BI Trends

How to Implement Business Intelligence: Key Steps And Best Practices

By Logi Analytics
Share on LinkedIn Tweet about this on Twitter Share on Facebook

Implementing business intelligence requires considerable planning and cross-functional coordination. Whether you’re implementing an embedded business intelligence (BI) solution to package as part of your software application or implementing a standalone solution to use in house, it helps to know what you’re getting into so you can plan for each phase of the effort.

In this article, we break down how to implement both embedded and standalone BI solutions.

What is BI Implementation?

BI implementation is the process of putting a business intelligence strategy into effect. Authorities differ on whether BI implementation includes the creation of that strategy or begins once the strategy is ready to be applied. For the sake of this article, we’ll err on the side of thoroughness and say it includes strategic planning.

Implementing Embedded BI vs Standalone BI

Before we get into the processes, it’s important to understand the key differences between embedded and standalone BI.

Embedded BI delivers analytics capabilities within the context of a business application. Large enterprises that build proprietary applications for their own employees might supplement these with embedded BI, for instance. Software-as-a-service (SaaS) providers also commonly implement embedded BI in order to provide their end users more value.

To implement embedded BI, the host application and analytics application must be integrated in some way. Logi Analytics integrates with the host using one of two open APIs (.NET and REST) in conjunction with one or more embedding methods (iframes or JavaScript). This deep integration is one of the biggest differences between embedded and standalone implementations.

Unlike embedded BI solutions, standalone or “traditional” business intelligence applications are self-contained. Users may access data ported in from other business applications, but they must log in to the BI solution separately and build their analyses there. Organizations use standalone BI internally to blend data from a variety of sources for a 360-degree view of their business.

Because these two types of BI serve such different use cases, we will be sure to indicate where their implementation processes may differ.

How To Implement Business Intelligence

Step 1: Form a Committee

BI implementations require time and internal resources. According to the Business Application Research Center (BARC), the average BI implementation can take anywhere from 3 to 7.2 months, depending on the implementing company and type of BI application.

Having the right people on the project from the start will help shorten your implementation timeline because it will reduce the amount of time spent bringing new members up to speed. It will also help preserve consensus, as high turnover in a group often causes initiates to shift.

Who you include on the team will depend on the size of your company and whether you’re in the market for an embedded BI solution or a standalone solution. Larger organizations will likely have more specialists on staff, so instead of recruiting a single database administrator (DBA) to the committee, they may have a DBA, a data quality analyst, and a BI infrastructure architect. So we’re going to make recommendations based on role rather than title.

As long as each of these roles is filled by one or more individuals, the team will be equipped to see the project through from start to finish. Its first task is to establish project goals.

Step 2: Identify Success Factors

Don’t skip this step! Without a clear understanding of what you’re seeking to accomplish, it will be difficult to focus your efforts and establish requirements.

Here are some tips and best practices for identifying and managing BI success factors:

  • Select success factors uniquely suited to your use case. No two organizations will have the same set of BI goals, so make sure you customize them to fit your situation.
  • Consider who the solution is for. When you factor in all stakeholders, you may discover you have multiple sets of success factors. What qualifies as success for the c-suite may look very different from success for the BI team supplying reports.
  • Make sure your success factors are measurable. If your objectives aren’t measurable, it will be much harder to determine whether or not you’ve reached them. It also helps to establish a baseline, if possible.
  • Decide when you will revisit and redefine your success factors. Determine ahead of time when you will review your progress towards these goals and select new ones.

Picking success factors might involve some user research. Take this opportunity to lay the groundwork for the next step, establishing requirements.

Step 3: Establish Requirements

There are three classes of BI requirements to consider: user, technical, and business.

User requirements reflect the needs of everyone who would be interfacing with the application: users, report authors, dashboard builders, analysts, application administrators, data stewards — everyone. Organizations deploying an embedded BI solution will need to draw from market research for this step whereas standalone buyers will need only internal research.

Next, you’ll need to take stock of your technical constraints. What data sources will need to be supported, for example? For embedded buyers, what application integration methods will you need? Consider what your existing software infrastructure looks like and what it will take to add BI to the mix.

Lastly, consider the business context. This goes beyond financial constraints to include staffing and timeline. How quickly are you looking to deploy, and how many people will it take to maintain the implementation? What will the total cost of ownership be, and what kinds of returns would make the purchase worthwhile?

A clear set of requirements in each of these three categories will make evaluating prospective BI solutions much easier.

Step 4: Evaluate BI Solutions

It’s generally a good idea to narrow your BI platform options down to a short list before testing each one against a proof of concept project. Make sure each candidate meets all non-negotiable requirements before you entertain a formal evaluation. Take full advantage of product demonstrations to ask questions that will help you either disqualify or advance a solution through the process.

Embedded BI buyers are welcome to use our Assessing Product Fit guide as a starting point; but again, organizations will want to supplement with their own requirements. Standalone buyers can use the guide as a springboard as well but will want to omit any categories that do not apply.

When you’ve identified the candidate that best fits your needs, it’s time to transition into the fifth and final step!

Step 5: Deploy

Each BI deployment is unique. How your team deploys will depend on the BI solution they’ve chosen, the state of their business data, how the solution will be hosted, who will have access to it, what those users need to get started, and how the deployment will be maintained over time.

The steps below describe what deployment might look like for an embedded BI implementer. Our team uses this generic process to build customized deployment plans for each client.

  1. Planning
  2. Permissioning
  3. Environment Configuration
  4. Storage Management
  5. Data Source Configuration
  6. Integration
  7. Report Building and Staff Training
  8. Deployment
  9. Post-Deployment
  10. Optional Features

There are nevertheless five things every BI deployment must address, no matter the context:

  1. Data Preparation: Before integrating with a data source, it is important to assess its quality, integrity, and structure. “Garbage in, garbage out,” as the saying goes. Your BI implementation will fail before it’s even begun if the data isn’t primed for analysis.
  2. Data Integration: Connecting your BI solution to the data sources users will analyze is a universal deployment step.
  3. Testing: It is always a good idea to QA test your BI implementation in a staging environment before making it generally available.
  4. Training: Users will need to be trained in the software upon deployment. Not only that, but anyone responsible for supporting users and/or the system will require training as well. Make sure all necessary educational materials — manuals, documentation, courses, tutorials, etc. — are developed and ready for use.
  5. Ongoing Support and Maintenance: All BI deployments must include process planning for ongoing system support and maintenance. This includes:
    1. Who users should contact for technical support and how.
    2. How issue escalations will be handled.
    3. Who will be responsible for data stewardship.
    4. Who will oversee user training.
    5. Who will administer the BI implementation.

Taking care at each of the five implementation steps above will help ensure your BI project’s success. But what might help most of all is not thinking of it as a project in the first place.

Jorge García, principal analyst of business intelligence (BI) and data management at TEC, says he recommends thinking of BI projects as initiatives because they don’t end upon deployment. They’re permanent.

“The systems evolve,” he explains. “We evolve with the systems, and the employee is evolving with the system. So we all need to look for that evolution and try to adjust consequently.”

Originally published March 4, 2020; updated on January 4th, 2022

About the Author

Logi Analytics is the leader in embedded analytics. We help team put business intelligence at the core of their organizations and products.