Decision Support Survival: How to Avoid Automating a Mess

There’s a paradox at work surrounding the IT tools and toolsets required for sound real property decision support, according to Ray Summerell, VISTA’s Vice President for Corporate Development. “The old systems don’t do what we need, replacing them is expensive, and we don’t have sufficient IT experience to upgrade the existing systems on our own. Meanwhile, reporting requirements unique to the federal government mean we can’t just go out and buy software. Processes change along with reporting authority and responsibility” he stated.

Summerell made those remarks at the January luncheon membership meeting of the Federal Real Property Association (FRPA), in a presentation titled “Decision Support Software, for Enterprise Implementations: A Federal Real Estate Survival Guide.”

“Sustaining good, valid data for decision support is essential,” Summerell said. “To do that we have to get the architecture right the first time. You can’t just add on additional applications if the underlying process is broken.”

This article summarizes Summerell’s presentation and explains the steps required to ensure that your department or agency can successfully develop toolsets to maintain valid data for real property decision support.

Ground rules for good data

When dealing with vendors or consultants, make sure everything is documented for you. Get it all in writing – how the company will deal with your scope of work, what the results will be, how you will access your data, and whether your work products will be owned by you or by a consultant.

“Getting it in writing” applies to your own organization as well. You need to establish a data collection and validation methodology that’s formal and structured enough to ensure consistent good data, but that’s flexible enough to adapt to new requirements and to allow for sustainable good data over time.

A data migration strategy is essential. Think twice before you begin. “The industry is good at automating a mess,” Summerell said candidly. Be critical about your data, and ask yourself whether the quality is good enough to justify migrating.

“Just because we’ve done something the same way for years doesn’t mean it’s not incredibly stupid,” Summerell quipped. You have to scrub the data and know your processes are worthy, he added. Good processes and good data will keep you from automating a mess.

Develop a pilot project; don’t try to implement a new system across your whole organization at once. You need an incremental rollout strategy to enable you to plan for changes, gain user acceptance, test the system at a manageable scale and validate the data. A small pilot project will give you a more immediate measurable result, which might make it easier to gain user acceptance and understand what will be required to scale up the system across your enterprise.

Know what’s necessary and what’s sufficient

Sustained quality information for decision support can require a high level of data detail. That detail can come with high associated costs. You need to be able to track the type of information required in any given scenario against the level of information you have – and need – to support asset management decisions.

The types of information you need are or should support your key performance indicators, and are necessary aspects of data. The level of information you require should have a relative importance to your ability to make decisions, which is a matter of sufficiency.

When looking at the level of detail in your data for decision support, you must ask yourself whether it is suitable or sufficient to support decisions. If the answer is yes, you then must ask whether you have invested adequately in the processes required to sustain that type of data for consistent decision support over time.

It is absolutely critical to understand what is sufficient detail in the data that relates to your operational key performance indicators. People who over-invest in data quantity also overs-pend on their processes. The result is an oversized system with too many data points and too many process components; that brings down the useful value of your decision support system overall.

Remember that you don’t need – and can’t support – the same level of detail and accuracy across every key performance indicator category in your data model. Do the upfront work to understand in which areas you need to invest to ensure success in your particular organization.

Tips for success

Leverage existing solutions from similar organizations. Gain insight from how these organizations balance key performance indicators against detail in their data. Even though each organization has its own key performance indicators, taking a page from your nearest neighbor’s playbook will help abbreviate your own learning curve.

Manage your costs. Determine and periodically refine your key performance indicators. Create baselines for further requirements. Minimize new features and beware of scope creep that might produce higher startup costs and greater program risk.

Plan for a quick success. Start with a pilot project that will give your organization a measurable result in 60 to 90 days. Choose the right architectural framework with a presentation layer that will enable your senior management to see slivers of data rather than drowning in detail. Don’t chase after dashboard capabilities until you are certain that your data and processes really sustain necessary key performance indicators.

Don’t expect a perfect fit right out of the box. It is always less expensive to buy a solution for decision support than to build one from scratch, but assume any off-the-shelf solution will only be an 80-to-85 percent fit. Be prepared to adjust your processes if you can or build additions to any packaged system.

Own your work and the supporting data. When working with consultants or vendors, you must be able to control the use and ownership of your data. Ultimately, that data, not the software, is what has value for your organization.

Remember that the end game here is not an exhaustive level of detail for every mundane aspect of your inventory. The goal is to make your organization run better, which means having sufficient levels of high quality data supporting key performance metrics to make sure that your real property decisions support your organization’s mission.

By understanding necessary metrics and sufficient levels of data for each of those metrics, you will be able to confidently justify all your budget-related decisions, control spending, and redirect funds to the actual purpose and mission of your organization.