This whitepaper is for business users seeking a greater understanding of the process for implementing search within their organization. Includes 10 steps to follow and a strong ROI business case.

Introduction

Given the ubiquitous nature of the search box on a standard webpage, one might be surprised by how little is actually known about the mechanics of sifting through data and what separates good search from bad. While it is rare to find any internet or intranet site that doesn't have at least a rudimentary search function, the quality of results returned by these simple boxes varies widely.

Looking at the internet, public search engines like Google, Bing and Yahoo have created expectations for how search should work. Employees take these expectations with them into their enterprise-level searches in the workplace. they're often frustrated.

Enterprise search is different than crawling the internet for content. Information resides in a wide variety of locations (shared drives, individual hard drives, document management systems, intranets etc.) and a wide variety of formats (documents, images, emails, charts and tables etc.).

The challenge, of course, is that a good enterprise search tool requires resources - financial and human - to set up and configure. Enterprise search, like office email or a telephone system, in an enabling technology; there isn't the same direct correlation to business objectives and ROI that is traditionally available (such as measuring the revenues from a direct mail campaign or the tangible benefits of a new accounting system). Where's the ROI in finding information quickly? While there are ways to measure the savings associated with knowledge workers spending less time searching for information (see Appendix for more on this) the measurements are less than perfect.

Does this mean you don't need to create a business case for enterprise search? Of course not. It does mean, however, that you can't rely entirely on a formal ROI justification for the investment. So how do you build a business case? First you must determine how search is currently used in your organization and where knowledge and information reside. Then identify which search scenarios promise the most value for the smallest investment. 

This paper

Step 1: Define the knowledge landscape

All enterprise search is, of course, not created equal. Therefore, building your business case should focus on directly addressing your most pressing business need. In order to determine what that need is, you must first explore where your organization's knowledge exists and how it's currently being searched. Figure our where key business information resides in your organization and identify the gap between current search deployments and where the real knowledge assets exist. How do you do this? For each key repository of corporate information, first determine if the content can be searched today and, if so, how effective the search is and how  many users are taking advantage of it. This information can help you identify where search can drive the most value.

At this point, it is probably useful to reach out to colleagues in other departments, such as IT, but doesn't belabor your investigation. This is intended to primarily narrow your field of options quickly. Eliminate most and focus on the promising ones. While each company will have a different knowledge landscape, this table should provide a useful starting point. We've included a few examples in each section to help you frame your own approach.

Define your landscape

Step 2: Explore most promising search scenarios

Once you've evaluated the existing search and knowledge landscape, the next step is to take a most detailed look at the most promising search scenarios. In general, the most promising opportunities can be grouped, often along lines like this:

  • Intranet and document management
  • Records management
  • Extranet, public site
  • Retrieving external (third-party) data
  • Specialty search
  • Supporting e-discovery
  • Surfacing line of business or enterprise applications

While the best-case scenario would see an enterprise search solution that encompasses all of these data sources, this is probably not realistic for most organizations. Instead, pick the scenario that has the ability to address the most important information needs for the most people as quickly as possible.

How do you do this? It's time to talk to people. While every organization is different, in general you'll be looking for the people in IT and the people who do the most information searching. Ideally you should include a minimum of one representative from each functional department in your organization. This will help shed light on real world requirements and minimize assumptions regarding each group's actual needs. 

Questions for the IT department

While this is certainly not an exhaustive list, the following questions will give you a better idea for work done to date and, hopefully, validate the findings gleaned from you earlier efforts. It also brings IT into the discussions at a fairly early stage which will be important down the road towards implementation.

  • What search solutions are currently in place?
  • Has anyone spent time tuning this system?
  • Does our licensing support additional data sources / end users?
  • How challenging will it be to implement security trimming for specific data sources?

Questions for the searchers

This is harder to define, of course, because it depends entirely on which search scenario you've chosen to take on. In keeping with our previous example of focusing on intranet / document management, the following cluster of questions could be useful and insightful:

Today, how do you (task questions):

  • Find contact information (known person)?
  • Locate expertise (unknown person)?
  • Find policies?
  • Find procedures?
  • Find current / past project information?
  • Find records?

Open questions:

  • How well do others understand how to find this information?
  • How easily are you able to find this information?
  • How sure are you that you've found accurate or timely information?
  • How confident are you that you've found all of the relevant information?
  • What is the biggest challenge with the current approach?
  • If you could change one thing about the way the current system operates, what would it be?

Closed questions:

  • Are you usually seeking specific information that you know exists? Or are you searching for information related to a general query or issue?
  • Do you make user of acronyms or ID numbers? Does this make it easier or more difficult to find information?
  • Are there specific documents, records or systems that you use all the time or are particularly critical for helping you find the information you need?
  • How often do you think you find the information you need with your first search?
  • How much of your day is spent searching for information?
  • How much time do you think you could save if the system allowed you to consistently find what you are looking for on the first or second try?
  • Are you familiar with the term Boolean search?
Question

Step 3: Define the benefits

Key to getting buy-in for any project, of course, is the definition of the benefits. Demonstrating tangible results will help illustrate the merits of any project which is ultimately what every business is looking for if they are going to invest resources. As discussed earlier, measuring the benefits of a well tuned, focused enterprise search solution with a strict ROI calculation can be misleading. However, this does not mean there are no demonstrable benefits. 

Culture shift - Many organizations are attempting to bridge internal barriers and develop a more collaborative culture. Effective enterprise search surfaces that most relevant information, regardless of who has produced it; data produced by other divisions or "silos" in the org chart might be returned and this facilitates cross-enterprise collaboration.

Efficiency - A good enterprise search tool (or suite of tools) can dramatically cut down the time spent hunting for information, leaving more time to actually use that information in support of the business' or organization's goals.

Effectiveness - Good search not only allows you to search faster, it allows you to search smarter. While it's easy (relatively) to measure the impact of time wasted conducting searches, it's harder to determine the impact of employees finding outdated or insufficient information. A good search solution can remedy that.

Consistency - A consistent approach to search means that all employees will be accessing information efficiently and effectively; people will see the same versions of policies and procedures. The end result is an organization with a consistent understanding of corporate priorities and processes that provides the same (correct) answer to customer inquiries. 

Flexibility - Good search doesn't just crawl known information, it can hep surface new sources of knowledge quickly. It can pull in from outside sources and start to internalize that knowledge.

Continuity - One of the biggest challenges facing organizations in every industry and sector is the looming demographic crunch as baby boomers head towards retirement. Good search as part of a knowledge management strategy can ensure that old knowledge can easily be transferred to new minds. 

Authority - A robust search solution ensures that priority is placed on returning current information ahead of legacy information. While the latter may be important for institutional memory, poorly tuned search solutions can lead to confusion about which policies or information is true authority.

Risk Reduction - Inability to find corporate information is the source of several types of risk. Employees may provide the wrong answer to customers or make decisions that violate corporate policy; the legal team may not be able to locate key information to support litigation efforts; the firm may find it difficult to demonstrate compliance with regulations. Enterprise search can help improve the risk profile for most organizations. 

The challenge is figuring our which benefits are the most important to you organization. First determine which are applicable; then decide which of the search scenarios identified above would offer the most value based on those benefits. 

Risk Reduction

Step 4: Evaluate costs of deployment

With a better understanding of the existing knowledge landscapes, search scenarios and the benefits that can be derived from a custom search solution for each, the final step is to determine the cost associated with deploying such a solution.

There are a number of factors at play:

Number of data sets - Are you trying to introduce a simple people search tool for an employee directory or something more comprehensive? The more types and locations of data you need to search, the more robust the solution has to be; this will have an impact on cost.

Scale - Closely related to the previous point, scale can have a major impact on cost. How many items need to be indexed? How often does the indexing need to be updated? This drives both software and hardware costs. 

Security and permission requirements - If everyone has access to the same information, this is not much of a factor. If each person gets results trimmed based on permissions and roles, it gets more complicated and, thus, more expensive.

Custom interface - This goes beyond simply something that looks nice. A custom, faceted interface, for example, makes search and navigation easier but it requires more work to configure. Out of the box solutions are lower cost but may not give you the full benefits identified above. 

Custom relevancy - Do you want your search solution to prioritize results differently for someone in human resources versus someone in IT or on the executive team? It's possible to do but, again, typically not out of the box.

Tuning - Cost depends entirely on the time required to do things like adding in synonyms, thesauri and dictionaries to capture wayward searches; incorporating common acronyms; introducing support for several languages; and determining best bets for common terms (overriding standard algorithms to return most valuable results to your organization.)

Take these factors into account and estimate whether implementing a search solution for each scenario would involve low, medium or high costs. 

Step 5: Complete the business case grid

Business Case Graph

How you proceed from here depends entirely on your organizational culture and what can be supported. Obviously the high value solutions are optimal but if they are also the highest cost it may be difficult to attain buy-in and, more importantly, get the necessary funds to proceed. In these cases, a medium value solution with a lower cost may be a good place to start, with an eye towards higher value in a second or third phase. 

The next step

Obviously building a business case is only the first step; once you have a solid sense of the business benefit enterprise search can deliver, you have to launch into the hard part of the exercise - selecting and deploying the search technology. Selecting a shortlist of vendors, implementing a proof of concept and planning a phased roll out strategy will take time and expertise. As will monitoring the use of the search system and tuning the results based on the real-world behaviours of searchers. Feel free to drop us a note if you would like to understand the nonlinear creations methodology for making enterprise search actually deliver the business value you identified in the business case; we're always happy to chat. 

Appendix A: A "formal" ROI analysis approach

Many organizations require a formal return on investment calculation as part of the budgeting or project approval process; the following is one approach to meeting this requirement. But, as we suggest in the body of this document, it ignores the real value of enterprise search. It is the equivalent of evaluating email based on the cost of saved long distance phone call fees - not exactly wrong, but kind of missing the point.

We first walk through the data you will need and then the math you will need to perform. We then close with an example that uses real-world figures.

The IDC Baseline

A frequently cited IDC report, the Hidden Costs of Information Work (May 2009), attempts to quantify the costs associated with common information worker tasks. The survey reveals that the task of seeking or gathering information - essentially search - is the second most costly activity undertaken by information workers. The following table provides an overview of their findings. 

Cost of information tasks

n = 706
Note: Cost per worker per year is based on an annual salary of $75,000, including benefits.
Source: IDC's information Worker Productivity Survey, October and December 2008, and IDC's LinkedIn Survey, 2009.

Search Fail

Gather or estimate key data

To perform an ROI estimate, you'll need to have the following information. In most cases, you will need to make estimates or guesstimates (one of the reasons we prefer a looser business case approach.)

Cost information

  1. How many "knowledge workers" does the company employ?

  2. What is the average fully loaded cost of a "knowledge worker" in your firm? Note IDC uses $75,000 in their calculations.

  3. Percent of time knowledge workers spend searching (we usually use the IDC figure of 8.8 hours per week / 40 hour work week = 22%.)

  4. Approximately how much will it cost you to deploy enterprise search? Be sure to include:

  • Software costs
  • Hardware costs
  • Deployment costs: internal costs and outside consultant costs
  • Annual maintenance costs for first three years
  • Annual internal costs for support
  • Change management and communication costs

Benefit information

  1. How much do you anticipate improving the effectiveness of search? As noted above, the IDC suggests that 42% of time spent searching is wasted. We generally estimate that a combination of more relevant search results and a broader set of indexed documents will generate a 50% improvement. If your current search infrastructure is very poor, then use a higher figure; if it is effective, use a lower estimate. We'll call this the enterprise search lift ration.

Do the math

Step one is to determine the current cost of knowledge workers seeking information within your organization.

The formula is:

The Formula

The next is to determine the total dollar benefit that the deployed search technology will deliver.

The math is straightforward:

Math

Most organizations will amortize the cost of enterprise search over three years but check with your finance office to confirm this for your company.

Assuming a three-year amortization cycle then you'll want to calculate the benefit delivered over those three years:

Three year amortization

You need to estimate the total cost of deploying more effective search:

Total Cost

Calculating the Return on Investment

With the hard part done, it is straightforward to perform the ROI calculation:

ROI

Some organizations use payback time to evaluate potential projects. To calculate this:

Payback Time

Finally, you may want to perform some sensitivity analysis - consider reducing the enterprise search lift ration by 25% , 50% and 75% seeing how this changes the outcome.

An example calculation

It is probably easier to follow this procedure if we use a real-world example. The scenario is invented, but the figures are representative of what we have gathered in client engagements.

Cost information

Example calcuation

Benefit information

In this example, we'll assume a 50% improvement in search effectiveness.

Doing the math

Search cost today = # of knowledge workers X avg. fully loaded costs X percent of time knowledge workers spend searching

Search cost today = 200 X 85,000 X 22%

Search cost today = $3,740,000

The next is to determine the total dollar benefit that the deployed search technology will deliver.

The math is straightforward:

Potential annual benefit = Search cost today X the enterprise search lift ratio

Potential annual benefit = $3,740,000 x 50%

Potential annual benefit = $1,870,000

Three year potential benefit = Potential annual benefit X 3

Three year potential benefit = $1,870,000 X 3

Three year potential benefit = $5,610,000

You need to estimate the total cost of deploying more effective search:

Total cost = Software costs + Hardware costs + Deployment costs (internal costs + outside consultant costs) + Annual maintenance costs for first three years + Annual internal costs for support + Change management and Communication costs.

Total cost = $250,000 + $20,000 + $350,000 + $112,500 + $25,500 + $10,000

Total cost = $796,000

Return on investments

Three year ROI = Three year potential benefit / Total cost

Three year ROI = $5,610,000 / $796,000

Three year ROI = 705%

Payback time in years = Total cost / Potential annual benefit

Payback time in years = $796,000 / $1,870,000

Payback time in years = 0.43 (or slightly more than 5 months)

Sensitivity analysis

As the following table suggests, this is a good investment even when we dramatically reduce the anticipated improvements in search effectiveness. Even if search is only improved by 10%, the result is a reasonable ROI over three years.

Sensitivity Analysis


comments powered by Disqus

Get your enterprise search in order

Want your own print-ready copy of this whitepaper?

Download the PDF