Seizing the Automation Opportunity
One of the first stumbling blocks of business process transformation is identifying meaningful automation opportunities. A typical process will have a management consultant, system expert, or Lean Six Sigma specialist to monitor manual processes for a set time, collect performance data, and make recommendations. Companies supplement this assessment with executive ideation sessions, operational managers’ workshops, and superuser interviews. The traditional way to speed up and improve the accuracy of this approach is to add more consultants over a longer period of time analyzing larger data sets. Yet, despite these efforts, not every process is observed and not all automation opportunities are identified and prioritized.
In this blog, we look at tools and algorithms to streamline ‘the discovery stage’ and recommend effective automation deployment in the front-office.
From manual to automated discovery
As an example, Alorica has developed an automation approach for the discovery stage, calling it ‘One-Click Automation’. Its unsupervised machine learning algorithms measure agents’ desktop utilization on a select group of 10-20 users. The bot monitors the users’ screen toggle, clicks, data entries and entry type, and tracks time per system and per screen, the time duration for performing each step, and the number of used fields. It counts errors, frequency, variance in the time and data entry quality, the end-to-end handling time, the complexity, and importance of the tasks. It also tracks the underlying communication between systems and exchanged information.
The bot does not capture personal data or other input information, remaining non-invasive to the systems. It also does not intervene with the users’ actions. Once the bot collects enough information (~100k data points per LOB, typically in two weeks), it creates a visual map of the observed processes. The process map visualizes the screens, the start and end points, the navigation between them, and the direction with drill-down functionality. For large product portfolios with multiple support processes (for example, in technical support), Alorica requires a bigger user pool to observe.
The automation specialists in the DCOE analyze the discovery bot results, estimate savings opportunities, and recommend and prioritize bot implementations with workflow business process, exceptions adjudicator, and look-up. Because the collected information shows the actual steps and the specific systems and fields to move the data, the created automation workflows make the bot training quicker and more precise. It also provides cost-benefit analysis and ROI calculation for each automation.
Implementations for a healthcare client
For a U.S. healthcare payer client, Alorica deployed the discovery bots on a statistically relevant number of agent desktops for three days, identifying 150 automation opportunities in multiple processes with tasks such as secondary claims research, multifactor case, clinical guidelines, and Citrix login. The algorithm analyzed the events to measure the number of agents who performed the particular process, its recurrence, and time taken. The team compared the results to the agent training guidelines, then Alorica prioritized with the client 11 automation initiatives that delivered the highest time and productivity savings and ROI and CX improvement.
Alorica ran the discovery bots on both for the client’s voice and non-voice work. For example, in the client’s back-office, the discovery bots analyzes processes for manual data entry from image and PDF claims received over emails and e-fax. Using third-party RPA & AI partner tools, Alorica created bots to capture and auto-populate fields regardless of the screen or source format while blocking certain data or document areas to meet client or compliance requirements such as HIPAA. It also has exception notifications to inform live agents to validate. Another benefit is that these bots runs 24/7. The automation deployed in early 2019 delivered ~70% headcount reduction of the process headcount who translated the data from the claims to the systems.
The company is now targeting other existing healthcare clients and showcasing their India DCOE as a ‘digital workbench’ for back-office processes, leading with automation. For automation opportunities in voice processes, Alorica is working with several travel, retail, BFSI, media and telecom brands.
Automation in the process trenches
The fundamental challenge of which tasks to automate has many dimensions such as high volume, time-consuming processes, sizable FTE involvement, high error rate, or simply, risk aversion by internal IT teams. To address the latter issue, Alorica automation deployments so far are on its own desktops without compromising, changing or modifying core client systems. The company absorbs the investment cost for running the discovery bots to set up a meaningful discussion with the client’s process owners on the possible cost-benefit and ROI.
A less discussed problem with RPA projects is the impermanent state of business processes with dynamic changes of internal rules, product/services, and system environments. The shelf-life of a bot will be quite short and its ROI smaller without human supervisors who in turn need to be prioritized and allocated based on business decisions. The automated desktop analytics model allows the deep learning algorithms to solve at least the first part of the challenge of identifying process impact. To mitigate automation program redesigns, Alorica’s solution is to focus on mature system infrastructure with a steady release roadmap.
Another challenge to automation is the incentives for the provider. Through the achieved cost optimization and improved performance, Alorica targets a bigger share of the client wallet, looking to take over higher value processes, and switching to performance-based pricing models. For the above-mentioned healthcare client, Alorica structured the pricing by processed claim, decreasing the overall cost by ~50%.
The value of partnering for automation
While broad statements for full automation attract media attention, the reality for most companies is on learning. Automation is not likely to change a process but improve it and increase customer satisfaction. Partnering with an outsourcing provider can deliver faster and better discovery in an ideal scenario, where the vendor has the process and domain expertise and quantitative discovery tools. The next step is to expand automation across both captive and supplier network centers to maximize the benefits.