Analytics Section of INFORMS NewsStanford team wins Syngenta Crop Challenge
Syngenta and the Analytics Society of INFORMS named Xiaocheng Li, Huaiyang Zhong and associate professors David Lobell and Stefano Ermon – a team from Stanford University – as the winners of the inaugural Syngenta Crop Challenge in Analytics. The team was awarded a $5,000 prize for their entry, “Hierarchy modeling of soybean variety yield and decision making for future planting plan,” which modeled a system for predicting soybean seed variety selection.Read More
Analytics Section of INFORMS NewsInnovative Applications in Analytics Award
The Innovative Applications in Analytics Award (IAAA), which recognizes creative and unique developments, applications or combinations of analytical techniques used in practice, has garnered significant interest in its brief life. The IAAA for 2016 was awarded to an MIT-led team for its submission entitled, “An Analytics Approach to the Clock Drawing Test for Cognitive Impairment,” during the recent INFORMS Conference on Analytics & Operations Research in Orlando. Fla.Read More
Analytics Section of INFORMS NewsWelcome to the Analytics Society!
The Analytics Section has become INFORMS’ newest society, and we are now the Analytics Society of INFORMS. Achieving society status recognizes our significantly increased membership base and our vastly expanded scope of activities, all of which were on display at the 2016 INFORMS Conference on Business Analytics & Operations Research in Orlando, Fla. We are still the fastest growing subdivision of the Institute!Read More
Predictive analytics in field service
Practical ways to drive field service, looking forward.
Susan is having a hectic Thursday at work. Busy days are nothing new for this 20-year veteran of the financial services industry. Susan works as the director of operations at the headquarters of Excellence Bank. A regional retail bank with 25 locations, Excellence Bank caters to consumers and small businesses in the state in which it was founded 75 years ago.
Susan has seen economic ups and downs, but the current recession is persisting far longer than she and her colleagues expected. On her way back from a quick lunch, she checks her smartphone and discovers that a technician from the bank’s server vendor is scheduled to show up at 3 p.m. for a repair. The vendor’s helpdesk failed to resolve the slow performance of this server – one of Susan’s direct reports has been on the phone with the vendor’s call center twice already – and has scheduled an onsite technician to address the issue.
Joe shows up at Susan’s site 10 minutes before the scheduled time – an experienced technician, he works for one of the local outsourcers Susan’s server vendor uses for onsite repairs. This has also been a busy Thursday for Joe. Susan is Joe’s fourth appointment of the day, with one more to go before he calls it a day. Joe came prepared and his appointment at Excellence Bank HQ went smoothly. Joe quickly confirmed, in line with what the second call center agent anticipated, that one faulty part was causing the performance issue with the server. He was carrying the replacement part and was able to restore the system to its normal performance level within half an hour, before leaving for his next appointment. The next day, Susan received an e-mail survey from the vendor, requesting feedback on Joe’s service visit and other related topics. She waited until Wednesday to ensure the continued performance of the server Joe fixed, then favorably completed the survey.
Can predictive analytics help Susan, Joe, their respective companies and other parties in this ecosystem? That’s the premise here. This article is the second of a two-part series highlighting some of the most practical and effective applications of predictive analytics in after-sales service. Part 1 focused on customer service; this article will focus on field service.
Simply put, advances in predictive analytics go further today to identify and quantify not just “what will happen and when,” but also ”why” by synergistically combining available datasets with domain knowledge and mathematical sciences. Leading researchers are even coupling predictive analytics with
decision management to provide predictions along with associated decisions – i.e., practical and timely action plans to take advantage of the predictions – and calling it predictive decision management™ or prescriptive analytics.
Customer service is the after-sales service provided by a company through contact centers. This service can be for technical or non-technical issues that a customer is experiencing with a product or service. Contact, usually inbound (i.e., a customer contacting a company), can be made by phone, e-mail, chat and the Web.
Field service is the after-sales service provided by sending an expert to a customer’s site to resolve an issue, technical or otherwise, that the customer is experiencing. This usually takes place after the customer has unsuccessfully attempted to resolve the issue with customer service, at which point field service is engaged to address the issue.
Understanding Field Service
The scenario described above involving Susan and Joe is typical in field service (see Figure 1).
Figure 1: Typical field service scenario.
For large computer hardware vendors – Dell, HP, Acer and others – it may take six or more functions to execute a successful field service business process: Customer service (contact centers), IT, service logistics, parts fulfillment, carriers and service providers.
Measuring services performance can be far more difficult than doing so for manufacturing. Underperforming service providers are quick to point out that services involve intangibles, and hence can’t be accurately measured and managed. The McKinsey article, “Measuring Performance In Services,” tackled this topic head on, listing best practices through which services executives can implement rigorous metrics to reduce variance and improve productivity. A follow-on article from McKinsey, “Improving Field Service Productivity,” highlights how real-time technologies and corresponding process adaptations are helping companies rein in “invisible” field service employees and further improve productivity.
This article focuses on metrics in the critical areas of customer satisfaction, customer loyalty and customer problem resolution. These metrics deal with effectiveness, not efficiency of field service, and thus are of significance to the company’s top- and bottom-line. We will also address one of the most critical cost metrics, cost of parts dispatch, in field service.
Customer: Solve my issue
Field technicians show up at a customer’s site to resolve an issue the customer is having with a product the customer purchased from the vendor. Three common metrics (also known as key performance indicators, KPIs), generally measured through customer surveys, are used to learn about the resolution of a customer issue:
- Resolve in one (Ri1) – Was the problem resolved in a single service visit?
- Resolve within two (Ri2) – Was the problem resolved within two service visits?
- Problem resolution (XPR) – Was the problem resolved?
These three metrics depend on many factors that come together to constitute a field service visit. Obviously, the preferred outcome is “resolve in one.” Anything a company can do to improve Ri1 will result in significantly higher satisfaction and lower costs.
Figure 2 shows an example of an advanced interactive simulation, originally developed for a large computer hardware vendor, showing the effect of several factors on XPR, Ri1 and Ri2. The factors displayed in this simulation are self-explanatory. It is also important to keep in mind that some of these factors, i.e., influencers, are interdependent. The factors are also displayed, on the screenshot, in their order of importance with respect to influence they exert on XPR, Ri1, and Ri2. [Note: All screenshots in this article have been modified to protect confidential information.]
Figure 2: Advanced interactive simulation, showing the effect of several factors on XPR, Ri1 and Ri2.
Enter predictive analytics. For a field service, it can help predict:
- specific issues customers will have with a particular product, and when they are likely to have these issues, far in advance of these issues occurring;
- XPR, Ri, and Ri2 – percentage values per month, per quarter – even before customers encounter issues for which they will contact the company for resolution;
- XPR, Ri1 and Ri2 – percentage values – per product per issue per technician per time horizon (of interest);
- optimum length of appointment to achieve Ri1, per product per issue per technician per time horizon; and
- a range of insightful custom predictions specific to each scenario of interest.
Customer: Make me happy, earn my loyalty
Customer satisfaction and customer loyalty are the pillars of any successful business. Your customers’ perception, accurate or otherwise, is your reality. Monitoring, interpreting and, most importantly, acting on customer feedback is critical to the success of any field service operation. Common metrics that companies use to measure (generally through customer surveys) customer satisfaction and loyalty include:
- Customer satisfaction: How satisfied are you with the service event that occurred on Oct. 15, 2010? Please rate from “not at all satisfied” to “extremely satisfied” on a scale (examples of scales: 1 through 9, 1 through 5, etc.)
- Tech support satisfaction: How satisfied are you with your experience with the company’s contact centers? Please rate from “not at all satisfied” to “extremely satisfied” on a scale.
- Company (also known as vendor or brand) satisfaction: How satisfied are you with the company (not just your recent support experience)? Please rate from “not at all satisfied” to “extremely satisfied” on a scale.
- Recommend likelihood (usually measured through net promoter score): How likely are you to recommend this company to a friend or colleague? Please rate from “extremely unlikely: to “extremely likely” on a scale.
- Repurchase likelihood: How likely are you to repurchase from this company? Please rate from “extremely unlikely” to “extremely likely” on a scale.
Figure 3 shows a screenshot of an advanced interactive simulation that captures how the major influencers, organized in order of their influence, affect the metrics in field service for a large computer hardware vendor.
Figure 3: Interactive simulation captures how major influencers affect metrics in field service for a large computer hardware vendor.
Enter predictive analytics, which can provide critical predictions for field service:
- each metric in customer satisfaction and customer loyalty – percentage values per month and per quarter – even before customers encounter issues for which they will contact the company for resolution;
- satisfaction and loyalty performance, per product per issue per technician per time horizon;
- dependency between Ri1 and customer satisfaction, per product per issue per technician per time horizon;
- optimum length of appointment to maximize customer satisfaction and loyalty, per product per issue per technician per time horizon; and
- a range of insightful custom predictions specific to each scenario of interest.
Company: Effective Field technicians
People are the key to running a successful and effective field service operation, and the field service technician is the “face” of the company to the customer.
How a company uses its field service technicians, optimizes their individual efficiency and manages the team’s overall performance are all crucial factors. Some companies include questions in the customer survey to learn more about their customers’ perception of their field technicians. Common metrics in this category include on-site technician’s timely arrival to customer site, professionalism, expertise in resolving customer’s issue and effectiveness in updating customer about the status of the issue before leaving customer site (see Figure 4).
Figure 4: Metrics in an effective field service operation include timely arrival, professionalism and expertise.
Once again, enter predictive analytics, which can deliver the following field service predictions:
- technician metrics – percentage values per month and per quarter – even before customers encounter issues for which they will contact the company for resolution;
- technician expertise, per product per issue per technician per time horizon;
- technician service time, per product per issue per technician per time horizon; and
- a range of insightful custom predictions specific to each scenario of interest
Company: Parts management is paramount
Multiple touches with a customer throughout a service event impact overall costs: initial contact, service parts, field engineers, event monitoring and managing, etc. However, one of the most critical (and most expensive) elements is the physical part that may be required to successfully complete the onsite repair. Even within the part costs, multiple elements must be managed and optimized (such as logistics costs, procurement costs, inventory costs). All of these elements together make up the overall cost per dispatch (CPD).
Consistently getting the right part to the right site at the right time is a difficult endeavor. Some vendors mandate that their parts suppliers maintain a massive inventory of slow-moving, long-storage parts. Others might require high parts turnover rates or frequent new product introductions from their parts suppliers. Effectively managing these inventories can be challenging.
The beauty of predictive analytics is that it can be deployed to evaluate overall costs as well as individual elements for the complex service parts ecosystems. Recently, a large computer hardware vendor used a predictive, interactive simulator (see Figure 5) to evaluate its service parts network that consisted of multiple service delivery models, multiple regions, multiple product lines and hundreds of thousands of individual parts. This robust simulator was able to illustrate the changes in CPD by varying a range of elements.
Figure 5: Large computer hardware vendor uses a predictive, interactive simulator to evaluate its service parts network.
For example, how much would the CPD go up by increasing the number of service parts in inventory? Or, what will be the change if the Latin America business grows by 10 percent? There are multiple inventory management and network optimization tools that can help model costs, but predictive analytics can bring it all together, for the first time, making it easy to see a complex set of levers being manipulated at once.
Predictive analytics can also help effectively manage inventories by predicting demand for short-term, new and life cycle parts and thus lower costs and drive customer satisfaction. By prioritizing issues and activities based on anticipated customer needs rather than simply maintaining inventory levels, a field service organization can favorably drive its satisfaction and cost metrics. Creating self-sustaining processes and systems that support the predictable flow of parts from a parts supplier to a vendor warehouse to a field technician to a customer site, a business can wrestle back control of inventories and the entire end-to-end parts process.
As with any process, the inclusion of relevant, real-time data dramatically improves prediction accuracy. For example, entry enabled by smartphone field units can improve visibility, tracking and execution across the enterprise – sales, marketing and finance departments can all get updates on inventory changes as they happen. This immediate updating of crucial parts logistics data can feed predictive models the inputs needed to return vital, immediate and actionable results.
Remember the fundamentals. Predictive analytics is not a panacea; the fundamentals of field service operations have to be addressed first. Examples include consistent service level agreements, streamlined and standardized business processes, rigorous data collection (data quality, timeliness, etc.) and strong management involvement. However, with these best practices in place, businesses can enjoy huge benefits by deploying Predictive Analytics in field service.
It is summer 2012 and Susan is now a senior director of business operations at Dominant Bank, a large multipurpose bank with hundreds of locations across the nation. The recession is finally a past memory. She took this job last year and moved to a local branch office; after two decades with smaller financial services outfits, it was time for a change. Susan’s new job, and the promotion that came with it, didn’t make her days any less hectic, however.
A VP from one of her server vendors is scheduled to show up this afternoon. Susan and her team didn’t have any recent problems with this vendor’s servers, so she was a bit surprised when she learned from her assistant that the vendor reached out and scheduled this meeting two weeks earlier.
David, VP of support operations at the server vendor, arrives on time with a couple of his deputies. Recognizing that Susan is a bit confused about the purpose of this meeting, David quickly gets to the point. David’s company wants to offer Susan complimentary hardware and service to preemptively replace a few servers in her branch’s back office. David believes these particular servers may begin to experience issues as soon as two weeks, and David’s company wants to ensure that Susan’s branch doesn’t experience any disruption.
Bewildered but reassured, Susan appreciates the offer and accepts it – she can’t afford a disruption in her back office. Next week Susan is going to meet her peers from other local branches to evaluate hardware and software vendors that can support Dominant Bank’s continued expansion in the region. She e-mails David before this meeting and asks him to get on her calendar again in two weeks.
Scott Brown (email@example.com) is a senior manager at Dell with 20 years of experience in end-to-end service delivery, including tech support, field services, service parts and logistics with Dell, Accenture, Exxon and the U.S. Army. He has consulted on supply chain strategies with Qwest, HP, AT&T and Motorola. Atanu Basu (firstname.lastname@example.org) is the CEO of DataInfoCom, an analytics software company headquartered in Austin, Texas. He has 17 years of experience in the semiconductor and the software industry. Dell and Microsoft are among DataInfoCom’s reference, and repeat, customers. DataInfoCom created Predictive Decision Management™ technology and has won an “emerging technology” investment award from the Texas Governor’s office for it. Tim Worth (email@example.com) is the senior manager of delivery operations at DataInfoCom. He has 20 years of frontline service industry experience with Dell (16 years in service analytics), Sallie Mae and Aditya Birla Minacs.