Top 8 critical success factors for a business intelligence implementation
Delivering a successful business intelligence implementation can be transformational. How? For a start, giving business users access to self-serve reporting empowers faster, more accurate decision making. Well-designed dashboards and KPIs lead to improved situational awareness and the ability to steer the business with agility. And a single version of the truth – usually stemming from a well-designed data warehouse –reduces conflict, errors and “invention of management information”. Hours of manual work producing Excel spreadsheets can be removed.
But with an estimated 70% of business intelligence implementations failing to live up to expectations, how can you ensure your BI project is a success?
These 8 critical success factors can help to ensure that your business intelligence implementation is completed on time, on budget, and delivers a stellar ROI—while also delighting your users.
1. Don’t boil the ocean.
With many enterprise software projects, there’s an “all or nothing” mentality which leads to high costs, long timescales, and unwieldy user requirements, which have often changed by the time the technology is delivered. But with a business intelligence implementation it doesn’t have to be like this. In fact, it actively shouldn’t be. Select an area of the business to focus on—for instance, one with the most to gain from delivering improved reporting, and with the biggest data analysis problems today. This might be a functional area like Sales, Supply Chain, Finance or Operations. Deliver this area first and fast. Gain momentum and credibility. Then take your ‘lessons learned’ into the next area of your business, and start there.
2. Focus on ease of use.
The audience with the most to gain from a successful business intelligence implementation is the business user. But others will benefit too: IT will no longer have to produce/amend reports for business users, for instance, and Finance won’t have to manually produce Excel spreadsheets any more. But these all rely on the business user being able to “self-serve”. And business users like things really simple. An easy-to-use front end, devoid of technical terms, table names, and SQL syntax, is critical to ensuring a successful business intelligence implementation. Choose a simple-to-use tool. Include a glossary of fields, measures and terms using the familiar nomenclature that the user already understands. Have explanations— built into the software—of what data means and where it came from. Finally, ensure that your delivery platform is one that the user is comfortable with—which today almost certainly means a browser.
3. Performance is key.
While it may have previously taken a whole day to produce a report, the reality is that any delay at all will still annoy users. This will hit your user satisfaction ratings, and in turn, impact the success of your business intelligence implementation—thereby killing adoption. In a world where a web search for ‘business intelligence implementation’ returns 41,400,000 results in 0.38 seconds, your users just don’t want to wait. And they don’t have to. Use modern technology like in-memory analytics, columnar databases, SSD disks and advanced caching to deliver blistering, sub-second performances over millions of rows of data—and ensure your users will love you forever!
4. Pick your technology carefully.
There are many ways of tackling business intelligence and self-service reporting, and many different technology options. Pick yours carefully, and don’t necessarily just go for the one that comes bundled with your existing software packages. While products may have the same badge, they are usually just separately packaged BI tools that require separate implementation. With that in mind it makes sense to evaluate them in their own right, and against alternatives which may better suit your requirements, skill-levels and budgets.
5. Understand where the cost is.
Evaluating the cost of a business intelligence implementation is usually an ‘iceberg model’ exercise. The software licence spend usually represents about 20% of the overall cost (the visible part of the iceberg). But a further 80% (the part invisible from the surface) will need to be spent on consultancy, hardware, ancillary software (databases, operating systems etc.) and training. Usually about 40% of the total project cost will then be required annually to maintain it all. So a £40k business intelligence software cost can easily expand to a £200k total project cost, and a £80k per annum running cost. There are exceptions to this rule, of course.
Matillion‘s business-intelligence-as-a-service model, for instance, wraps the entire cost into a one-off fixed price setup and a monthly subscription fee—no hardware, software, or further consultancy required. But most BI software is not like this, so understand where the cost is before you begin.
6. Nominate advocates.
At Matillion, we believe that one of the best things that we do as part of our business intelligence implementations is to identify and work with advocates. These are domain experts who understand the reporting requirements in the business. Often, they’re the very people who beforehand had been hand-cranking management reports using Excel. They have a lot to gain from BI, so they’ll be on your side. They help to get the reporting requirements right, and can test and reconcile numbers. Once the business intelligence implementation is complete, they become your ‘super-users’, promoting BI use, and helping others.
7. Reconcile, reconcile, reconcile.
The best way to instantly kill your business intelligence implementation is to deliver numbers that—in your users’ eyes—don’t add up. They’ll use the system once, lose trust in it, and never log-in again. To avoid this, ensure you reconcile numbers carefully back to trusted sources. Where calculations are included, explain these calculations to the user (ideally through the user interface itself e.g. with a ‘tool-tip’) so they know and understand what they’re looking at.
8. Ensure executive involvement.
At Matillion, we can usually tell at the beginning of an implementation exactly how well it’s going to go, purely by the level of executive involvement. Quite simply, business intelligence implementations with a high level of line-of-business-executive involvement invariably work better and faster than those without.
Executives can get things done, make decisions, and allocate resource. They’ll also be the ultimate arbiters of project success, so if they expect to be happy at the end of the implementation, they should get themselves involved in it. Finally, there’s no substitute for leading from the front. If users see that the executive level isn’t interested, they won’t be either.
Implement these 8 critical success factors in your company and your business intelligence implementation will be off to a flying start.
If you’re currently considering a business intelligence implementation or would like to improve reporting in your organisation, download our Free E-book using the button below, or request a demo of cloud business intelligence software Matillion BI today.