not sports, music of interest, sports interest, And where they live.
Based on customer and agent data, Afiniti software uses AI regression model parameters to calculate how a particular agent successfully meets business goals (sales, subscriptions, cross-sells, retention, etc.). The calculated agent probabilities are compared to each other based on the difference.
Finally, based on the commitment and operational rules of the contact center service level, Afiniti software made an optimal agent pairing. It follows all contact center rules: queue length, call distribution between agents, and more.
Importantly, the Afiniti solution works under the constraints of the contact center operating rules, and it does not necessarily have to be beneficial to a certain agent. In the large contact centers involved in Afiniti, each agent's goals are set to best meet the business goals of the business. Moreover, because it is fair to measure its own performance, Afiniti will treat each agent fairly.
Training and continuous learning
Afiniti software takes two weeks to four months of training and learning time (primarily for privacy and security reasons) before it can start working.
Considering that a single contact center agent may need to pick up about 700 calls per month, or about 2800 calls every four months, self-training may seem like a long time. If a company has set up 100 different customer types, then this is equivalent to 28 calls per seat per seat per customer type, which is actually not a lot of data for training machine learning systems. .
Therefore, once the system is up and running, the performance of the daily agent is continually fed back into the regression engine to update the system during the next day's call. This continuous learning mechanism is useful because agent performance and customer preferences may change over time, and events that temporarily change the needs of certain types of agents may suddenly appear.
Constant feedback loops also help train the new seat system. The attributes of the new agent are entered into the system as soon as they are recruited, and the Afiniti software begins to calculate the possibility of the agent as soon as the best match is considered. Of course, with time and experience, the matching performance of the agent should be improved.
Prove Afiniti's own performance: how Afiniti makes money
As mentioned earlier, Afiniti is a very confident technology that can install its AI-based agent-customer matching system to the contact center for free. After the customer and Afiniti agree on the performance metrics of the system, Afiniti charges based on its actual effectiveness. For example, if an Afiniti customer wants a higher sales conversion rate, Afiniti will be paid a certain percentage based on the higher sales that it can prove.
Afiniti uses statistical sampling to prove its performance: every hour, the Afiniti system closes the matching function of the client's agent within a certain period of time. It is then compared to the agent performance when the Afiniti system is started. In this way, both Afiniti and the customer can see the impact of the Afiniti solution on the expected results and the payment based on actual system performance.
Proof of its validity
A case study on the Afiniti website proves that T-Mobile describes how it turned to Afiniti solutions in the face of paying users seeking higher conversion rates. According to T-Mobile, the Afiniti system adds $70 million in additional sales each year. In other case studies, other multi-billion dollar companies have shown that the percentage of sales growth is inversely proportional to the dollar value. However, in a large company, a small increase in sales is equivalent to a large sum of money.
Afiniti positions its software in a large contact center. The company claims that all major US telecommunications companies use its technology and currently has 150 deployments.
CB Insights, a technology market intelligence platform provider, recently ranked Afiniti as the number one of the 100 most promising private artificial intelligence companies in the world. Based on its previous round of financing, Afiniti is worth about $1.6 billion.
AI is clearly entering the communications and collaboration arena, adding intelligence to contact center routing; providing smart chat bots and virtual digital assistants; or starting video conferencing based on facial recognition. Afiniti's story resonates because it uses artificial intelligence (AI)/machine learning (ML) in production, and its revenue model is so transparent and straightforward.
Wall Power Switch And Socket,Remote Control Electrical Outlet,Battery Powered Light Switch,Electrical Outlet With Switch
ZHEJIANG HUAYAN ELECTRIC CO.,LTD , https://www.huayanelectric.com