10 Keys to Successful BI Implementation
An IT industry executive discusses some best practices to keep in mind while implementing BI to ensure that the data is useful and gets used.
Join the DZone community and get the full member experience.Join For Free
Why is it so challenging to get Business Intelligence (BI) adopted, implemented, and leveraged in an organization? Unlike CRM and ERP tools, which are essentially data collectors, BI requires just as much thought to implement and sometimes even more resources than its transactional counterparts.
Yet many organizations continue to struggle to ensure that it is being used effectively.
Why? Some say that it is because BI is a non-critical component of the collection of tools that a company uses to conduct business. Can't invoice a customer? That's a show-stopper, some say. But if you can't access a dashboard, then there's no crisis, they continue. While it is true that, in IT, priority has historically been given to transactional systems over analytical ones, is this still the case?
It may be that many organizations are challenged in this way because, if necessary, users can employ a BI workaround when things get tough, giving the impression that BI is of lesser value. If BI is unavailable, for example, users might still be able to pull some data from exports and data feeds and get some of their questions answered and even make some decisions using Excel or similar tools. Or they can even simply show their manager the raw numbers.
Secondly, company culture has something to do with it. Transparency is not always desirable, so there can be a natural tendency to block automated reporting to management before there is a chance to sanitize, understand, and potentially even change the numbers (sometimes for valid reasons) before the upper echelon gets to see them.
Still, assuming that some organizations are ready and willing to embrace BI as a core system, why do so many of them let go of structured reporting after a period of time? How can you ensure that those metric-rich dashboards that you worked so hard to create, providing Key Performance Indicators (KPI) for your entire organization, do not go to the recycle bin just a few months later?
Here are ten things you can do to ensure that BI sticks around in your organization:
Make sure that you're not stuck with a middle-out approach when the in-house Business Analyst needs to fight with the decision-makers on top to get funding and approval of their time to implement BI, as well as fight with the people at the bottom who resist change and transparency.
Ensure that your boss or customer wants you to build it, and build it for them, first and foremost. Middle management will agree much faster if the top tier is already using BI to manage their way of working.
2. Data Visualization
Data visualization is just one aspect of Business Intelligence. It can be cumbersome, or it can be simple. It can be a multi-page report with lots of filters, or highly focused on key metrics of importance with prompts you to find out more. Choose what is best for your users.
Keep your data visualization cool, fresh, and modern. Add a variety of small but useful features that keep the users interested in viewing it again and again. Do not confuse data analysis with performance management. Data analysis is when you are still digging into the data, and you're not sure what you are looking for or even what you want. Performance management begins with knowing what you want to measure and allows you to continue doing so frequently and constantly.
Don't get carried away with charts if tables can do the job. Don't get carried away with complex pivot tables if a single-number indicator can do the job.
Add time and dimension filters only if they're needed — not "just in case" someone might want them. Stick to basic metrics that are easy to understand. More complex ones or derived metrics can be calculated for the infrequent times they are needed.
3. Build on Success
Do not try to turn your BI into an all-out centralized reporting solution. Solve problems and answer questions one at a time. Start slow, show value, show agility, and don't spend money and time unless you have to. Avoid "RFP" types of implementations. It is a clear sign that if you are doing an RFP that you do not know what you want. Be clear about what you are trying to achieve and do it, even if you have to do it first on paper or in Excel.
Once you are satisfied and your users want more, then find the tool to automate it and distribute it. Then, move on to the next one. Take one win, one person, or one department at a time.
Prototype your dashboards and reports with your users. Sit with them and draw things out with PowerPoint or Excel or on paper and guide them to answer one question: "What are up to three metrics that you would like to have immediate access to at any time?" Drill into those. Find out the relevant time frames and dimensions and how best to visualize the data. Ask them what details would be interesting for them to see afterward.
Don't let them stray into adding more or "nice-to-haves." If they're not sure what they want to see or how they want to see them, then there's a good chance they don't really know what they want yet.
6. Ease of Access
If you build it, they will NOT come. Don't waste time building that internal portal with all those cool login screens and logos and news panes. They will not use them. Instead, push reports and dashboards to users' phones, watches, email, or Slack.
Find a scheduler that automates publications of dashboards and delivers them on time. Avoid publishing too frequently. Otherwise, it will be seen as spam, and it will quickly be deleted, sight unseen, by a mailbox rule.
Use the Alert function of your BI tool to send the dashboard only at certain times or under certain conditions, not all the time. Doing so will help increase that email or messages importance in the eyes of its receivers.
6. Speed of Data
Everyone loves live metrics, and data delivery speed is one of the keys to success. The quickness between the time new data is available and the time when dashboards that report that data and its effects are delivered is critical to ensure that your end users trust and rely on them. There is hardly anything more exciting to watch than a user group waiting for a set of dashboards to arrive in their inbox to get feedback on how they performed. If you have enough data volume, consider refreshing data more often so that changes are visible throughout the day.
Login. VPN. Password. Connect. Login. Password. Connect. Ahhhh, forget it!
Avoid obstacles. Security is important, but the more steps you have between initial device login and presentation of data, the more chances you have of building a soon-to-be-dead BI. Look for Single Sign-On (SSO) options that allow the user to sign on to the device quickly, and that is enough to access the dashboards (fingerprint, maybe?). Try to centralize security so that the users don't have to remember multiple login credentials. You will see the usage rise as a result.
Delivering data on mobile devices forces you to think hard about what you are trying to show. There's not enough space or functionality to put a pivot table capable of slicing and dicing data in a million ways. And even if there was: don't! Mobile devices are perfect for focusing on metrics and fulfilling many of the requirements I mentioned above, including ease of access, simple security, and pushing data.
9. Embed and Integrate
If CRM and ERP are core to the functioning of your organization, I recommend embedding BI in those tools and ensuring that you can navigate in and out of one system without leaving the actual system. For example, if you use Salesforce for your CRM, then integrate your dashboards inside the application. Make the data quite actionable. For example, if a user selects a chart or data on the dashboard, provide a parameter to take them back to Salesforce to show more detail.
10. Build Heroes
Don't be the only person responsible for BI in your company. Yes, everyone thinks you are the go-to person who can get the data they need, and that's nice. You get to be a hero for a while. But once the workload becomes unmanageable, you'll be seen as the one who is always late, who controls the BI access, and who doesn't have time to build that other dashboard that a department so badly needed.
Instead, be the hero who makes other people successful. Build a cross-departmental team of people who are responsible for, trained in, and own the BI in your organization. They can help even if the dashboards don't directly concern their department. They'll learn new tools and processes-something that makes professional team members more satisfied with their job. Their presence will be especially helpful when the workload peaks.
Another advantage is that if someone is not around, the team can help out. An integrated team within the business helps to reduce costs. The business also benefits because, believe it or not, you might not be around some day. Building in redundancy is valuable.
Now you have ten ways to make sure BI sticks around for a long time in your organization — even after you are no longer responsible for it.
Published at DZone with permission of Telmo Silva, DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.
Building a Headless E-Commerce App Using OceanBase and Python
Top 10 Pillars of Zero Trust Networks
Integration Architecture Guiding Principles, A Reference
Security Challenges for Microservice Applications in Multi-Cloud Environments