How Enterprise Search Will Fundamentally Change the Way Companies Work

Software-as-a-service (SaaS) tools allow companies to access software through the cloud on an as-needed basis. The scope and use of these tools have ballooned over the past 25 years, especially in the past 15, as companies increasingly turn to SaaS to help them increase productivity with smaller teams and lower overhead costs. As enterprise software shifted from on-prem (where highly secure hardware was deployed at a company’s physical data center) to the cloud (where the software is hosted by cloud providers like Google or AWS), it’s enabled businesses to access a larger quantity and higher quality of “consumerized” SaaS tools, including Slack, Dropbox, GSuite, Github, and more. As the number of search tools proliferates, knowledge workers find it more difficult to locate information from within the multitude of SaaS tools they might use at work. A study by McKinsey & Co shows that the average knowledge worker spends nearly 20 percent of the workweek searching for information. A good enterprise search engine can tackle this problem by making knowledge across all SaaS apps accessible in one place, saving valuable time for both employees and businesses.

Any company with a significant number of knowledge workers stands to benefit from implementing enterprise search, yet not all enterprise searches are created equal. A robust, well-functioning enterprise search engine understands a company’s use of language, maintains the permissions of its source SaaS apps, crawls entire documents (instead of just titles, for example), and maintains a quickly updating index. As artificial intelligence (AI) sees a resurgence with generative AI models like large language models, it’s important for enterprise search engines to leverage cutting-edge research to make enterprise search even more powerful or risk being eclipsed by competitors.

Evolution of SaaS tools in the workplace

SaaS tools have always existed, but were once quite expensive, placing them out of the reach of many businesses. This changed with the rise of cloud-based computing. Large companies including Google, Amazon, and Microsoft began offering off-the-shelf, pay-as-you-go access to hardware through the cloud at a low price. Before the advent of the cloud, businesses would need to buy expensive hardware and set up data centers themselves, a process that required money and expertise they didn’t necessarily have. On-prem enterprise software was also slow to scale as the need for hardware increased. With the cloud, companies with just software knowledge can rent hardware and deploy and scale their apps on the cloud. This allows more companies to build higher-quality apps more quickly.

Newer SaaS tools also tend to be “consumerized,” or clean-looking and easy to use. This shift was driven by younger employees who grew up using technology and are accustomed to fast, functional apps with good user interfaces (UI). For example, while old-school messaging apps looked archaic, Slack builds mobile-first technology which resembles and exceeds the functionality of consumer equivalents like WhatsApp and Messenger.

These shifts created an explosion in use and today, many companies use hundreds of SaaS tools, making proper management more important than ever. It is crucial to understand which SaaS tools best serve a company’s needs, and to monitor and keep careful track of these tools and their functions.

Want More Tech News? Subscribe to ComputingEdge Newsletter Today!

How enterprise search tools work

Enterprise search, which allows employees to find information throughout the organization, is an example of both a SaaS tool and an important part of SaaS tool management. Enterprise search tools can be broadly divided into external and internal categories. Internal search tools can be further subdivided into those that are for personal use and those that serve the company as a whole.

The basic elements of enterprise search are similar to those of a commercial search engine such as Google. Enterprise search tools crawl pages, though they use an application programming interface (API) to access the company’s documents instead of crawling the public web. They then index these pages, retrieve results relating to user queries that are permissible to be seen by the user, and rank the results in order of relevance. Ranking is refined through data collected about user activity. UI is important to enterprise search tools, especially in an age of consumerization. Each user should also have unique permissions, defining which pages they are able to access. Some of the potential benefits of enterprise search tools include:

• They reduce the time knowledge workers spend looking for information. Enterprise search provides a significant boost to a company’s productivity and innovation. Employees can easily find the answers they previously asked coworkers for. For example, if an employee heard an acronym in a presentation and was embarrassed to ask about it, the employee can easily search for it to find the full term and view all the documents in which the term appears.
• They improve security and governance. Enterprise search tools enable centralized management of a company’s knowledge resources.
• They increase transparency and trust across departments. Knowledge workers often cite feeling out of the loop about what’s happening in the company as a reason for dissatisfaction. Enabling everyone in the company to see documents from throughout the organization can help to increase trust and approval of management.
• They reduce data sprawl and redundancy. If employees can quickly search all existing documents, they are much less likely to create duplicate resources.

Challenges and necessary considerations to implementing enterprise search tools

The implementation of enterprise search is not without potential pitfalls. To ensure the full advantages of a strong enterprise search tool, systems professionals can prioritize the following:

• Robust, thorough searching and indexing. It is important for the tool to search entire documents, not merely their titles, and to have access to all of the documents in the enterprise, and the ability to add new documents as they are created.
• Fluency in the company’s language. The tool needs to be familiar with any field-specific jargon and unusual acronyms.
• Ranking. This can present a larger challenge in enterprise search than in commercial search because enterprise search engines do not have access to massive amounts of data with which to refine their ranking algorithms. Instead, the signals used to rank documents in enterprise search are based on a deep understanding of the user, including their role, current projects, employees they work with, and managers they report to. Enterprise search ranking requires deep, fundamental natural language understanding of the organization’s unique concepts, something most general natural language processing software isn’t equipped to do out of the box.
• Customization and user-specific permissions. It’s important for the tool to be tailored to suit each person or team’s needs and that the information remains secure.
• Distribution and UI. It’s not enough to just have a new search portal—it’s also important to meet users where they are. This includes ensuring they can search from search bars on other tools or tag a chatbot in the company’s chat tool.

Search result synthesis with large language models

Commercial search engines like Bing are experimenting with using large language models (LLMs) to generate answers to user queries directly. Similarly, language models have the potential to transform enterprise search—yet there are unique challenges LLMs sometimes hallucinate, or state false information as if it is fact, and the cost of giving wrong answers in a workplace environment is high. It is difficult to train a language model on the limited amount of click feedback data generated by an enterprise search. The costs associated with LLMs are significant, and in an enterprise search context, language models often have issues with leaking permissions.

Regardless, many companies, including Glean, are working on addressing these problems and creating an LLM-based enterprise search. By presenting a tool that can provide employees with instant, comprehensive answers to their questions, the introduction of language models into enterprise search will fundamentally and permanently change the workplace. For example, instead of having to scour through hundreds of results, a user can receive a summarized text answer within seconds.

The importance of enterprise search tools is often overlooked, as many have tried and failed to develop a well-functioning product in this space. There are many aspects of a robust enterprise search engine and getting just one of them wrong can spell trouble. A functional enterprise search tool must index all of a company’s content in a permissions-sensitive way and be customized to individual users and teams. The good news is, in the midst of an explosion of SaaS products, enterprise search tools have the potential to boost productivity by as much as 20 percent by slicing the 10 hours a week knowledge workers spend looking for information. With the recent advances in AI, these tools are poised to further transform what a workplace could look like, paving the way for a future in which employees never again have to be embarrassed to ask a question at work.

About the Author

Debarghya (Deedy) Das is on the founding team of Glean, a unicorn enterprise search startup, where he leads several engineering teams in search and intelligence. Prior to that, he built search products for Google across New York, Tel Aviv, and Bangalore. He writes about technology and the tech industry on his blog and Twitter and has been featured in global news publications. He is also an independent tech consultant and active angel investor. For further information, contact dd367@cornell.edu.