Getting started with manufacturing business intelligence software systems

Getting your new manufacturing business intelligence software system off the ground and working can be a challenge. Learn strategies to prepare and analyze your business intelligence data.

Getting started with business intelligence (BI) software systems can be particularly challenging for manufacturing

companies. The problem faced by many organizations with multiple manufacturing units starts close to home -- they've probably grown themselves into something barely recognizable.

How to get started with BI in manufacturing
Find out if BI can help manufacturers improve efficiency

Read how to build a business case for BI in manufacturing

Learn key questions for manufacturing BI tools evaluation

"Say, for example, that my company became large by acquiring 10 different manufacturing companies, and they all have their own customer list, their own product naming conventions," said Boris Evelson, principal analyst of Business Intelligence for Cambridge, Mass.-based Forrester Research. "How do I make sure I am not stepping on my own toes? How do I know my factory is calling the same product by the same name as my factory in another country?"

The answers lie in your business intelligence data -- and data preparation, it turns out, is critical.

Business must own the data

No intelligence tool can work without data, and if the data isn't accurate and specific, the tools used will only spew more meaningless data. Surprisingly, the first step isn't building a data warehouse or creating specialized databases. No, the real preparation starts far sooner than many expect.

"We always say the very first step is to admit that what you do as a CEO or COO -- in addition to owning widgets and factories -- is that you own data," Evelson said. "When business owners look at their data problems as a technology problem, they say, 'My data is all over the place. It's not synchronized. What is my IT doing wrong?' That, to me, typically is the first indication that a company isn't going to be successful doing business intelligence."

After ownership is defined and embraced, the steps become clearer.

1. Define your data "IT is there to help you, but you have to come up with the definitions. What is a widget? What is a customer contact?" Evelson said. "Try getting two or three people in a room to define a product or customer profitability metric . . . you have to understand how profitable each customer is. Tracking revenue from each customer is easy, but understanding how much you're spending on a customer is a very difficult task. You can calculate your cost over raw materials, but electricity, real estate, work force . . . that's not attributable to a single customer. How do you allocate all of your costs across your company? You'll get different opinions."

"A business has to have a very clear understanding of what they are going to measure and how they are going to measure it, and then, and only then, can they hand it off to IT," Evelson added.

2. Aggregate your data Many traditional BI software systems start with a dedicated data warehouse. But as tools get better at accessing data from various source systems, data warehouses are becoming less necessary.

"It doesn't matter what you call it, but you need some kind of integrated way to have all your data in one logical place -- not necessarily physical -- but it has to be brought together so you can, for example, relate your financial data to chart data, or relate North American data to European data," Evelson said. "You can build a virtual data warehouse."

Companies can evaluate new technologies designed for BI, he said, as opposed to older relational databases that may have been designed decades ago and augmented over the years.

"There are some new technologies that came from the ground up specifically for data warehousing -- things like columnar databases or inverted index databases, which are especially useful for any company looking to analyze structured and unstructured data," Evelson said.

Clearly, companies will need to map out their data architecture early.

3. Clean your data Meanwhile, manufacturing-focused companies need some sort of master data management plan to ensure that their BI data is accurate -- they simply can't make decisions based on faulty data.

"All this data aggregation, data synchronization, and data cleansing is a very tough exercise -- that's the bulk of the difficulty in BI environments. Once you get the data in one place, and it's clean, you can start using it," Evelson said. "That first part is huge, and that's what companies are spending millions of dollars on."

Principles for business intelligence success for manufacturers

Of course, in any major undertaking, the rules are rarely hard and fast.

"There are no strict, well-defined methodologies -- the whole process is much more an art than a science," Evelson said. "It's all about lessons learned, not repeating someone else's mistakes. You can't really buy a strategic Business Intelligence 101 textbook."

Still, there are solid principles for success. Here are some important ones:

Call in the consultants: While it may be theoretically possible for an organization to create its own manufacturing-focused business intelligence system, it's going to be much more difficult if it doesn't hire consultants.

One best practice, according to Evelson, is "to work with consultants who have done this before, who know best practices, who have already accumulated a long list of potential pitfalls, so when they come to you, you pay them so you don't repeat the same mistakes."

Don't try to boil the ocean: Manufacturers have to take small steps, and plan waypoints into their journey.

"Pick a metric you can implement in a few weeks," Evelson said. "Try to show your key stakeholders something tangible, because if they don't see something tangible within a few weeks, people lose interest, lose focus, and business requirements change. You can't have long-term strategic planning without still delivering something tactical every few weeks."

From KPIs to real-time data -- use only what you need: Manufacturing intelligence that's focused on the production line can be particularly useful if the data is real-time or near real-time. But for strategic analysis and decisions, data that's a few days old may be of little use. Real-time dashboards are impressive, but a company has to consider the metrics it's measuring -- in some instances, it might not be worth the cost of managing real-time data. Still, if the goal is to better understand manufacturing performance, "the further you go away from a line or process, arguably the less valuable your data becomes," noted Simon Jacobson, a research director for Boston-based AMR Research.

Build flexibility: "One of the problems with large BI rollouts, if you build your system without flexibility, and it takes a year to roll out a cross-functional system, then something is going to change," said Matthew Littlefield, senior research analyst for Aberdeen Group.

Buy some BI and data management tools: Organizations shouldn't reinvent the wheel, but rather buy some tools. They may be bolt-on products from the enterprise manufacturing system or the ERP system, or a combination of the two with some best-of-breed solutions thrown in. There's no way around it, though: Selecting applications from vendors is going to take a lot of time.

"There's two major places we've seen people using BI -- one is to analyze and connect production data to corporate performance, and the other is to connect maintenance and downtime data and then to analyze that data," Littlefield said. "So on an asset maintenance side, if you're looking to find trends and reliability -- reduce downtime -- very often that downtime data is stored in ERP or data warehouses, and the analysis tools that come with the BI system can help you do that."

It all depends on the metrics an organization is using, and it will want to make sure that the possible vendors have tools that have been used in similar situations with similar types of data.

Where to start with business intelligence for manufacturers -- getting your hands dirty

The last three key lessons learned are strategic rather than tactical, and they ensure that a company navigates toward the areas that will reveal the best return on investment.

Look for variability: Many organizations get stuck finding the best places to start, but there's a simple rule of thumb that always helps: "It's a matter of attacking the biggest point of variability and getting visibility into that," Jacobson said.

Find points of leverage: Where to start will always depend on a combination of factors that will vary by company, industry and geography. Market factors create opportunity, too.

"When crude oil was $150-plus, logistics and supply chain efficiency were critical, but that's less so now," said Dan Miklovic, vice president of Manufacturing Industries Advisory Services with Stamford, Conn.-based Gartner Inc. "In general, the best advice is -- after a few pilots to validate you know how to use the tool and build some credibility -- go after the areas with the greatest leverage. If labor is a major contributor to cost, use business intelligence to improve labor productivity -- i.e., increase volume or lower cost or both. If energy is a major factor, go after that."

Change your philosophy: To use BI tools and principles successfully in dynamic, global manufacturing environments, companies may need to tweak their basic worldview.

"Adopt a philosophy of continuous improvement," Miklovic said. It really is making a cultural shift and not just implementing a tool."

Chris Maxcer is a freelance writer.

Dig deeper on Manufacturing ERP and business intelligence

Pro+

Features

Enjoy the benefits of Pro+ membership, learn more and join.

0 comments

Oldest 

Forgot Password?

No problem! Submit your e-mail address below. We'll send you an email containing your password.

Your password has been sent to:

SearchOracle

SearchDataManagement

SearchCRM

SearchSAP

SearchBusinessAnalytics

SearchSQLServer

SearchContentManagement

SearchFinancialApplications

Close