Coveo recently conducted a survey asking approximately more than 600 IT professionals about the current state of enterprise search. A substantial 84% reported a lack of alignment across different enterprise search deployments. While this  is not uncommon as different departments like to own their search, we wanted to unpack why this happens – and the ramifications you should watch out for. 

We’ll lay out how you end up with  fragmented experiences for all users — as well as difficulty in long-term maintenance and growth.

Causes of Lack of IT Alignment

Why does this happen? It arises from departmental needs, siloed data, and a lack of IT resources. But it also prevents employees from finding data easily — or at all, our survey found. But companies are trying to address the issue, with some 53% investing in better search because of remote and hybrid work.

A graphic illustrates 84% report a lack of alignment across different enterprise search deployments.

Tons of Content, Lack of Solutions

In 1995, there were roughly 50,000 websites, each made up of up to about 10 or so pages, on the web – and even then finding a particular document on the web was problematic. In 2022, there are roughly 50 billion websites, and each of these sites may have thousands of pages. 

In other words, the typical corporate website today has more pages in it than existed on the entire internet 30 years ago.

What makes this growth particularly notable is that the types of content — and as a consequence, the audience of that content — also diverged over time. Someone in accounting is far more likely to search for invoices or sales figures, a marketing person is likely to need information specific to previous campaigns, and engineers may look for blueprints or code repositories.

This often means that different departments often will seek different solutions for search, believing (with good reason) that since their particular search needs are likely to be different, the search engines that they choose should be different, too. However, this can be expensive, both in terms of licensing and maintenance, for most moderate to large companies.

Our survey supports this, finding that investment in enterprise search is rising, particularly at big companies where 85% reported spending more on search in the last 12 months. 

A graphic illustrates 85% have increased investments in enterprise search technology in the past 12 months.

Enterprise Search Doesn’t Have To Be Hard

Increasingly, having multiple search engines is unnecessary with the advent of machine-learning based systems, such as Coveo. These incorporate metadata about the user into the search process, such as: 

  • the department the user is in; 
  • the kinds of work that they do; 
  • and their role within the organization, 

Combining this kind of profile information, behavioral data, and content returns results that are specific to what the user is looking for. 

This kind of search learns from the user in real time and is capable of learning from previous queries of similar users — meaning that it’s possible to train the search engine to a reasonably high level even before the first search inside an enterprise is made. 

Search evolves over time as more content, more users, and more departments come online. For example, when multiple people from accounting search for and select specific content, the search engine begins to show those documents to similar users who conduct comparable searches — in other words, what the engine knows about the user and what the user has done is factored into search results for similar users and searches.

This learned behavior means that there is less need for specific rules engineering, which translates to less fragility when transitioning search from department to department or replacing sunsetted search systems.

This is especially important given that 54% of survey respondents report that significant changes in search requirements capabilities was a driving force in enterprise search investment in the preceding year. 

A graphic illustrates search stakeholders report a wide range of compelling events for enterprise search investments.

Enterprise Search Talent Getting Hard to Find

Flexible machine learning search solutions also mean that you need fewer — or perhaps even no — “search engineers” tweaking the search system to optimize it for one or another kind of target audience. This is an increasingly important consideration since 95% of IT teams struggle to find enterprise search talent, according to our survey.

Machine learning options offering low-code configurability alongside full code access that enables developers to extend and customize search solutions are one way to overcome the talent gap. 

For example, low-code solutions can sit on top of a chatbot-like interface. They can use contextual clues about the searcher and the kind of questions being asked to determine the best matches for queries as well as the best mechanism for presentation of that information.

More and more data stores hold heterogeneous data, making it difficult to present to users in a uniform way. By capturing contextual information, search engines can make all data, including heterogeneous data, more accessible, especially over time. Coveo makes this possible through 50+ connectors to third-party systems, which provides the ability to query external systems transparently.

In addition, creating separate search engines tuned to different departments is likely to increase the number of data silos. This presents a problem when a user needs a document or dataset that would be within another department’s purview — with siloed data, they wouldn’t even see the documents. 

And there’s a difference between weighting documents so that those mostly likely to answer a query are on top while others are lower and not being able to show content at all. Machine language-based solutions, such as Coveo, make it easier to identify and locate those particular resources.

Bringing Departments Together Provides Relevance Everywhere

Intelligent search goes beyond simply looking for content in an index. It relies on user context and historical patterns as well as more traditional text tools to determine relevance, maintains several feature dimensions to be able to surface content that may be useful in multiple domains, and it does so transparently in real time, without the complexity inherent in manual tunable systems. 

It’s one of the best ways to align departments in getting to the data they need in the most efficient manner possible.

Interested in more insights from our enterprise search report? Download a copy.

Get the full reportReport: 4 Strategies to Overcome Obstacles and Improve Search Relevance
Share this story:

About Kurt Cagle

Kurt Cagle is the managing editor of Data Science Central, the largest community of data science practitioners on the Internet, and is publisher and editor of The Cagle Report. He is also an editor and senior ontologist for the IEEE Spatial Web specification, focusing on metaverse, digital twin, and related shared world technologies. He lives in Seattle, Washington.

Read more from this author