Your current enterprise search engine provider announced an end-of-life date for the software. The vendor proposed a more modern engine to migrate to. You consider their offering and your organization's requirements. At that point, you may encounter one or more of the following scenarios:

  • You would go from a fixed licensing model with predictable cost calculations, to a variable pricing model.
  • You would change from a search engine packaged with all the components you need, to integrating services from the vendor’s new platform, in whichever ways are needed for your search applications.
  • You must maintain the current search engine while migrating to the new one if you decide to do so.

Your situation may vary from the above, of course. Perhaps you have a search engine from a company recently announced as being acquired (an example is the acquisition of the Attivio search engine by ServiceNow, announced in Oct 2019). Would the parent company continue the engine’s offering and its enhancements, or retire it?

Maybe you acquired an organization that has valuable digital assets residing in a different search engine; one more to add to your list of search software to manage. This increases the pressure on your organization to reduce cost and complexity by choosing one engine to power all of your custom enterprise search functionalities. By custom, I mean the search functionalities you implement and maintain, not the ones already embedded out-of-the-box.

Your enterprise search strategy in the age of AI and Cloud

Evaluating and selecting a search engine is hard, and I dare to say that it has become harder lately with the revolution fueled by cloud, analytics, and Artificial Intelligence (AI) technologies. But thanks to these technologies, modern enterprise search engines have evolved into intelligent search engines (also known as question-answering systems, cognitive search, or insight engines) that provide answers and not just lists of results.

I began writing this blog while I was, once again, helping an organization select a new search engine. They were at the first leg of the journey: charting out their path to get to where they wanted their enterprise search to be. This post is intended to provide you with practical steps to get started. 

<<< Start >>>

smarter enterprise search strategy

<<< End >>>

5 steps to define a practical enterprise search strategy

STEP 1: Where are we? Assess the situation

Chances are that you already have a modern search engine. Yet, it's not simply a lift-and-move or upgrade-to-the-latest-release kind of project because:

  • The way it was implemented prevents the search engine from getting the most out of its own capabilities;
  • As deployed, the search engine can't benefit from powerful technologies, like cloud or AI;
  • Some prior architectural decisions limit you from adding new requirements or serving new search applications; or
  • Describe your issue here…

Whatever your current situation is, think of it as a transportation need. You can't get proper driving directions or a ride-share without providing your current location. So, start by properly identifying where your search engine is at. That is the search system’s current conditions, not its server, data center, or cloud location.

STEP 2: Where are we going to? Define the destination

If you haven’t yet, this would be the right time to put on your future vision goggles. Where does your search solution need to be five or more years from now? You certainly have some shorter-term needs for search related functionality. Still, thinking about the longer term would frame your selection process to choose a search engine that would still serve your evolving needs, hopefully including those you cannot really envision today. If you need some inspiration, read this recent blog from our Innovation Lead. I’ll just summarize the key enterprise search trends that were discussed in that post:

  1. Neural networks and search engines
  2. Semantic search
  3. Document understanding
  4. Image and voice search
  5. Knowledge graphs

There are plenty of other features that you may consider in defining your search vision: cognitive search, insights and analytics, contextual search, natural language processing (NLP), etc. I suggest you move beyond the buzzwords and try to understand what each concept means within your context. The key is to look forward yet staying realistic. For instance, would you benefit from a search engine that allows you to incorporate your own AI when your organization has no current or planned AI capacities? Perhaps built-in AI search solutions or AI services partners may work better than building your own in this situation.

During this step, the maturity model below can be a guideline for assessing your situation and defining your vision for enterprise search.

<<< Start >>>

enterprise search maturity model

<<< End >>>


STEP 3: What's in it for us? Set clear expectations

By now, you should have a clear start and end of your journey. As a side effect of visualizing future search, you likely thought, and hopefully wrote down, your reasons and benefits. I’m not talking about specific requirements here. Below are some areas for consideration at this stage of the process:

  • More self-serve opportunities for end-users
  • Less infrastructure for the search engine
  • More automated management and operation
  • Increased productivity
  • Beyond better search: discovering data and generating insights

There will be time later to drill into the details of each area. Remember that we're just clearly defining expectations and purposes of your effort. If you feel lost while travelling to your destination, they will remind you why you started the trip in the first place.

STEP 4: What would be our stops along the way? Plan the journey

As I wrote this part, I got curious about the word “journey.” I learned, through a major Internet search engine of course, that the word has Latin and French origins. In both, the word referred to something that happened in a single day. A trip lasting a day. Initially, that was its use in English as well. Its modern use though, is often for something that takes longer than a day. Also, there may be stops, obstacles, or changes along the way. This is why I associate it with search applications. Just the implementation alone can take weeks, even for small, straightforward search indexes, with multiple milestones from inception to production.

Coincidentally, as I was thinking about stops in our enterprise search engine selection journey, I heard someone talk about "landmarks" instead of "milestones" in the context of a project. The word “milestone” implies measurement. You can estimate how much effort, time, and other resources it would take to get to each project milestone. We aren't there yet. Landmarks instead, represent places you want to visit while traveling but you don’t necessarily know how or when to get there. You need to figure out those details as you go and before you depart towards your next landmark.

I suppose yours is not a simple search application. You've read this far because choosing the next enterprise search engine for your needs is complicated. Even identifying the right candidate technologies to compare can be challenging. I suggest that you identify the landmarks you’d like to reach throughout the journey. For now, don’t worry about the order. Simply list them out so you can refer to them as you define things more precisely.

Here are some general examples of landmarks for changing search engines:

  • Select a search engine (open source like Elasticsearch and Solr; commercial engines like SharePoint Search, Sinequa, etc.; or cloud-based search engines like Google Cloud Search, Amazon Kendra, Azure Search, Microsoft Search, etc.)
  • Augment your search team
  • Migrate existing applications to the new search engine
  • Add more search applications
  • Enhance search with basic AI, such as content enrichment, AI-based relevance ranking and tuning, clustering, or others already included in modern search engines
  • Incorporate advanced AI, such as advanced Natural Language Understanding and Processing (NLU/NLP), your own machine learning models, integration with third-party AI libraries or services (Vision, Audio, or others), etc.

Try to define more specific landmarks than the examples above for your specific situation. And these landmarks don't necessarily have to be in this order every time. You may decide to migrate critical search applications first, implement some AI functionality for those applications, then go back to migrating another subset of the existing applications. Or you may find that you’d want to start by adding staff to your search team or joining currently separate staff members into a single, centralized search team. Doing so would free up some time from your senior search team member to work on selecting the next engine. 

STEP 5: Visualizing the journey

Time to put together everything we have to this point. We have the main elements to plot our trip on a map now:

  • Starting point
  • Landmarks
  • Destination

You probably even know what pictures to take along the way: your expectations for the journey. You may notice we don’t yet have specific directions to reach each landmark. That’s okay for now. We have a reference to guide us from beginning to end, along with the main turns and stops.

Some of the landmarks on your list may be unique to your organization. This process should help you realize that search is more than the responsibility of the search team. It should be a digital asset that will benefit many parts of your organization. Therefore, deciding which order to visit them likely requires coordination and involvement from multiple business functions. You should have also a communication tool to support that process and keep stakeholders engaged while renovating search.

The next leg of the journey: Choosing your enterprise search engine

Although search engine selection may not be next landmark to visit for you, at some point it will. That’s a part of the journey we know very well, as I’ll describe in my next blog post.

Accenture’s Search and Content Analytics can help accelerate the steps described above, while capturing input for selecting a search engine. We’ll do that by gathering the pain points, expectations, and key requirements from your multiple stakeholders and provide practical recommendations so that your organization can make an informed decision. 

Read my next blog post on how to select a search engine and connect with us to learn how we can help. Happy finding!

Carlos Maroto

Functional & Industry Analytics Manager – Accenture Applied Intelligence

Subscription Center
Subscribe to Accenture's Search and Content Analytics Blog Subscribe to Accenture's Search and Content Analytics Blog