Imagine navigating rush hour traffic while blindfolded and in a new car. You don’t know all the controls, and you didn’t bother to ask. Disaster! It’s a reality for many businesses struggling to harness their BI tools.

In this article, we delve into the nitty-gritty of business intelligence implementation. It includes nuggets of wisdom and experience from my interview with Anurag Sanghai, Principal Solutions Architect with Intellicus Technologies.

Compare BI Software Leaders

Article Roadmap

• What Is Business Intelligence?
• What Is Business Intelligence Implementation?
• How To Implement BI
• Benefits
• Cost Considerations
• Trends
• Next Steps

A mining company in Southeast Asia was struggling with low visibility into operations due to siloed data. Metrics were slipping through the cracks as they struggled to combine information from various sources into a unified store.

Their business intelligence implementation was proving insufficient, and they needed help. We need to examine BI closely to understand how they tackled the issue.

What’s business intelligence, and why is it important?

What Is Business Intelligence?

BI is a set of technologies, tools and techniques that make data available for enterprise use. Since data analysis is at its core, you could say it’s a superset of enterprise analytics.

Anurag Sanghai, Principal Solutions Architect at Intellicus Technologies, prefers to call it data intelligence, explaining that business intelligence might not be adequate.

Because when we talk about any enterprise or when we talk about data, I think that is the most important asset any organization can have.

Business intelligence primarily focuses on using data analysis to support business decision-making whereas Data intelligence encompasses a broader range of activities, including data collection, storage, processing, analysis, and interpretation. It involves the entire lifecycle of data within an organization, from its acquisition to its transformation into actionable insights. This comprehensive approach ensures that data is effectively managed and utilized to derive valuable insights.”

How does data intelligence extend to end users?

Sanghai explains further

I envision data intelligence extending to end users through the development of a cohesive ecosystem. This ecosystem would integrate disparate applications across the organization, breaking down silos and enabling seamless correlation of data to create a single source of truth. With everyone accessing the unified system, users save time and effort, fostering efficiency. Additionally, it promotes collaboration and visibility, enabling users to share insights and understand broader objectives. Ultimately, this approach democratizes data access and insights, empowering every user within the organization to make informed decisions.”

Interestingly, he feels data democratization doesn’t justify the full scope of self-service analytics.

I refer to it as analytics democratization or BI democratization, expanding beyond the previous notion of data democratization. It’s not just about granting access to data, it’s about delivering value derived from that data. True democratization is providing insights effortlessly, on-demand, and timely. This comprehensive approach, which I term as data intelligence, imbues intelligence into every stage of the data journey.”

Understanding the need for self-service BI is critical before business intelligence implementation. I’m sure it’s on your BI requirements checklist.

Compare BI Software Leaders

What Is Business Intelligence Implementation?

A BI software implementation involves setting up workflows for data integration, mining, analysis, visualization, dashboarding and reporting. It follows a strategy that outlines preparing for, choosing and beginning to use a BI platform.

Having a plan helps you get your ducks in a row before embarking on deployment. It involves creating a budget, adding stakeholder input, identifying goals, choosing software and rolling it out.

Every BI implementation is unique, as every business has specific needs. With the variety and volume of data that comes in, data integration becomes Big Data Integration.

With thousands of ready-to-go BI solutions on the market, it begs the question — should you build or buy?

Build vs. Buy

Building in-house gives you greater freedom and flexibility, but with great power comes great responsibility. Get ready to plan ongoing maintenance, troubleshooting and upgrades.

Buying a readymade BI solution frees you from infrastructure and maintenance but might give you less control over managing and presenting your data. Tableau, Power BI and Qlik Sense are examples.

If you want more, customizing the platform is always an option.

You can also opt for a custom deployment with an implementation partner. A custom BI deployment will give you a tightly bound solution that serves your end-to-end data needs, from ingestion to analytics.

You can outsource reporting and dashboard projects as well if you don’t have the technical expertise.

What’s under the hood of a BI solution? Sanghai broke down the business intelligence implementation process for us.

How To Implement BI

BI implementations often fail to deliver. For an architecture with so many moving parts, things can go wrong. Data experts attribute it to the gaps in capturing requirements and identifying use cases.

Defining use cases in consultation with stakeholders, including the business owner, is a non-negotiable best practice.

Second, champion a data culture every day. Implementing a new system is like a relationship – you’ve got to put in the work daily.

It’s easier to continue working the way we always have, with the same results. Doing better means doing some things differently. Teams must put their faith in data and make it the driving force for decision-making.

Additionally, invite other teams to the table. Which departments will your BI solution connect — human resources, accounting, sales and marketing? Capture their input when defining requirements.

BI Architecture

Your ETL application connects to sources, extracts data and stores it in your data store. Your BI tool should have built-in data quality management for accurate insight.

For storage, Sanghai referred to the medallion architecture — a type of lakehouse setup. As your system process and stores the data, it increases the data accessibility, quality and business value from bronze to silver to gold.

Ask your deployment partner which model they use.

A cherry on the cake would be integrating a multi-dimensional processing engine. It pre-aggregates data for fast querying, even letting you explore large data volumes in less time.

This data feeds traditional reporting and advanced analytics which uses machine learning for deeper insight. Predictive modeling supports data-backed decisions by forecasting trends.

Here’s what business intelligence implementation entails.

Phase 1: Requirements Gathering

It’s the phase when you lay the groundwork for business intelligence implementation. Define your business needs, end goals, pain points and nice-to-have features.

Assess where you are in your BI journey. List the questions you want the solution to address. How are you answering them today? What are the gaps you’re looking to address?

Requirements feed into goals, and backtracking from your key objectives could tell you what you need. A discussion of how to set goals will help you define your requirements better.

Goal Setting
Goals should be SMART — specific, measurable, achievable, relevant and time-bound. Consider this goal.

Increase customer satisfaction score by 10% (as measured by Net Promoter Score) within the next year by using BI to identify and address customer pain points.

This goal is:

• Specific: It targets customer satisfaction scores and identifies the measurement method, Net Promoter Score.
• Measurable: It defines a clear target improvement of 10%.
• Achievable: The goal is ambitious but achievable within a year.
• Relevant: Improving customer satisfaction aligns with overall business objectives.
• Time-bound: It sets a timeframe of one year for achieving the improvement.

This SMART goal provides a clear direction for the BI implementation. How?

It mentions using BI to identify customer pain points. The BI solution should collect customer data using surveys, reports, and reviews or connect to systems that do.

The goal is more than gathering data. It’s about using the information to take action. The BI solution should provide actionable insights. What functionality will you need to achieve this?

What would be a not-SMART goal? Consider the following two objectives.

Goal 1: Reduce overall costs with business intelligence.

Goal 2: Reduce campaign cost by 15% within the next quarter using BI to identify and eliminate underperforming channels.

Goal 1 is vague and lacks specifics on what costs to target or by what amount. Goal 2 is measurable and describes how you’ll achieve it. It’s time-bound and relates the target to the method, BI.

Other Considerations
Evaluate the five Vs of your data. What value will the data impart?

Which data source will you connect to? What are your expectations regarding peak performance and scalability?

• Will incoming data be clean or need to be cleansed within your BI ecosystem?
• Does your analytics layer require OLAP?
• Will you deploy on-premise or on the cloud?
• Will you need zero downtime? Would a blue-green deployment be feasible?
• Which dashboards and reports will you need?
• How will you track project progress?

Compare BI Software Leaders

Phase 2: BI Product Selection

Intellicus Technologies educates clients about the capabilities of their solution at the onset. They call it the BI awareness stage. It’s an excellent way for clients to see the system in action and fine-tune their requirements.

Sanghai explains the reasoning behind this initiative.

So rather than solely inquiring about their business requirement… we collaborate with them. We guide them on the technical front and offer consultancy. We create awareness about analytics and BI… encouraging them to expand their ask and wish list beyond their existing knowledge, as many are still reliant on legacy systems or Excel using legacy systems… even within larger enterprises. We demonstrate the possibilities enabled by technology.”

Plus, it gets the ball rolling on stakeholder engagement.

Read our BI adoption strategies article to understand how employee buy-in can go a long way in boosting adoption.

Phase 3: Assessment and Planning

For a successful deployment, we recommend a top-down approach. It focuses on delivering measurable results from the first development milestone itself.

This methodology — Collaborate, Build, Operate, and Transfer (CBOT) — provides a high-level view of how the solution will work early on. It gives the client confidence their investment will pay off. Plus, it helps identify gaps early on.

As Sanghai says, the development team should see the value creation as clearly as the business owner.

…let’s imagine your BI system as a set of seamlessly integrated components… All working together towards a singular goal delivering the valuable insights you need… In a top-down BI approach, we start by clearly defining those desired outcomes… Just like reverse engineering a complex system, we then map out the data flow needed to generate the insights that will help you achieve those goals. This ensures that every element of your BI system is strategically designed to provide real business value from the very beginning.”

Define the technical and functional requirements.

• Determine the ETL processes to load data into a data warehouse/lake.
• Accommodate for semi-structured and unstructured data. Read our article on Types of Big Data to know more.
• Design a data governance framework with policies on data access, ownership and quality.
• Identify who will access the BI solution, what their roles are and what data they need.
• Classify your data based on sensitivity — public, confidential and highly confidential. It’ll help determine the required level of security on a case-by-case basis for specific data sets.
• Plan user training besides the available vendor resources. Self-reliant teams will deliver faster.

Calculate the cost of development. Skip to the Cost Considerations section to know more.

Phase 4: Development and Deployment

Establish Connections

Set up your data warehouse and connect it to your data sources and BI platform. Create data models and map them to your BI tool.

If you don’t need vast storage, a semantic layer is a good option. It’s a business layer that pulls data from where it resides using metadata and indexing, like in a book. It retrieves data faster and standardizes business terms.

Professional Services

Many BI solutions have ready-to-go dashboards and report templates. You can also add custom dashboards based on your unique KPIs and requirements.

If you don’t have resources in-house, some vendors offer dashboarding and professional reporting services.

App Building

You should be able to build apps over your BI platform if it’s extensible.

A retail clothing store might want an inventory app on top of its BI system. It would act as a middle layer for your BI tool, collecting data from the store’s POS, CRM and inventory management systems.

Deployment

Deploy the BI system to a production environment. An Agile development method keeps the feedback loop short and helps debug the solution as it rolls out. It involves developing the solution in short sprints while keeping the client in the loop.

Read more in our article on Agile.

User Training

The best BI software vendors will actively train your staff in self-service analytics. Reducing dependence on the vendor is in everyone’s best interest. Comprehensive training sessions and a thorough handover document will be effective.

Phase 5: Results Evaluation

Creating a list of KPIs for your BI team can lead you down a rabbit hole if you’re not careful. Without rock-solid metrics, it could become difficult to justify the time and effort invested.

Here are some suggestions for tracking your progress.

Internal KPIs

• Tracking usage is a good indicator of the BI solution’s effectiveness. Gather user feedback about the dashboard’s functionality, ease of use and overall value. It isn’t a quantifiable metric unless you use a rating scale or net promoter score.
• Calculate the number of tickets per X amount of revenue. Your support ticket backlog will tell you if users are happy with the business intelligence implementation.
• Compare the time employees spend on projects vs. resolving issues. The nature and number of improvement requests give you a fair idea of the success of your BI deployment.

Business KPIs

What’s the business impact? Have revenue figures improved since your solution went live?

Revenue generated and cost savings can be good indicators, but you must tie them back directly to the deployment. You can use a KANO analysis to link buyer satisfaction to your implementation.

Read our Guide to Building a Successful Business Intelligence Strategy to know more.

Compare BI Software Leaders

Primary Benefits

When business intelligence makes the industry go around, it’s no wonder investment in BI software is increasing. Fortune Business Insights predicts the BI software market value will rise to $63.8 billion by 2032.

BI organizes data to support reporting and business analytics. By filtering data, you can view performance and market trends and focus on specific business areas. Modern BI software helps calculate risk and predict events.

• Maximize ROI: Save money by identifying money sinks using dashboarding and reporting. Get a clear picture of where to invest to get the maximum gains. Reduce wastage by pulling the plug on inefficient processes.
• Stay Competitive: Apply market research to your offerings. Use customer data to your advantage. Pivot and adapt to buyer needs, an essential trait in a fickle market. Offer products and services others haven’t thought of yet.
• Benchmark Performance: Learn how to improve by comparing your progress with others in your industry. Set measurable goals by defining KPIs. Keep employees motivated with goals they can aspire to.
• Manage Operations: ERP and project management are critical areas benefiting from operational BI. Control costs and improve supply chain management by tracking inventory levels, suppliers and delivery data.
• Delight Customers: Satisfy buyers by designing products that align with their tastes. Surveys, social media feedback and customer service calls offer critical insights. Earn loyal clients and avoid churn with the correct data.

Read our Business Intelligence Use Cases article for BI success stories.

Cost Considerations

The best things in life are free, but business intelligence implementation can cost you a tidy sum.

Deployment expenses include hardware acquisition and maintenance, user management, and dashboard development. You’ll need a database server, another server to account for redundancy and a client-side server for end users.

Additionally, developing custom features and the ability to handle additional workloads will add to the cost.

• Check if the BI platform offers flexible options, such as core and user-based licenses. Core-based licensing is cost-effective for more users.
• On-premise deployments are yours to control but have a higher cost of entry, and costs start from $80K.
• Cloud BI software is available for monthly and annual subscriptions, but reliance on the vendor introduces variables you can’t control. Subscription costs vary depending on the number of users, source connections and data volumes.
• Subscribing to a readymade BI solution can cost you $500 to $6K annually. Power BI Premium per user (PPU) is available for $20 monthly.
• Usage-based subscriptions allow you to pay as you go.
• Your license requirements can vary by the type of users. Power-user licenses will cost more than those for reporting consumers.
• A perpetual license does have a higher cost of entry, but it gives you ownership and has lower long-term costs. You may need to pay for additional support and upgrades.

Big data drives many software trends. The semantic layer provides rock-solid support for LLMs, paving the way for businesses to build custom language models.

Compare BI Software Leaders

Trends

Semantic Layer

Data warehouses can be expensive, and third-party data adds to the cost. However, large data volumes are crucial for advanced analytics like machine learning.

A semantic layer can help manage these costs by simplifying data access and potentially reducing the load on your data warehouse servers.

It also allows for easier scaling of the underlying data storage by providing a consistent view regardless of where the data resides.

Sanghai tells it like it is.

Just imagine your data like a massive archive of documents. A data warehouse is like a central filing system for all those documents, meticulously organized but perhaps a bit complex to navigate. A semantic layer, on the other hand, is like a well-designed index. This index allows you to quickly find the information you need without needing to delve into the intricacies of the archive itself.”

The best part is you don’t have to move data from where it resides.

Thanks to linked datasets, your end users see contextual data views, though the results live in their source. The semantic layer retrieves data from sources, performs joins and transformations, and delivers the results at super-fast speed.

Whether a semantic layer will suit your business depends on your data, users, audience and analytics level.

Case Study

Let’s circle back to the mining company. Unified data views topped the list of their goals for BI implementation. They identified the following requirements.

• Data silos were spread across different locations and multiple legacy systems were in place
• The need for manual intervention, and checks with data quality issues.
• Lack of standardized data processing, quality checks, reconciliation & validation.
• Inability to scale. Performance and data management issues.
• Not leveraging advanced ML-based analytics.

They collaborated with Intellicus and leveraged their unified data engineering platform to develop and modernize their end-to-end analytics ecosystem.

This comprehensive BI implementation empowered all 12 business units and their departments with self-service analytics.

Executives gained real-time insights through interactive dashboards and reports, enabling proactive decision-making based on the latest data. Proactive alerts kept everyone informed of critical trends and events.

Is the semantic layer suitable for every organization size? Sanghai thinks so. Large companies will benefit from its ability to manage complex data volumes while mid-sized companies can plan for growth with this technology in their corner.

Domain-Specific LLMs

The rise of machines—Gen AI and LLMs—has spurred the adoption of the semantic layer and spawned a new era.

Anurag clarifies:

A semantic layer plays a crucial role in organizing data through tasks like performing joins, translating data terms, and creating hierarchies…This structured data proves immensely beneficial for training Large Language Models (LLMs). While LLMs are capable of being trained on raw data, the presence of a semantic layer offers a more organized and consistent format, potentially enhancing training efficiency… It’s essentially a synergistic relationship where the semantic layer and LLMs complement each other effectively…. The semantic layer provides the structured data necessary for training, while LLMs, in turn, deliver insights in a human-readable format…. This symbiotic relationship amplifies the value of both components, resulting in more effective data processing and analysis.”

Compare BI Software Leaders

Next Steps

A successful BI deployment will deliver rich rewards in due course. Be patient and adhere to a data culture in your workplace. Highlight the successes of BI projects, however small, and incorporate the learnings the next time around to keep growing.

Start right. Get our free BI requirements template to select a BI platform that matches your unique needs.

How was your business intelligence implementation experience? Let us know in the comments.

Contributing SME

Anurag Sanghai is an experienced IT professional and thought leader with over 14 years of expertise in consulting, architecting and implementing state-of-the-art data analytics and business intelligence solutions for global clients.

He’s worn many hats, including those of Full Stack Engineer, Data Architect, Engineering Manager and Solutions Architect. His experience includes developing enterprise data warehouses, analytics, trends and forecasting systems, and data visualization initiatives.

An avid innovator, Sanghai is a U.S. patent holder.