When it comes to our field service operations, we’ve learned that knowledge is power. This is especially important when products become more complex and information becomes more difficult to find.
Our primary business is centered on the development, manufacture and sale of semiconductor production equipment (SPE), flat panel display (FPD) and photovoltaic cell (PV) production equipment. In an effort to support a diverse customer base, our 12,000+ employees (research & development, manufacturing, sales and service) are strategically located all over the world. However, this abundance of siloed information and resources posed quite a challenge for Tokyo Electron America (TEA), in which my division is the North American sales and field engineering arm of Tokyo Electron Limited.
With more than 500 field service engineers, we maintain a tremendous array of highly configurable and complex manufacturing equipment. In order to address complex customer cases, the TEA field engineers traditionally needed to manually identify, normalize and extract their necessary knowledge through decentralized and disparate systems, comprised of information housed in various formats with individual specifications. This led to lengthy data retrieval times, and inefficient dissemination of important product knowledge throughout the team. In fact, we estimated that during a peak month, duplicate and lost knowledge was costing the company more than $80,000, which does not include comparable impacts on field service engineer’s satisfaction levels.
Each of our 6+ SPE product lines is manufactured in separate factories in Japan and the U.S. — with each serving an industry niche. This led to nuanced and localized practices in each product’s engineering, manufacturing and service information infrastructure. As a result, our team employed a variety of small-scale solutions that could negotiate the various and localized character of primary information sources. Compounding the complexity was the multi-lingual nature of the information and the highly granular, layered permission systems governing access to information. Whenever there was a push for a unified solution, it quickly faded when no solution seemed available within our budget, organizational and geographical constraints.
Like any world-class service organization, our management team was highly aware of service performance and challenges in meeting contracted service levels. One challenge that continually rose to the surface was worst-case-scenario events occurring on advanced technology, first-of-kind tools where normal engineering or service information was not readily available. Even in these scenarios, our service team was amazingly able to achieve 95 percent compliance to service levels – but we wanted to do better.
Over several months, the management team isolated actions and parameters contributing to the 5 percent of events that exceeded time limits. Upon analysis, it was apparent that if knowledge retrieval could be improved, these worst-case scenario events could be shortened by as much as 33 percent.
However, while the problem was plain to see, the solution was not simple. The information important in these scenarios was patchy and disparate. Additionally, the amount of new knowledge concerning these target tools dramatically increased daily. In one month alone, we had nearly 200 different, complex documents produced for only one product. The requirement for 200 field service engineers to read and then file such a large number of documents contributed to the costs of maintaining high service levels. We needed to better structure our organizational assets to enhance performance and competitiveness.
The problem of the rapidly accumulating data was just the tip of the iceberg. The larger problem was identifying, normalizing and aggregating the multitude of primary and secondary knowledge sources that were highly local to return meaningful results. TEA’s digital knowledge consisted of equipment manuals, manufacturing, engineering and service information authored in a variety of formats and built to an assortment of specifications. Even when one of our engineers knew where to look for information, retrieving that information in a timely manner was a formidable challenge. In practice, groups would often assign a single field engineer with a talent for finding information as the “documentation” person who would find technical information since it required deep skills and a fair share of heuristics.
It became apparent that our service team needed a powerful knowledge management solution that would find technical information where it resided and federate it quickly and in a logical fashion. We needed a solution to speed the identification and retrieval of critical engineering and service information from a growing and highly decentralized global knowledge infrastructure.
The first attempt at implementing Thunderstone, a Google-like enterprise search tool, proved more difficult than initially imagined. We learned that the science behind this search engine technology looks really easy, but if you try to develop the capabilities on your own, it’s very difficult. We had one bright software engineer tasked with making this work, and it was a constant struggle and steep learning curve.
Our initial enterprise search solution was eventually shelved because of its inability to successfully index all of the knowledge repositories and applications. Additionally, the “commodity” search solution was not able to negotiate and respect the multiple layers of permissions set by role and user level. As a result, search results were unsecure, incomplete and took an unacceptable amount of time to return results.
The team went back to the drawing board to find the next level of search solution — a comprehensive and advanced information access solution that could extract and interpret the taxonomies and logic built into the various repositories and applications. After an intense selection process that involved several deep and detailed proof of concepts, the team unanimously selected Coveo.
By architecting a heterogeneous solution around Coveo’s unified indexing technology, the team brought online a knowledge system that was both quickly adopted by end users and immediately improved targeted goals.
For TEA, the impact of rolling up technical information from various, highly localized sources into a single, unified search has been about adoption. Our field service engineers can be short on tolerance for breaking-in and utilizing new systems, but the Coveo solution was met with resounding end-user success. The solution is simple, but casts a wide net and provides an intuitive interface to dive deep into issues while pulling consolidated, correlated knowledge from a variety of sources. Our searches can get very granular – down to the tool serial number, while aggregating the system history.
With Coveo, TEA’s engineers have a consolidated, unified view of knowledge comprised across the company’s complex product lines. This allows them to reduce repair times and improve customer retention — effectively generating a higher return on existing knowledge. Upon analysis, we’ve concluded that this knowledge retrieval improves time-to-repair incidents by 28 percent and enables the achievement of 98 percent compliance for service level agreements, which is still improving.
In a complex industry, we’ve found that harnessing our collective knowledge is the best way to reduce time-to-repair, improve efficiencies and deliver higher quality service. If your organization is facing a knowledge gap or wanting to stimulate innovation, we encourage you to invest in a search-powered knowledge management system.