Data Crafts

Data Crafts Logo
bi implementation

Business Intelligence is a set of components that feed data to people in organizations to inform decision-making. It is a complex and nuanced subject, and organizations face many challenges during the adoption process. In our post, we tell about the most common ones we observed and share some ideas to mitigate them.

There are three main groups of BI challenges:

  1. People Challenges: challenges related to team members, available skills, and the company’s culture. People often underestimate the importance of having the right talent, skills, and environment to plan, execute and maintain business intelligence initiatives.
  2. Process and Management Challenges: even having the right team, challenges will arise without the proper strategy, documentation, and processes. Without them, the team members won’t be aligned, and the results will be suboptimal.
  3. Technology Challenges: challenges related to the choice, implementation, and management of the tools used to deliver the BI program.

So, let’s go ahead and discuss the most common BI implementation challenges shared by our business intelligence services experts.

BI Challenges


BI Challenges

People Challenges

1. Having the right talent

You need the right team and skills to run a successful BI program. Too often, we see teams with junior members or people with “close” skill sets (e.g., analysts and front-end developers) put in charge of business intelligence programs or assigned heavy responsibilities. This isn’t the right approach and will invariably lead to severe problems.

The first thing to be aware of is that this talent is scarce and expensive. There is no way around it, so be ready to pay expensive wages. If your program is big enough, you can reduce the costs by including junior members. If you want to do so, always ensure that the program lead is knowledgeable and that the team members’ experience is balanced. Too many people to manage will keep the leaders busy training other people and take too much of their time for architecting and planning.

The second thing is to identify the roles you’ll need to run the Business Intelligence implementation and balance the team according to their knowledge. We borrowed the following image from the dbt blog. It clearly illustrates data team roles according to their position in the spectrum that goes from raw data to reporting/action/insights:

An image showing the spectrum of data jobs and what output they generate.


Extracted from How to find a role in Analytics Engineering? on the dbt blog

You can use other spectrums to balance the team, like production vs. communication and business vs. technology. The key idea here is that team skills must be balanced to achieve the business objectives and allow the data to provide insights and drive action. The exact team members will vary depending on the company’s size, experience working with data, the communication flows between departments, and other factors that will determine what skills have the most value for the team responsible for the BI implementation.

One last key to success is the hiring process. I can’t stress enough how important this is. When organizations rush to hire the people they need when they need them without planning, they end up not having the right skills, cultural fit, or experience level. In this type of work, people are not interchangeable. The difference between an ok person that does the job and a person that excels at it is HUGE. The right person can deliver 10x the outcome of just an adequate one. The right four people will provide a result that is orders of magnitude better. So, ensure you invest enough in creating the job description and selecting the right candidate.

2. Getting the key people involved

In contrast to the previous point, this challenge refers to getting the right people from each department involved with the business intelligence program. BI initiatives serve information to all the departments in the organization. However, not everyone in the other departments is equally literate in data. Selecting the right people to interface with each department will go a long way toward communicating requirements and adding value to them. Conversely, choosing the wrong people will hinder the success of the program.

Whether you are a small organization getting people involved ad hoc or a big corporation creating an Analytics Centre of Excellence, make sure you talk to the right people and be purposeful with who you get involved. The exact shape of the team will depend on your organization, but there are a few green flags for good candidates:

  • They are data literate,
  • They understand the objectives of the organization and their team,
  • They are good communicators,
  • They are team players,
  • They have good relationships with both their bosses and their peers

If they check all the boxes, they understand what you are attempting and why and become a great ally to promote and champion your work.

3. Lack of adoption and resistance to change

There will always be people who don’t want to adopt the new tools. There are many reasons why they resist change: they are comfortable with their current systems, believe the new platform will be challenging to learn, feel clumsy with it, don’t trust the data, or prefer what they currently have based on their experience. Regardless of the reason, it would be best if you don’t try to force them to use the tool, as it will upset them, lower their morale or, in the worst-case scenario, increase their resistance to adopting the tool. There are other ways to lure them to your side.

The first and most obvious is getting them involved from the beginning (and forever more). Some people fear being left aside, not having their needs accounted for, or are upset because the organization adopted the new tools without consulting them. You can reverse that fear if you ask them to provide input to the process. This is the most efficient way to mitigate their concerns. If things go well, you might get them excited about participating, develop a sense of ownership, provide valuable input and become your best advocates.

If they don’t want to participate, a powerful incentive is to showcase your successes to the rest of the organization. On one of our most successful projects, we started working in a team of only 4. Two months in, we had a couple of notable wins. Our main stakeholder shared it with the broader organization, and it was like magic: we started getting requests from people to help them execute their ideas. Shortly after, we got so busy that we had to instaurate a weekly council with over ten people to discuss and prioritize ideas and initiatives. Everyone wanted a bite of the success. We had become the golden team, and everyone wanted to be part of it.

4. Preventing siloed work

While people are expected to collaborate in small groups and teams within organizations, sometimes they work too isolated. When this happens, communication is minimal, and there is no collaboration between teams. In the best case, this situation causes inefficiencies and hinders synergies between them. In the worst cases, groups have incompatible objectives that introduce competition and conflicts among people. This creates frustration, demoralizes the teams, and can damage the organization’s performance.

One of the BI Implementation Challenges are team silos

So, what can you do to avoid silos and get people and teams to collaborate on BI implementation projects? A general rule is to be empathic towards other groups and people. If we make an effort to understand why other members are acting the way they are, what they are afraid of, and what motivates them, we’ll be able to:

  • Understand what drives them,
  • Identify what incentives are in place,
  • Understand how they misalign teams and people,
  • Design and share an incentive structure that better aligns the teams with the business,
  • Speak to people in a language they understand and appreciate.
  • Most people in any organization strive to contribute to its success (unless they are toxic and shouldn’t even be there). It is a matter of empathizing with them to understand their goals and pains so that we can help them feel motivated, comfortable, and happy while collaborating with the broader organization and us.

Other tactical initiatives can help, like recurring meetings to share objectives, performance, wins, and additional information that brings everyone together; creating cross-department roles and projects; bringing leadership together on regular meetings; and putting mechanisms in place to gather feedback from team members.

5. Establish a data-driven culture

The adjective data-driven means that progress in an activity is compelled by data, rather than by intuition or by personal experience.


Having a data-driven culture means the organization has the proper incentive structure to motivate each individual toward being data-driven. Most companies we work with are what I’d call “selectively” data-driven: they are data-driven as long as data tells them a story they like. As soon as it limits them in any way, they stop caring.

This is not the people’s fault. This happens because the organization incentivizes people in the wrong direction. We learned from Freakonomics‘ cheating sumo fighters and drug dealers that people respond to incentives.

If you incentivize developers to release an application faster, why would they care about its performance after the release? Or having the correct measurement in place to assess the application’s performance? If you put pressure on an eCommerce manager for a low conversion rate, why wouldn’t she try to release a checkout update as soon as possible, no matter if she A/B tested it? Why wouldn’t she try to get the most flattering picture when analyzing the data?

To establish a data-driven culture, the organization must design a reward system that motivates people to experiment, measure, and learn, that celebrates learnings over good results, allows people to fail and try again, recognizes contributions with ideas, and allows people to be creative and proactive without fearing the consequences of failure.

Process and Management Challenges

1. Justify investment ROI

A person looking at an investment graph. Calculating investment ROI is one of the most complex BI implementation challenges.

Unfortunately, it is common for directives to ask for an ROI estimation of business intelligence initiatives. We understand that reasoning: BI projects represent significant investments, and they want to understand what they’ll get in return. There are a couple of reasons why this conversation isn’t productive:

  • As Evan Lapointe once told me, “Asking about the ROI of data investment is like asking about the ROI of having a higher IQ.” You learn faster and can solve more complicated problems with a higher IQ. Similarly, you make better decisions effortlessly by accessing timely, accurate, and relevant data. While it is hard to calculate precisely how much better these decisions will be and how much they will positively impact your organization, it is hard to argue that making every decision better will considerably improve the overall results.
  • Data should be considered a utility. Nowadays, no functional organization questions the value of having utilities or using computers. Suppose you want to compete in the current time and day. In that case, you must use electricity, water, climatization, and computers, or (unless you are into a particular sector) you’ll run out of business soon. In that sense, data should be considered a utility as it can and will make a difference and give you an edge over your competitors.

If the two reasons above don’t convince your audience (which is likely), then you’ll have to present some numbers. Whenever we have to do this, we estimate the cost savings or marginal gains of:

  1. Time saved: how many person-hours will you save yearly?
  2. Productivity gains: how much more will you be able to get done
  3. Faster access to insight/decision-making: how much can you save if you stop a bad-performing campaign? How much more can you sell if you identify an opportunity faster?
  4. Better marketing investment allocation: What is the additional profit of shifting X dollars from a campaign/channel delivering Y percent more ROAS during a year?

This will not be accurate, but it will build a compelling case. Just make sure the estimates are sensible.

2. Define a BI strategy

It is surprisingly common to see companies starting to implement Data Warehouses or adopting BI solutions without planning. They assign a technical person to do the work and expect it to meet the business needs.

While technical people and teams are great at solving challenges, they benefit a lot from working with someone who can identify and understand the problem the organization needs to solve and translate it into terms that give them clarity. This aligns the development with the business needs, maximizes its value, reduces frustrations, and improves stakeholder relationships.

Additionally, that technical person or team is often busy with other activities like developing and maintaining production systems or doing the front-end work. They will approach this work as any other development task or a side hassle, trying to close it as soon as possible without incorporating all the work required to understand the business needs and design a solution that meets them.

There are many reasons why it is crucial to create an effective business intelligence strategy and implement it. Dedicate time and resources to identify the business goals, list the data sources, develop a plan for data analysis and reporting, identify the key performance indicators, communicate the strategy to the stakeholders, and review and update it over time. This will put you on the right path to getting the data and insights you need to achieve your goals.

3. Define KPIs

This relates closely to defining a BI strategy, but it isn’t quite the same. The BI strategy describes the goals in a broad sense and explains how to achieve them at a high level. In turn, the KPI establishes how to measure success against a specific goal.

For example, a goal defined in a BI strategy for a publisher’s website could be “To increase the traffic to our website.” In KPI terms, we can translate that to the number of users, number of visits, number of page views, or number of ad impressions. If you chose the number of users as a KPI, you might focus on increasing the ad spend, the SEO efforts, or our link-building strategy. If you select the number of page views, you could add internal links to your content, split the content into multiple pages or add functionality to your website. You might add more ads to every page if you want more ad impressions.

As you can see, your chosen KPI will significantly impact the measurement strategy and the actions it derives. It is vital to pick them carefully and ensure they are aligned with the business goals and values.

4. Disjointed definitions

Have you ever heard, “Wait, does gross margin include shipping? I thought it only included taxes”, “My leads in GA4 don’t match my leads in Salesforce”, or “We spend hours discussing the definition of revenue.” If so, your organization is most likely handling different definitions for each metric. We remember when we updated the gross revenue calculation three or four times, thinking that the client was hesitant about which one to use. Ultimately, we realized the requests came from people in two different departments. Every time we updated the gross revenue calculation according to one department, the other department noticed a discrepancy with their definition and requested an update.

None of them was wrong; they were using different metric definitions. Once we realized this, we jumped on a meeting and agreed to create two metrics: Gross Revenue after Taxes and Gross Revenue before Taxes (apologies if any financial folk is reading this). Afterward, everyone understood the value they were looking at, and there were no more discussions or doubts about it.

Handling diverse definitions will confuse people, introduce inefficiencies and reduce trust in the data. Make an effort to define, describe and document what each metric means to avoid undesired side effects.

5. Generating and maintaining documentation

No one likes generating documentation, and fewer people want to read it. As disliked as it is, it is fundamental to bring together all the previous points in this post. Words are lost in the air, so if you create a strategy, KPIs, and metric definitions but don’t document them, people will forget them, new employees won’t be aware of them, and they will start creating new definitions.

There is an overwhelming number of options to create and host documentation. Depending on the size and sophistication of your organization, some options might be better than others. If you work for a corporation, you might research tools like Atlan, Azure Data Catalog, or an internal knowledge base. If you are a startup, you might prefer to use a Google document, a Gist file, Confluence, or a spreadsheet. Whatever you do, make sure it:

  • It is accessible to everyone
  • Supports versioning
  • It is comprehensive
  • It integrates with your workflow

If you are not sure how to proceed, start small and share. We promise people will use whatever you give them as a reference and start asking questions and providing suggestions.

6. Support

There are several challenges that organizations may face when trying to get support for their business intelligence (BI) systems:

  1. Finding the right resources: It can be challenging to find the right support resources, such as technical support, training materials, or consulting services, that are specific to the BI tools or software being used
  2. Limited vendor support: Some vendors may not offer comprehensive support services or only offer support during certain hours or for a limited duration
  3. Language barriers: If the BI tools or software are from a vendor based in a different country, language barriers may make it difficult to get support
  4. The complexity of the BI system: If the BI system is highly customized or uses advanced features, it may be more challenging to get support, as it may require specialized knowledge or expertise

To overcome these challenges, it can be helpful to research and compare different support options to ensure that you are getting the best value for your money. It may also be helpful to build internal expertise on the BI system by training your team or hiring staff with relevant experience. There are a few different ways that you can get support for your business intelligence (BI) implementation:

  1. Reach out to the vendor or provider you used for your BI tools or software. Many vendors offer support services, including technical support, troubleshooting, and training resources.
  2. Look for online communities or forums where you can ask questions and get help from other users familiar with BI tools and software.
  3. Consider hiring a consultant or consulting firm that specializes in BI implementations. They can provide guidance and support throughout the process.
  4. Consider training your team on BI best practices and how to use the specific tools and software you have implemented. This can help to ensure that your team has the skills and knowledge needed to use the BI system effectively.
  5. Lastly, make sure to take advantage of any documentation or resources that are provided with your BI tools or software. These may include user manuals, online guides, and training materials.

7. Ensure a healthy communication

Effective communication is essential for successful business intelligence (BI) implementations for a few reasons:

  1. Ensuring everyone is on the same page: BI implementations often involve multiple stakeholders, such as business users, IT staff, and executives. Ensuring everyone is clear on the project’s goals, objectives, and timeline will help ensure it is successful.
  2. Sharing information and knowledge: BI implementations often involve collecting, integrating, and analyzing data from multiple sources. Sharing this information and knowledge with relevant stakeholders can help to ensure that the BI system is being used effectively and to its full potential.
  3. Collaboration: BI implementations often require collaboration between different teams or departments. Effective communication can help facilitate this collaboration and ensure the project stays on track.
  4. Avoiding misunderstandings and conflicts: Miscommunications or misunderstandings can lead to delays or conflicts during a BI implementation. Ensuring everyone knows what is expected and what is happening can help prevent these issues.

Overall, effective communication is key to ensuring that a BI implementation is successful and the BI system is used to its full potential. There are a few steps that organizations can take to improve communication during business intelligence (BI) implementations:

  1. Establish clear lines of communication: Identify the key stakeholders and decision-makers involved in the BI project, and establish clear communication channels with them. This can include regular meetings, email updates, or other forms of communication.
  2. Define roles and responsibilities: Clearly define the roles and responsibilities of each team member or department involved in the BI project. This will help to ensure that everyone knows what is expected of them and who to go to for specific tasks or questions.
  3. Communicate regularly: Schedule regular meetings or check-ins to keep everyone informed about the BI project’s progress and address any issues or concerns that may arise.
  4. Use clear, concise language: When communicating about the BI project, use clear, concise language that is easy for everyone to understand. Avoid using technical jargon or acronyms that may not be familiar to everyone.
  5. Encourage feedback and questions: Encourage team members to ask questions and provide feedback throughout the BI project. This can help ensure everyone is on the same page and that any issues or concerns are addressed on time.

Overall, effective communication is essential for successful BI implementations. By following these steps, organizations can help to ensure that everyone is informed and that the BI system is being used to its full potential.

Technology Challenges

1. Choosing the right tools

Choosing the right Business Intelligence tools can be daunting, especially if you have never done so. There are many aspects to consider, and it is not always easy to conciliate them. Here is a list of critical factors you take into account when choosing the right business intelligence (BI) tools for your organization:

  1. Business needs: consider the specific business needs the BI tools will need to address. This includes the type of data being analyzed, the types of analyses that will be performed, and the stakeholders who will be using the BI system.
  2. Data sources: list the types of data sources that the BI tools will need to integrate with, such as databases, spreadsheets, or cloud-based data stores.
  3. Data visualization: look for BI tools that offer various data visualization options, such as charts, graphs, and maps. This will help to make the data more meaningful and easier to understand.
  4. Ease of use: Choose BI tools that are easy to use, with intuitive interfaces and comprehensive documentation. This will help to ensure that your team can quickly get up to speed on the system.
  5. Scalability: Consider the scalability of the BI tools, as your organization may grow and need to handle more significant amounts of data or more users over time.
  6. Integration with other tools: Consider whether the BI tools can integrate with other tools or systems that your organization is using, such as customer relationship management (CRM) or enterprise resource planning (ERP) systems.
  7. Cost: Evaluate the total cost of ownership, including any upfront licensing fees, maintenance costs, and training expenses.

Weight these factors according to your needs and preferences. At some point, you will have to compromise some of them in favor of others, but doing so consciously will help you prioritize what is more important to you and your organization.

2. Controlling the costs

The platforms, programs, and services that compose a BI implementation can be costly, especially if you use cloud-based solutions that scale “automagically.” Before moving to monitor and control costs, establish a budget for the BI implementation and stick to it as much as possible. This will help you make informed decisions about tools or services to invest in. Once you have decided on a budget, you can move on to other decisions that will impact the cost of the implementation:

  • Evaluate multiple options: don’t just go with the first BI solution that you come across. Shop around and compare different options to find the best fit for your organization’s needs and budget.
  • Consider open-source solutions: research open-source BI tools or software, often free or low-cost. We wouldn’t recommend using only an open-source option, but complementing a SaaS ETL platform like Stitch or Fivetran with a self-hosted instance of Airbyte can save you tens of thousands of dollars a year.
  • Train your team: Invest in training for your team to ensure they can use the BI system effectively. This can help to reduce the need for ongoing support or consulting services.
  • Monitor the system: add cost monitoring to avoid undesired surprises. Recently, our client found herself in a situation where she has to migrate a dozen connectors out of an ETL solution or pay a significant bill to upgrade her ETL plan. She would have had more time to react if she had been screening the volumes.

By following these steps, you can help to control the costs of your BI implementation and get the most value for your investment.

3. Maintaining a healthy implementation

Having a valuable Business Intelligence program doesn’t end with the implementation. You should plan how to keep it healthy over time, updating it, fixing the issues that inevitably appear, helping people use it, and adding new connectors when necessary. To maintain a business intelligence (BI) implementation, it is essential to:

  1. Keep the BI system up-to-date: keep the BI system software and security patches up-to-date to ensure that it runs smoothly and securely.
  2. Monitor system performance: regularly monitor the performance of the BI system to identify and address any issues that may arise. This may include monitoring the system’s uptime, response time, and data quality. Adding alerts can be beneficial as well, especially for time-critical reporting.
  3. Manage data quality: maintain the quality of the data being used in the BI system by implementing data cleansing, validation, and enrichment processes.
  4. Train users: provide ongoing training and support to users to ensure they can use the BI system effectively and to its full potential.
  5. Establish best practices: Establish best practices for using the BI system, such as guidelines for data entry, reporting, and analysis.
  6. Review and optimize: Periodically review and optimize the BI system to ensure that it meets the organization’s needs and is used to its full potential. This may include adding or removing features, integrating new data sources, or making other changes to the system.
  7. Explore tools that can help maintain, monitor, and keep the implementation up to date: A proper dbt implementation, for instance, will go a long way in monitoring the data quality, its freshness, and the data source connectors.

These activities will help ensure that your BI implementation runs smoothly and effectively over time. Be mindful that keeping things running takes planning, time, and energy.

4. Make the system accessible

To maximize the impact of BI in your organization, you have to make it as accessible as possible. The more people use it, the more insights they’ll find, the more questions they’ll ask and the more information will be shared and democratized:

  • Ensure that the data being used in the BI system is available to all relevant stakeholders, including business users and executives. This may involve integrating data from various sources, such as databases, spreadsheets, or cloud-based data stores.
  • Avoid using technical jargon or acronyms when communicating about the BI system or the data it contains. Use clear and concise language that is easy for everyone to understand.
  • Offer training and support to users to ensure they can use the BI system effectively and to its full potential. This may include training on using specific BI tools or software and best practices for data analysis and reporting.
  • Use data visualization techniques like charts, graphs, and maps to make the data more meaningful and easier to understand.
  • Choose BI tools and software that are easy to use, with intuitive interfaces and comprehensive documentation. This will help to ensure that users can quickly get up to speed on the system.

5. Data quality problems

Bad data quality can be devastating for a BI program. If people stop trusting the data, they will spend more time asking questions and seeking assurance than using it. This will keep the BI engineers busy and won’t drive action or add value to the organization.

Data quality issues can often arise from errors made during data entry, such as typos, transposition errors, or incorrect values being entered. They can also arise if the data is not formatted consistently or in a format incompatible with the system being used.

Another reason data quality issues can occur is if different data definitions are used for the same data elements. For example, if one team defines “customer” as a business and another defines it as an individual, there may be confusion when the data is combined. Data that isn’t kept up-to-date can become stale or inaccurate, leading to data quality issues. Data quality problems can arise when integrating data from multiple sources if the data isn’t cleaned or appropriately transformed.

By identifying and addressing these potential sources of data quality problems, organizations can help to ensure that the data being used in their systems is accurate and reliable, increase trust and help people use data with confidence.

6. Combine data sources

Combining data from multiple sources can be challenging for a few reasons:

  • Data formats: Data from different sources may be in different formats, such as CSV, Excel, or a proprietary format. This can make it challenging to combine the data in a meaningful way.
  • Data structure: The data from different sources may be structured differently, making it difficult to integrate and analyze the data as a whole.
  • Data quality: The data from different sources may have different levels of quality, making it difficult to trust the accuracy and reliability of the combined data set.
  • Data security: Ensuring the security and privacy of the data can be challenging when combining data from multiple sources, as it may involve handling sensitive or confidential information.
  • Data governance: Establishing and enforcing data governance policies and procedures can be challenging when combining data from multiple sources, as it may involve working with different teams or departments with different rules and protocols.

Overall, combining data from multiple sources requires careful planning and management to ensure that the data is accurate, reliable, and secure. Plan the integrations carefully, and check the quality and accuracy of the results before publishing them to production.

7. Query performance

The query performance can start decreasing as data grows in size and complexity. This might not be an issue initially, but users will get frustrated if the response becomes too slow. Many factors influence query performance and response speed. There are several components that you can check to improve query performance:

  • Optimize the database design: Proper database design, including using indexes and proper data types, can significantly improve query performance.
  • Use appropriate query techniques: Use appropriate techniques, such as using the right join type, filtering early in the query, and avoiding unnecessary calculations.
  • Use stored procedures: Stored procedures can be faster than ad-hoc queries, as the query plan is stored in the database and does not need to be re-evaluated each time the query is run.
  • Optimize the schema: Optimize the schema of the data being queried to reduce the amount of data that needs to be processed. This may include denormalizing the data or using summary tables.
  • Use query hints: Use query hints, such as FORCE ORDER or FORCE INDEX, to give the database more information about how to execute the query.
  • Use appropriate hardware: Using proper hardware, such as fast processors, plenty of memory, and fast storage, can also help to improve query performance.

There might be other reasons why specific queries are slow, like system bottlenecks, high demand, bad provisioning, or even network connection. If you are experiencing issues and you can’t find the origin of the problem, you should get the help of a professional.

Final Thoughts

Business Intelligence implementation can be a complex process. To succeed in it, organizations must approach the challenges holistically and take a comprehensive approach.

This includes creating a clear BI strategy that aligns with the organization’s goals, investing in data quality and data integration, and ensuring proper data governance. Additionally, it’s crucial to continuously monitor and evaluate the BI system to ensure that it’s meeting the needs of the organization and making adjustments as necessary. By addressing these challenges, organizations can unlock the full potential of BI and gain valuable insights to improve their decision-making and overall performance.

If you’re looking to implement BI in your organization, don’t hesitate to contact us. Our team of experts can help you design a comprehensive BI strategy and provide the support you need to ensure a successful implementation.

Leave a Reply

Your email address will not be published. Required fields are marked *